Handbook of Research in Mass Customization and Personalization (2 Volume Set)

Vol. 1 Strategies and Concepts 7378.V1.tp.indd 1 5/14/09 3:26:58 PM This page intentionally left blank Vol. 1 Str...

Author: Frank T. Piller

32 downloads 1584 Views 16MB Size Report

This content was uploaded by our users and we assume good faith they have the permission to share this book. If you own the copyright to this book and it is wrongfully on our website, we offer a simple DMCA procedure to remove your content from our site. Start by pressing the button below!
Report copyright / DMCA form

DOWNLOAD PDF

Vol. 1 Strategies and Concepts

7378.V1.tp.indd 1

5/14/09 3:26:58 PM

This page intentionally left blank

Vol. 1 Strategies and Concepts

HANDBOOK OF RESEARCH IN MASS CUSTOMIZATION AND PERSONALIZATION edited by

Frank T Piller RWTH Aachen University, Germany

Mitchell M Tseng The Hong Kong University of Science & Technology, Hong Kong

World Scientific NEW JERSEY

7378.V1.tp.indd 2

•

LONDON

•

SINGAPORE

•

BEIJING

•

SHANGHAI

•

HONG KONG

•

TA I P E I

•

CHENNAI

5/14/09 3:27:00 PM

Published by World Scientific Publishing Co. Pte. Ltd. 5 Toh Tuck Link, Singapore 596224 USA office: 27 Warren Street, Suite 401-402, Hackensack, NJ 07601 UK office: 57 Shelton Street, Covent Garden, London WC2H 9HE

British Library Cataloguing-in-Publication Data A catalogue record for this book is available from the British Library.

HANDBOOK OF RESEARCH IN MASS CUSTOMIZATION AND PERSONALIZATION (In 2 Volumes) Volume 1: Strategies and Concepts Copyright © 2010 by World Scientific Publishing Co. Pte. Ltd. All rights reserved. This book, or parts thereof, may not be reproduced in any form or by any means, electronic or mechanical, including photocopying, recording or any information storage and retrieval system now known or to be invented, without written permission from the Publisher.

For photocopying of material in this volume, please pay a copying fee through the Copyright Clearance Center, Inc., 222 Rosewood Drive, Danvers, MA 01923, USA. In this case permission to photocopy is not required from the publisher.

ISBN-13 978-981-4280-25-9 (Set) ISBN-10 981-4280-25-9 (Set) ISBN-13 978-981-4280-26-6 (Vol. 1) ISBN-10 981-4280-26-7 (Vol. 1) ISBN-13 978-981-4280-27-3 (Vol. 2) ISBN-10 981-4280-27-5 (Vol. 2)

Printed in Singapore.

Shalini - Hdbk of Res in Mass Customization.pmd 1

11/5/2009, 4:14 PM

Volume 1

Strategies and Concepts

v


Foreword & Acknowledgments This book is the third in a series of publications that present the latest advancements in research on mass customization and personalization. Starting with Tseng & Piller (2003) and continuing with Piller, Reichwald & Tseng (2006), we again could collect the thinking of some of the leading scholars and practitioners in the field. In comparison to the previous editions, this is the most comprehensive collection of writings on mass customization ever. This inspired our publisher to name it the "Handbook of Research in Mass Customization & Personalization". The contributions in this handbook were inspired by the 4th World Conference on Mass Customization and Personalization (MCPC 2007), a biannual academic event that gathers the international research and practice community interested in mass customization, held in October 2007 at the Massachusetts Institute of Technology (MIT), hosted by the MIT Smart Customization Group (Mitchell et al. 2007). The conference also included a business seminar held at HEC Business School in Montreal, Canada. The participant roster of the conference represented the interdisciplinary nature of customization and personalization drawing from a wide range of interest from hard core engineering, fashion design, architecture, retail, business strategy to psychology. The papers in this book reflect this richness and scope. Our authors come from diverse schools in leading researech institutions as well as from business practice or consulting firms. Such a voluminous work is not possible without the help from many individuals. At MIT, we sincerely want to acknowledge the support and help by Prof. William Mitchell, Ryan Chin and Betty Lou McClanahan from the MIT Design Lab and the MIT Smart Customization Group. As co-chairs and organizers, they were providing leadership for the MCPC 2007 at MIT, hence paving the way that the research presented in this book could be assembled in the first place. From more than 200 conference contributions, an editorial committee selected the papers included in this handbook. All papers were subject of an additional review process. Along with the feedback authors received in the conference, the manuscripts were modified and edited to the collection of papers presented here. There are too many reviewers to name them here individually, but we want to thank them all for their great service to our community. At RWTH Aachen, Frank Steiner coordinated the editorial and publishing process of this large project as the executive editor and provided valuable assistance to us. Dealing with more than 100 authors and coordinating more than 50 papers is a very demanding and time-consuming task. In addition, we thank our

vii

viii

HANDBOOK OF RESEARCH IN MASS CUSTOMIZATION & PERSONALIZATION

publisher for their patience and continuous support for this project. It was a real pleasure working with World Scientific Co. on this book. Our final thank, however, deserve the authors and contributors to this handbook. Only due to their willingness to contribute their latest research and thinking, this project has been possible. We thank them for their patience and compliance in addressing all the numerous demands and requests that such a book project demands. We believe that the research presented here provides a comprehensive and rich introduction into the various aspects that make mass customization one of the most promising business strategies for this century. Frank T Piller & Mitchell M Tseng

Contents Volume 1 Foreword and Acknowledgments................................................................................... vii Introduction: Mass Customization Thinking: Moving from Pilot Stage to an Established Business Strategy ...................................... 1 1

Strategic Aspects of Managing Mass Customization & Personalization............ 19 1.1

From Mass Production to Mass Customization: Hindrance Factors, Structural Inertia and Transition Hazard

21

How to Implement a Mass Customization Strategy: Guidelines for Manufacturing Companies

44

Media Market Inertia: A Potential Threat to Success of Mass Customization

65

Operationalizing Mass Customization – A Conceptual Model Based on Recent Studies in Furniture Manufacturing

79

1.5

Beyond Mass Customization: Exploring Features of a New Paradigm

97

1.6

Is the Best Product a Unique Product? Exploring Alternatives to Mass Customization with the Online Community of Threadless

118

Before Pine and Dell: Mass Customization in Urban Design, Architecture, Linguistics, and Food

139

1.2 1.3 1.4

1.7

2

Consumer Studies & Marketing Aspects ............................................................ 159 2.1 2.2 2.3 2.4 2.5 2.6

Typology of Potential Benefits of Mass Customization Offerings for Customers: An Exploratory Study of the Customer Perspective

161

The Co-Design Experience: Conceptual Models and Design Tools for Mass Customization

181

Why Consumers Are Willing to Pay for Mass Customized Products: Dissociating Product and Experiential Value

208

Sneakerheads and Custom Kicks: Insights into Symbolic Mass Customization

226

E-Customization: Research and Applications from the Cognitive Learning Theory

240

Modularity as a Base for Efficient Life Event Cycle Management

263

ix

x

HANDBOOK OF RESEARCH IN MASS CUSTOMIZATION & PERSONALIZATION 2.7

3

4

275

Building the Solution Space: Product & Process Design for Mass Customization........................................................................................ 295 3.1

Towards a Knowledge Support System for Product Family Design

297

3.2

Product Family Modeling: Working With Multiple Abstraction Levels

319

3.3

Market-based Strategic Platform Design for a Product Family Using a Bayesian Game

338

3.4

Knowledge Based Configurable Product Platform Models

357

3.5

Change Prediction for Mass Customized Products: A Product Model View

376

Making Manufacturing & Supply Chain Management for MCP Work.......... 401 4.1

An Agility Reference Model for the Manufacturing Enterprise: The Example of the Furniture Industry

403

Overcoming Configuration Process Complexity of Highly Customizable Components

427

4.3

Mass Customization of Responsive Automated Assembly Cells

451

4.4

A Prioritization Algorithm for Configuration Scheduling in a Mass Customization Environment

487

Procurement Mechanisms for Customized Products

513

4.2

4.5 5

Bundling, Mass Customization, and Competition under Consumption Uncertainty

Rapid Manufacturing for Mass Customization.................................................. 535 5.1 5.2 5.3 5.4

Extreme Customization: Rapid Manufacturing Products that Enhance the Consumer

537

e-Manufacturing – Making Extreme Mass Customization Real by Laser-Sintering

555

RepRap: The Replicating Rapid Prototyper: Maximizing Customizability by Breeding the Means of Production

568

Customization of Consumer Goods: First Steps to Fully Customizable Fashionable Ladies' Shoes

581

VOLUME 1: STRATEGIES AND CONCEPTS

xi

Volume 2 Foreword and Acknowledgments................................................................................... vii Introduction and Overview........................................................................................... 591 1

Customization & Personalization of Services .................................................... 601 1.1

How to Master the Challenges of Service Mass Customization – A Persona-Based Approach

603

Mass Customization in Wireless Communication Services: Individual Services and Tariffs

622

Unraveling the Service Innovation Dilemma: The Promise of Network Embeddedness

646

1.4

Emotional Design Techniques in the Personalization of Services

665

1.5

One Size Fits All, Made-to-Measure, and Bespoke Tailoring: Challenges in Building an Attractive Service Portfolio

683

Mass Customization for Individualized Life-long Learning: Needs, Design, and Implementation

698

A Mass of Customizers: The WordPress Software Ecosystem

717

1.2 1.3

1.6 1.7 2

Beyond Bespoke Tailoring: Mass Customization in the Apparel Industry...... 729 2.1 2.2 2.3 2.4 2.5 2.6

3

Virtual Fit of Apparel on the Internet: Current Technology and Future Needs

731

RFID Diffusion in Apparel Retail: How Consumer Interest and Knowledge Lead to Acceptance

749

Discard "one size fits all" Labels! Proposal for New Size and Body Shape Labels to Achieve Mass Customization in the Apparel Industry

771

Developing Considerate Design: Meeting Individual Fashion and Clothing Needs Within a Framework of Sustainability

813

Customized Garment Creation with Computer-Aided Design Technology

833

A Case Study in Personalized Digitally Printed Clothing

852

Mass Customization in Architecture and Construction .................................... 867 3.1

Customizing Building Envelopes: Retrospects and Prospects of Customization in the Building Industry

869

xii

4


Mass Custom Design for Sustainable Housing Development

892

3.3

Customization in Building Design and Construction: A Contribution to Sustainability

911

Applications of MCP in Various Contexts .......................................................... 941 4.1 4.2 4.3 4.4

4.5 5

The State of the Art of Mass Customization Practices in Finnish Technology Industries: Results from a Multiple-Case Study

943

Opportunities and Challenges of Furniture Manufacturers Implementing Mass Customization

965

Mass Customization in the Ophthalmic Lens Industry: Progressive Addition Lenses for Your Visual Map

977

Towards a Mass-Customized, Full Surround Simulation of Concert-Theater Effects When Listening to Music Presented on a Pair of Earphones

996

Simulation Models to Demonstrate Mass Customization Strategies

1005

From Mass Customization to Open Innovation .............................................. 1021 5.1 5.2 5.3 5.4

User Innovation and European Manufacturing Industries: Scenarios, Roadmaps and Policy Recommendations

1023

Bridging the Innovation Gap: From Leading-Edge Users to Mass Market

1044

Ordinary Users and Creativity: Fostering Radical or Incremental Innovation?

1059

Modeling and Evaluating Open Innovation as Communicative Influence

1081

Introduction: Mass Customization Thinking: Moving from Pilot Stage to an Established Business Strategy Frank Piller RWTH Aachen University, Germany Mitchell Tseng Hong Kong University of Science & Technology, Hong Kong

Mass customization regards heterogeneities of demand among different customers not as a threat, but as a new opportunity for profits. To capture this value, however, a firm has to obtain a specific set of capabilities to address the challenges of such a business. In this chapter, we first define the core concepts of mass customization and personalization and briefly discuss their background and state of implementation in industry. We also present a set of challenges that many companies are facing when entering a mass customization business. We argue that companies have to obtain competences along three sets of distinctive capabilities to address these challenges. The term 'mass customization thinking' is introduced to denote practices in companies that follow these capabilities in order to profit from customer heterogeneities. In the second part of this chapter, we provide a comprehensive overview of the research presented in this handbook.

The Way Towards Mass Customization Along with Joseph Pine (1993), we define mass customization as "developing, producing, marketing and delivering affordable goods, and services with enough variety and customization that nearly everyone finds exactly what they want." What one needs, when one needs it. Or, to say it in a different way, mass customization aims at producing goods and services catering to individual customers' needs with near mass production efficiency (Tseng and Jiao 2001). To apply this apparently simple statement in practice however is quite complex. As a business paradigm, mass customization provides an attractive business proposition to add value by directly addressing customer needs and in the mean time utilizing resources efficiently without incurring excessive cost. This is particularly significant at a time where competition is no longer just based on price and

1

2


conformance of dimensional quality. Today, customization has been well accepted in several key industries such as agriculture harvesting machines, trucks, commercial airplanes, elevators, computer hardware, buildings and others. For examples, in the machine tool industry, every machining center is equipped with different options and features to fit into the production requirements of the companies that make the purchase. For years, we could observe a similar patter in this industry that guided the way of companies towards mass customization:

It starts with a segment of the market that has not been served well. Certain types of products are needed to fulfill this market gap.

Companies then design not a single product, but a platform that can be configured to address the requirements of this potential market

The marketing department will then come up with a campaign to communicate the unique differentiation of the products that have been perceived as market needs.

The sales department interacts with customers to translate customers' needs with machine specification and to configure a product that fits well with the customers' requirements and can be delivered when the customers need it. At the same time, and often more importantly, the total cost has to be within the budgets available to the customer.

Next, production and distribution have to figure out a way to produce and to manage the logistics for the necessary components and assemblies and to complete machine efficiently. However, uniqueness of components often translates into set up time and additional cost. This may run against the required budget limitation and lead time. To address these challenges, techniques like flexible scheduling, modularity, commonality and others are applied to counterbalance the additional cost.

In a final step, often years after product modularity and flexible processes in manufacturing and logistics have been established, the firm starts to increase the efficiency of the customer interaction process when taking the order. A product configuration system is introduced to better communicate to customers what is available and to match customer requirements with existing solutions of the manufacturer.

This sequence of events presents several challenges to accomplish the seemingly conflicting goals of mass customization: on the one hand to satisfy divergent needs of customers and on the other to accomplish efficiency comparable to


3

volume production without the economies of scale. Similar issues in trading cost and time with uniqueness also apply to downstream operations such as outbound logistics, installation, service, maintenance, and recovery. These challenges include tasks along all stages of the value chain. The characteristic of a good ("smart") mass customization system is that it addresses these challenges in a meaningful way. This is where the three capabilities of successful mass customization, which we will describe in the next section, come into place. But let's first have a closer look on the challenges when addressing diverse customer needs individually. Speed and lead time: Products produced in a mass production system are manufactured in batches with sufficient quantity to justify the fixed cost such as money spent for set up. They often come with inventory in warehouses in different nodes of a supply chain, including the shelves at the point of sales. Thus, products are readily available from off the shelf. The actual net lead time for these products spanning from product design, material acquisition to product delivery is often in months if not years. However, customers finding the most "tolerable" products from off the shelf in retail stores often perceive the lead time between perceiving a need and its fulfillment relatively short. For custom products, it is however not practical to follow the same approach by building up inventory along the supply chain. But still, customers may expect to have the same order of magnitude in short lead time by extrapolating the expectation from standard items onto the custom goods. In many industries (especially in consumer markets), this gap between the customers' expectation and the physical limit of time required by a customized production often is significant. Customers' needs: Contrary to the traditional belief, customers often do not know exactly what they want. They lack specific knowledge regarding what is available or feasible from manufacturers or suppliers in the value chain, let alone the relationships among product features and variables. For instances, a buyer of a price of customized furniture may not know what is the standard size of plywood and hence she may not select the proper width of a table that is more economical to produce. Such implicit domain knowledge is not easy to be made explicit in a customization system. This is very different in mass production. Here, manufacturers design and produce a set of products in quantity according to the perceived needs of customers. Customers then make their own selection among this set of products from the manufacturers and their competitors. Customers have to live with their choices and designs made by the manufacturers. Economies of scale: Customization to individual customers' needs intuitively leads to small quantities and higher varieties; hence it becomes difficult to reach

4


the necessary scale of economy. Various techniques have been developed to counter such natural tendency. For examples, although the finished products can be vastly different, they may contain components or subassemblies that are identical for volume production. Customers can then configure to their needs with components that are produced with efficiency in the back end. Another approach is to sequence the products or subsystems in such a way that similar or identical materials can be accomplished with scale economies. Postponement is one of techniques that have been deployed to address this challenge. The idea is to place variety as late as possible in the fulfillment cycle. But how to postpone and where to set the decoupling point is a major challenge for many firms. Value: Offering choices may not automatically be of value to customers. Previously, some manufacturers boasted billions of selections to customers. But most users do not appreciate the choices that are not of their interest. Additional choices often create confusion. In consequence, customers need longer to understand the differences among these choices. This additional mental load can discourage commitment of sales, impose extra sales effort or even chase customers away. Thus, it is important that the product variety offered to the customer matches the perceived value. Synchronizing choice between features, product attributes and options with customer appreciation and willingness to pay is a major challenge in the customization business. Complexity: With high variety and small lot sizes, the tasks of scheduling, organizing, and managing categories, schedules, and division of work can become daunting. Complexity drives additional cost (particularly in the overhead) that could defeat the efficiency goal set by mass customization. Various algorithms and IT solutions have been introduced to manage complexity and to make a customization system more manageable. However, understanding and implementing these methods still is a large challenge for many companies. The Capabilities of Mass Customization Companies that master mass customization successfully have found an integrated way to address these challenges. This requires them to gain competences around a set of three core capabilities that are driving a sustainable mass customization business. The key to profiting from mass customization is to see it as a set of organizational capabilities that can supplement and enrich an existing system. While specific answers on the nature and characteristics of these capabilities are clearly dependent from industry context or product characteristics, three fundamental groups of capabilities determine the ability of a firm to mass customize. Following Salvador et al. (2009), we call them Solution Space Development,


5

Robust Value Chain Design, and Choice Simplification. The methods behind these capabilities are often not new. Some of them have been around for years. But successful mass customization demands to assemble these methods to capabilities in a meaningful and integrated way.

Solution Space Development. First and foremost, a company seeking to adopt mass customization has to be able to understand what the idiosyncratic needs of its customers are. This is in contrast to the approach of a mass producer, where the company focuses on identifying “central tendencies” among its customers’ needs, and targets them with a limited number of standard products. Conversely, a mass customizer has to identify the product attributes along which customer needs diverge the most. Once this is understood, the firm knows what is needed to properly cover the needs of its customers. It can draw up the “boundaries of its playground”, clearly defining what it is going to offer and what it is not – the firm’s solution space is defined. Mass customization implies by necessity the development of vast solution spaces, thus escalating the cost and complexity of understanding customer needs, in terms of spotting differentiating attributes, validating product concepts, and collecting customer feedback.

Robust Process Design. A second critical requirement for mass customization is related to the relative performance of the value chain. Specifically, it is crucial that the increased variability in customers’ requirements does not lead to significant deterioration in the firm’s operations and supply chain (Pine et al. 2003). This can be achieved through robust value chain design – defined as the capability to reuse or re-combine existing organizational and value chain resources to fulfill differentiated customers needs. With robust process design customized solutions can be delivered with near mass production efficiency and reliability.

Choice Navigation. Finally, the firm must be able to support customers in identifying their own problems and solutions, while minimizing complexity and burden of choice. When a customer is exposed to too many choices, the cognitive cost of evaluation can easily outweigh the increased utility from having more choices, creating the “paradox of choice”: too many choices reduce customer value instead of increasing it (Huffman and Kahn 1998; Piller 2005). As such, offering more product choices can easily prompt customers to postpone or suspend their buying decisions, and, even more worryingly, to classify the vendor as difficult to deal with and hence undesirable. Therefore, the third requirement needed to ensure successful adoption of mass customiza-

6


tion is the organizational capability to simplify the navigation of the company’s product assortment. We call that choice simplification. Table 1: Three capabilities to make mass customization work. Mass customization capability

Examples of approaches to develop Mass customization capabilities

Solution Space Development: Capability to identify the product attributes along which customer needs mostly diverge

Innovation toolkits: Software applications that can empower large pools of customers to translate their preferences into unique product variants by themselves, enabling each of them to highlight possibly unsatisfied needs Outcome-driven innovation: Methods to identify latent customer needs in an analytic way and to transfer those into product functionalities. Customers experience intelligence: Definition of adequate processes to continuously collect data on past customer transactions/ behaviors/ experiences and to translate this data into information on customer preferences

Robust Process Design: Capability to reuse or re-combine existing organizational and value chain resources to fulfill a stream of differentiated customers needs

Flexible automation: Using modern "digital" manufacturing technologies that enable high variance in operations at low switching cost. Process modularity: Reusing and recombining existing organizational and value-chain resources to fulfill differentiated customers’ needs Adaptive human capital: People are a necessary element of a mass customization strategy, especially for their capacity to deal with new and ambiguous task.

Choice Navigation: Support the customers in identifying their own solutions, while minimizing complexity and burden of choice

Assortment matching: Negotiating the characteristics of an existing assortment with a model of the customers’ needs in order to propose possible solutions to the customers without requiring significant effort and time in the search process Fast-cycle trial-and-error learning in co-design toolkits: Empower customers to build models of their own needs and to learn about appropriate solutions to their needs by interactively testing the match between these models and the available options. Embedded configuration: Developing smart products that “understand” how they should adapt to usage conditions and re-configure itself accordingly to a user profile.

Table 1 summarizes some of the methods companies can apply when building these capabilities (for a more detailed discussion, refer to Salvador et al. 2009). We find that companies that have found individual means to implement methods and approaches to match these three capabilities are succeeding in their mass customization endeavor. Other companies are just working along one of these capabilities. This is a good strategy as well. Mass customization has to be seen as


7

a process rather than a destination. It is not about achieving a “perfect” state of mass customization. What matters to most companies instead is to continuously increase their overall capabilities to define the solutions space, to design robust processes, and to help customers navigate available choices. A company may already profit tremendously from just implementing better, say, choice navigation capabilities to match diverse requests of customers not familiar with the product category. We call this understanding "mass customization thinking". It provides a way to profit from heterogeneities of a firm's customers. Mass customization thinking means to build the three capabilities outlined before and to apply them for designing a value chain that creates value from serving different customers differently. The contributors to this handbook provide a much more detailed discussion of these capabilities. While the nature of such a handbook with multiple authors prevents us to provide an integrated framework among all the diverse papers presented in this here, we believe that our three-capability-framework will help the reader to see the larger picture between the different streams of research presented in this book. Personalization versus Mass Customization Before we provide an overview of this research, we want to comment on a long debate, apparently never-ending debate. What is the difference between mass customization and personalization? Kasanoff (2009) recently provided a good definition of personalization: "After years of trying to simplify [the definition of] personalization, I finally got it down to two words: Personal = Smarter. The more you customize, the smarter you get. The smarter you get, the more competitive you become. It really is that simple. Doing it, of course, takes a lot of work." According to his definition, personalization is using technology to accommodate the differences between people. Done right, it's a win/win strategy for providing a better outcome for both the service provider and the individuals involved. For example, if a doctor gives a patient a test to determine which treatment will work best for her before the treatment starts, that's personalization. Likewise, if a company gives their clients the option to tell their service center when and how to contact them, that's also personalization. Mass customization is a process for implementing personalization. In some respects, personalization is a goal and mass customization is the way to accomplish that goal. But we need to be careful about defining or debating semantics. Both personalization and mass customization push a company towards being more responsive to the marketplace and thus being more nimble. Both result in a firm that can react faster and more effectively

8


to volatility. Both enable a company to build defendable competitive advantages, because both require a firm to track, understand and accommodate the needs of its customers. In the end, it is not the term, but the result and value created by applying these concepts. However you are going to define these concepts, making them work is what matters in the end. And this is exactly where the chapters in this handbook start. An Overview of Volume 1 of this Handbook: Strategies & Concepts Volume 1 of this handbook is structured along the value chain. After a broader discussion of the mass customization concept and its implementation in industry, Chapters 2 to 5 follow the value chain by looking on mass customization from the perspective of marketing, product creation and design, and manufacturing. The latter aspect is extended by a special focus on rapid manufacturing (3D printing), a technology that is seen as a key enabler of new solutions for mass customization manufacturing. Chapter 1: Strategic aspects of managing mass customization & personalization The book opens with a number of conceptual and empirical contributions that discuss mass customization as a business strategy and its implementation in practice. The first papers address the inertia of companies and markets alike to move towards mass customization – starting the book with one of the largest present challenges in making mass customization work. In Section 1.1, Fabrizio Salvador and Manus Johnny Rungtusanatham review the factors why the state of practical implementation of mass customization does not match the extent of its discussion in the literature. The authors discuss hindrance factors and structural inertia that make the transition from mass production to mass customization difficult. Doing so, this section also addresses many of the characteristics of a mass customization system. The authors build on an extensive case study of a manufacturing facility as it seeks to transition from mass production to mass customization. They propose five factors hindering the move towards mass customization. This theme is continued in Section 1.2 by Erlend Alfnes and Lars Skjelstad who study a number of mass customization implementations in industrial organizations in Scandinavia. They identify three performance objectives, low cost, short delivery time and the degree of customization, and show how companies moving towards mass customization manage to balance these objectives.


9

While the previous sections have argued from the perspective of the company, Section 1.3 looks on the market inertia which become a potential threat to the success of mass customization. Detlef Schoder, Johannes Putzke and Kai Fischbach use the media market to explain why customers and users may not adopt mass customization to an extent as previously predicted. Based on entrepreneurial experience by the authors in setting up an individualized printed newspaper, the rich prospects of mass customization in content-related industries are contrasted with the lack of market take-off. They share an inside story of this venture and derive a number of exploratory explanations for the market inertia of mass customization adoption. Section 1.4 extends the arguments in this section by proposing an integrated conceptual model to operationalized mass customization. Using the example of furniture manufacturing, a leading industry in mass customization, Emmanuel Kodzi Jr. and Rado Gazo synthesize insights from previous research to conceptualize a value-delivery framework for making mass customization work. This section also provides an insight how mass customization can improve the competitiveness of an entire industry. As the authors argue for the furniture sector, this industry has traditionally pursued a concept of competitiveness based on either price or on high-quality differentiation in a craft-business sense. They show how mass customization can provide a new value preposition that seems to be more applicable for the majority of the US market. We believe that this argumentation holds true for many other sectors in developed economies as well. The final sections of this chapter extend the perspective and look beyond mass customization. Nicola Morelli and Louise Møller Nielsen argue in Section 1.5 that mass customization is just one of many strategies of a new paradigm of customer co-creation. Technological and organizational developments are stretching the capabilities of industrial systems, which are now able to address the needs of smaller and more diversified target groups. At the same time though, substantial transformations in the social and economic conditions of our societies are challenging the basic assumption of the existing production systems. This is creating extreme differentiations in demand patterns and is changing the role of customers in the production process fundamentally. The section outlines the characteristics of a new integrated value paradigm and explores methodological directions for addressing the new perspectives by further research. "Is the best product a unique product," asks Adam Fletcher in Section 1.6. He presents the results of study undertaken with the online t-shirt manufacturer Threadless and its virtual community. The aim of this study was to look at an industry where it is technically possible to deliver a "pure" mass customization

10


experience and then to look at different approaches to see what they offer the consumer. Threadless' business model aggregates opinions of user submitted designs and manufacturers the most popular into limited t-shirts (following a mass production model). Fletcher investigates why this model is an attractive proposition for customers and the company. His results challenge a number of assumptions which can be found in the wider mass customization literature. They also remind us that mass customization is not the ultimate business strategy, but (just) one of several strategies that can provide a viable response to today's changing business environment (Ogawa and Piller 2006). Section 1.7 concludes the chapter with a more general and broader discussion of the foundations of the mass customization concept. William Mitchell and Ryan Chin argue that long before Pine and Dell, mass customization thinking has been present in architectural and urban design. Long before B. Joseph Pine II presented a viable economic strategy around the concept of mass customization and Dell executed its custom build-to-order strategy, combinatorial theory and generative systems have been employed in biological systems, grammatical sentence structure in linguistics, and architecture and urban design. The paper traces the conceptual roots of mass customization through the examination of its historical precedents, citing work of Aristotle, Mitchell, Durand, Newell, and Simon. It applies this thinking on the grammar of culinary arts and discusses the limitations of such combinatorial methods. To overcome these limitations, the authors lay out a conceptual framework for achieving high levels of customization. Chapter 2: Consumer studies & marketing aspects Chapter 2 provides a deeper investigation of the customer perspective of mass customization and personalization. It addresses both the capabilities of choice navigation (selling systems for MCP) and solution space design (factors driving customer value in MCP). Hans Bauer, Anja Düll and Dennis W. Jeffery propose in Section 2.1 a typology to structure the potential benefits of mass customization for the customer. Based on an extensive review of the literature and twenty indepth expert interviews, they suggest that style (form) customization especially is capable of generating symbolic, emotional, hedonic and epistemic benefits. Customizing a product with regard to its fit and functionality mainly offers possibilities for the generation of functional benefits, such as quality and comfort, but also has positive effects on the physical health. Furthermore, personal and economical benefits are of importance. Using a focus group methodology, the authors also derive additional insight into moderating factors of a purchasing decision of a mass customization offering.


11

The co-design process of a user defining her custom product in a toolkit is a central element of any mass customization offering. Kate Herd, Andy Bardill and Mehmet Karamanoglu address this process in Section 2.2 and discuss conceptual models and design tools for mass customization. The authors find that the notion of designing for co-design is still relatively under-researched. Co-design can be seen to consist not only of specific activities during product creation, but also include the entire purchasing experience for the customer co-designer (see also Müller and Piller 2004). This section presents a conceptual model and an approach to support design for co-design, encompassing issues of increased emotional connection and positive customer experience. In Section 2.3, Aurelie Merle, Jean-Louis Chandon and Elyette Roux also investigate the additional value generated by mass customization from the consumer perspective. This value can be expressed by the willingness of consumers to pay a premium for mass customized products. In line with the previous sections, the authors conceptualize the perceived value of mass customization into two clusters, the product value and the mass customization experience. In an empirical experiment, they test their integrated model. This quantitative research perspective is supplemented by a qualitative study of customer perception of customization, using the extreme case of Sneakerheads, an online community of sneaker enthusiasts. The study by Michael Giebelhausen and Stephanie Lawson (Section 2.4) provides deep exploratory insight into the symbolic value of mass customization. Members of the "Sneakerheads" community demonstrate their infatuation with sneakers via activities ranging from creating catalogs of custom shoes to buying and selling rare athletic footwear online. A series of in-depth interviews utilizing the Zaltman Metaphor Elicitation Technique (ZMET) provides a better understanding of how issues such as art, self-expression, exclusivity, peer recognition, and counterfeit goods interact with the mass customization of symbolic products by category experts. The remaining sections of this chapter provide a focused view on selected marketing topics. Muhammad Aljukhadar introduces cognitive learning theory as a concept to model and explain mass customization in online markets. Whereas consumers are heterogeneous with regard to cognitive learning styles and strategies, cognitive learning theory proposes several high levels categories that can be used to segment consumers for different customization applications. The findings suggest a positive effect for the congruency between consumer learning styles (strategies) and online message format (content) on communication efficiency, recall, attitude, and decision making. The section proposes a number of

12


applications of cognitive learning theory in the domain, including online consumer segmentation, information content and format customization. In Section 2.6, Florian Siems and Dominik Walcher show how the concept of (product) modularity, a key feature of a mass customization system, can also serve at the base for efficient life cycle management. The authors introduce the idea of "life event cycles" in order to enable a long term relationship between companies and their customers. The final section (Section 2.7) of this chapter represents some of the recent research on formal modeling of mass customization in the marketing sciences. Luca Petruzzellis and Ernesto Somma develop a mathematical model to explain when mass customization is superior to a conventional bundling strategy within a duopoly of differentiated goods and consumer uncertainty. The focus of this section is on information goods, i.e. products with large amount of digital content, which facilitates the customizability of the products. Chapter 3: Building the solution space: Product & process design for MCP A core capability of mass customization is to define a solution space, consisting of the product architectures and corresponding process structures that allow the firm to meet the heterogeneities in the target market in an efficient and effective way. This solution space has to be stable during the execution of a mass customization process – a core characteristic to achieve mass production efficiency – but at the same time flexible enough to address the diverse needs and demands of the customer base. The first paper in this chapter (Section 3.1) develops a knowledge support system that can enable a better design of product families for mass customization. Seung Ki Moon, Xiaomeng Chang, Janis Terpenny, Timothy Simpson and Soundar Kumara describe how such a system supports the knowledge representation, knowledge discovery, and recommendation for product family design. The authors define an ontology to represent products as functional-based hierarchical structures and to describe cost information related to product design. Fuzzy clustering is employed to partition product functions into subsets for identifying a platform and modules within a given product family. Using a case of a family of power tools, the authors demonstrate the application of their method in practice. Section 3.2 continues on the theme of product family design for mass customization. Kaj Jørgensen introduces a method for product family modeling that observes multiple abstraction levels. He argues that customer driven product configuration is concentrated on decisions, which are relatively invariant throughout order processing. Higher abstraction levels are typically related to the identification of basic functionalities of the product and to considerations about


13

the ability to perform functions demanded by the customer. With the modeling approach introduced in this section, the focus of product configuration can be shifted to identification and definition of attributes instead of modules and components. Section 3.3 proposes a methodology for strategic platform design in a product family. The idea of Seung Ki Moon, Timothy Simpson and Soundar Kumara is that game theory allows for better modeling of situations of uncertain market environments. A Bayesian game is used to decide strategic equilibrium solutions for selecting modules in a product family design. To demonstrate the implementation of the proposed approach, the authors continue the case study of power tools introduced already in Section 3.1. Hans Johannesson and Stellan Gedell present a knowledge-based model for configurable product platforms (Section 3.4). In a research project in the automotive industry, a modeling procedure and a new fully configurable platform model concept have been developed. It consists of a set of linked sub-systems which are configurable, generic, and autonomous. The authors describe the development and implementation of this model and comment on its performance achievements in platform design. The final paper of this chapter (Section 3.5), written by René Keller, Claudia Eckert and P. John Clarkson, introduces a model and method to predict the changes for a product (component) over the live time of this product. Using the automotive industry as an example, they discuss how change requests from new customer requirements, coming often late into the process, can be analyzed for potential costly knock-on effects on other components. The authors introduce the Change Prediction Method (CPM) to assess knock-on change risks to support companies in evaluating proposed changes. The method allows for an improved solution space planning in mass customization systems. Chapter 4: Making manufacturing & supply chain management for MCP work Following the value chain structure of this part of the book, Chapter 4 and 5 are dedicated to selected methods and tools of manufacturing and supply chain planning in mass customization companies, addressing the capability of "robust value chain design" (Salvador et al. 2009). Chapter 4 is primarily discussing planning issues, while Chapter 5 will look on advancements in manufacturing technologies. To start the chapter with a general perspective on mass customization manufacturing, Riadh Azouzi, Sophie D'Amours and Robert Beauregard connect the idea of agile manufacturing with mass customization in Section 4.1. Using again an

14


example of the furniture industry (see Section 1.4), their paper proposes an agility reference model to represent the specific capabilities required by along the value chain. Agility is described along three perspectives: flexibility, responsiveness, and autonomy. It is shown that each perspective addresses a distinct set of competences, which required the application of technologies with specific properties. In two case studies, the authors analyze the agility properties of manufacturing technologies in use and the corresponding customization strategies. Section 4.2 addresses the issue of complexity in mass customization manufacturing. Erik Oestreich and Tobias Teich introduce a flexible configuration model that supports the generation of free definable descriptions and configuration dialogues, which can be used to identify individual components instead of using unique item numbers. Their method, developed and tested in the German automotive industry, also considers the fact that in many situations there is additional demand for standard parts, needed for the final assembling of an individual component. Flexible manufacturing in the automotive industry also is the field of study in Section 4.3. Here, Ulrich Berger, Ralf Kretzschmann, Veronica Vergas and Sarfraz Ul Haque Minhas present an approach for flexible assembly of custom components. They argue that present solutions for production control at the manufacturing cell level as well as the plant level have certain disadvantages, such as manufacturer dependent programming of industrial robots and difficulties in implementing synchronized robot simulation and program execution. The authors discuss different approaches to overcome these limitations. In Section 4.4, Ashok Kumar and Frank Piller introduce a prioritization algorithm for scheduling the production of an arbitrary number of product configurations when production budgets and time are limited. The authors first develop three measures of the value associated with each configuration of the product. Using these measures, a linear programming model is developed to find the optimal sequence of configurations to maximize the total value of the production over a given period. Given the dynamic nature of configuration demands and a constantly evolving system state, two efficient heuristic solutions are developed to solve this problem. The final section of this chapter extends the perspective and takes a supply chain management view (Section 4.5). Songlin Chen and Mitchell Tseng address an issue that has not been covered in the existing literature on mass customization in a larger extent, procurement strategies for customized products. The authors develop a conceptual framework based on contract theory and axiomatic design theory to characterize the essential decisions involved in procuring customized products. Based on the framework, they identify key barriers that prevent


15

customers from effectively tapping into the value of customization. To overcome these barriers, the authors explore alternative procurement mechanisms. Chapter 5: Rapid manufacturing for mass customization Chapter 5 provides a detailed view on an advanced topic of manufacturing for mass customization: rapid manufacturing. This term represents a family of new production technologies (also called "3D printing") that can revolutionize the way how mass customization products are delivered by fast, flexible, and cost-effective production directly from electronic data. On the one hand, rapid manufacturing promises a design freedom for customized products that cannot be reached with conventional technologies. At the same time, rapid manufacturing can also achieve lower costs. Saving on molds reduces time and costs. Economies of scale are fading thus liberating manufacturing decisions from lot size optimization, forecast accuracy, and break even points. Rapid manufacturing promises to overcome these constraints of present manufacturing systems, but also introduces new challenges. In Section 5.1, Christopher Tuck, Min-Huey Ong, Helen Wagner and Richard Hague provide a detailed introduction into rapid manufacturing and its applications in the mass customization domain. Combining results from several research projects, the authors present an integrated process view of generating a customized product using rapid manufacturing techniques. Their sample case is that of a customized motorcycle seat that is adapted to the driver’s body. Methods and issues associated with manufacturing personalized seating are explored and the service requirements for motorcycle seat consumers are identified. Christof Stotko and Andy Snow present in Section 5.2 a practitioner-focused view on the applications and technologies of laser-sintering as one example of rapid manufacturing technologies. The article provides an introduction into the technologies behind laser-sintering and shares results from a number of use cases in various industries. The authors also discuss potential future applications that can extend the present state of customization. While the technologies discussed in the two previous sections demand rather advanced machinery with high machine cost, Ed Sells, Sebastien Bailard, Zach Smith, Adrian Bowyer and Vik Olliver argue in their paper that rapid manufacturing technology also can be provided at very low cost, even meeting DIY applications in the home (Section 5.3). The authors present insights from the RepRap project, a "replicating rapid prototyper". This is a filament-deposition rapid prototyping machine that has been designed to manufacture the majority of its own parts. All other parts of the machine are standard materials easily available.

16


RepRap is intended to maximize the customizability of both the products that it makes and also itself. While this machine may not yet be used for manufacturing of components in industrial settings, it is a great tool to educate future users about the opportunities of customization. In Section 5.4, another extreme case of mass customization based on rapid manufacturing technologies is presented. Marc van der Zande, Sjors Bergmans, Nico Kamperman and Bart van de Vorst show how rapid manufacturing can be used to produce a fully customizable ladies' shoe – in one single manufacturing process. Such an application requires deep knowledge of footwear requirements, production processes, material properties, and design opportunities. The article summarizes the main achievements of the project and opportunities to transfer this learning onto other product categories. Volume 2 of this Handbook: Applications & Cases Volume 2 of the handbook provides a focused view on applications of mass customization & personalization in diverse industry settings. It contains a number of extensive case studies of specific mass customization implementations. These case studies discuss the findings presented in Volume 1 in an integrated way and discuss how the three bundles of capabilities of a sustainable mass customization system have been applied in different companies. But beyond just demonstrating "best practices" and learning from case studies, the papers in this part of the handbook also provide new conceptual, methodological and theoretical contributions with a distinctive industry focus. While the fashion industry (Volume 2, Chapter 2) has been the focus of research in mass customization since a long period of time, the construction industry (Chapter 3) has only recently become an object of study. The same holds true for the vast area of services and intangible products (Chapter 1). This chapter provides insights into specific challenges and methods required to customize services efficiently for a larger customer segment. But also furniture, optical lenses, industrial machinery, or high-end home entertainment are fields of mass customization application discussed in this volume (Chapter 4). The final chapter of this book then bridges the topic of mass customization & personalization with a closely related topic, open innovation and customer co-creation in the new product development process (Chapter 5). For a more comprehensive overview about the content of Volume 2, please refer to the introduction in that volume.


17

References Huffman, C. and Kahn, B. E. (1998). Variety for sale: mass customization or mass confusion? Journal of Retailing. 74(4): 491–513. Kasanoff, B. (2009). "Personal=Smarter". Interview in the Blog "Mass Customization & Open Innovation", April 02, 2009, http://tinyurl.com/nkc985. Mitchell, W., Piller, F., and Tseng, M. (eds.) (2007). Extreme Customization: Proceedings of the 2007 MCPC Conference. Cambridge, MA: MIT Design Lab. Ogawa, S. and Piller, F. (2006). Reducing the Risks of New Product Development. MIT Sloan Management Review. 47(2): 65–71. Piller, F. and Müller, M. (2004). A new marketing approach to mass customization. International Journal of Computer Integrated Manufacturing. 17(7): 583–593. Piller, F. (2005). Mass Customization: Reflections on the state of the concept. International Journal of Flexible Manufacturing Systems. 16(4): 313–334. Piller, F., Moeslein, K. and Stotko, C. (2004). Does mass customization pay? Production Planning & Control. 15(4): 435–444. Pine, B. J. (1993). Mass Customization. Harvard Business School Press, Boston, MA. Pine, J. B. II; Gilmore, J. H., and Boynton, A.C. (1993). Making mass customization work. Harvard Business Review. 71(5): 108–118 Reichwald, R. and Piller, F. (2009). Interaktive Wertschöpfung [Interactive Value Creation]. 2nd edition. Wiesbaden: Gabler. Salvador, F., Forza, C. and Rungtusanatham, M. (2002). Modularity, product variety, production volume, and component sourcing: Theorizing beyond generic prescriptions. Journal of Operations Management. 20(5): 549–575. Salvador, F., de Holan, M., and Piller, F. (2009). Cracking the Code of Mass Customization. MIT Sloan Management Review. 50(3): 70–79. Salvador, F., Rungtusanatham, M., Akpinar, A. and Forza, C. (2008). Strategic capabilities for mass customization: theoretical synthesis and empirical evidence, Academy of Management Best Paper Proceedings. Tseng, M. and Jiao, J. (2001). Mass Customization. Handbook of Industrial Engineering, edited by G. Salvendy. 3rd edition, New York: Wiley: 684–709. von Hippel, E. (1998). Economics of product development by users: The impact of "sticky" local information. Management Science. 44(5): 629–644.

Author Biographies Prof. Dr. Frank Piller leads the Technology & Innovation Management Group at RWTH Aachen University. He also is a co-founder of the MIT Smart Customization Group at the Massachusetts Institute of Technology, USA. Before entering his recent position in Aachen in spring 2007, he worked at the MIT Sloan School of Management and has been an associate professor of management at TUM Business School, Technische Universitaet Muenchen (1999-2004). His research focuses on mass customization, open/user innovation, and methods to increase the efficiency and effectiveness of the innovation

18


process. As a founding partner of Think Consult, a management consultancy, he helps his clients to serve their customers better by using truly customer-centric strategies. Contact: mass-customization.blogs.com | [email protected] Prof. Mitchell M. Tseng, Ph.D. is Chair Professor and Director, Advanced Manufacturing Institute, Hong Kong University of Science and Technology. He also is an Adjunct Professor of MIT-Zaragoza Logistic Program. Prof. Tseng started his industrial engineering career in developing key enabling manufacturing technologies for IT industry. Some of them, such as configuration systems for computers, diamond machining for polygons in laser printers, are still widely used in industry. After serving in industry for two decades, he joined HKUST in 1993 as the founding department head of Industrial Engineering. He is an elected fellow of the International Academy of Production Engineers (CIRP), and American Society of Mechanical Engineers (ASME). Professor Tseng is internationally known for his research in Mass Customization and Global Manufacturing. Sponsors of his research include AT & T, Astec-Emerson, Esquel, Honeywell, Lucent Technologies, Intel, SAP, Rockwell International, Liz Claiborne, Motorola, Nokia, GAP, Ford Motor, Norvullus, Tecton, Synocus, Yuesan, OOCL, Novellus, Ove ARUP, HK Air Cargo Container Limited, and various government agencies in Hong Kong, Mainland China and the EU. Contact: ami.ust.hk | [email protected]

1

Strategic Aspects of Managing Mass Customization & Personalization

19


1.1

From Mass Production to Mass Customization: Hindrance Factors, Structural Inertia and Transition Hazard Fabrizio Salvador Instituto de Empresa, Spain Manus Johnny Rungtusanatham University of Minnesota, United States

Research on Mass Customization has largely overlooked the issue of organizational change associated with the Mass Production-to-Mass Customization (MP-to-MC) transition. To address this gap in the literature, we conduct a quasi-longitudinal case study of a manufacturing facility belonging to a division of a Fortune 1000 discrete manufacturing firm as it seeks to transition from MP to MC. We empirically derive five factors hindering the MP-to-MC transition within the research site. We propose five corresponding analytical generalizations explaining how and why these hindrance factors relate to the MP-to-MC transition hazard (i.e., how and why they threaten the likelihood of a successful MP-to-MC transition). To lend credibility to these theoretical insights, we then juxtapose the five factors and analytical generalizations against the more general constructs and prescriptions of Structural Inertia Theory. We conclude with a discussion of the scientific and pragmatic significance of the findings and opportunities for future research. This paper is reprinted with permission from the editor of the Production and Operations Management Journal, where it was originally published in 2008, Vol. 17, Issue. 3.

Introduction The goal of Mass Customization (MC) is to develop, produce, market, and deliver ". . . affordable goods and services with enough variety and customization [such] that nearly everyone finds exactly what [he/she wants] . . ." (Pine 1993, p.48). During the last two decades, academic research has advanced current understanding of the MC phenomenon from multiple perspectives. We know, for example, what the working principles of MC are (e.g., Pine 1993); what the benefits and limitations of MC are (e.g., Squire et al. 2006); what alternative forms of MC can be pursued (e.g., Salvador et al. 2002, 2004); and what technocratic solutions are available to implement MC (e.g., Trovinger and Bohn 2005).

21

22


One important issue – that of organizational change when implementing MC in a Mass Production (MP) operating environment – has largely been overlooked (Duray 2002). While this issue of organizational change is not as critical a concern when a manufacturing firm has the option of pursuing MC in a "Greenfield" context, it presents serious challenges to Mass Producers needing to effect an MPto-MC transition. In such a transition, Mass Producers can expect simultaneous changes with respect to product architectures (e.g., Huang and Kusiak 1998); manufacturing processes, technologies, and control systems (e.g., Su et al. 2005); and supply network configurations (e.g., Krishnan et al. 1999). Because this product-manufacturing process-supply chain redesign profoundly alters the core of an organization, these changes are likely to be met with resistance and, in the worst case, may not be effected (Amburgey and Kelly 1993). In this paper, we focus on understanding the challenges of implementing MC within an existing MP environment. We empirically identify five factors hindering the MP-to-MC transition and articulate five corresponding propositions explaining how and why these hindrance factors relate to MP-to-MC transition hazard (i.e., how and why they threaten the likelihood of a successful MP-to-MC transition). We then theoretically validate the five hindrance factors and corresponding generalizations by mapping them onto the antecedents and tenets of Structural Inertia Theory (Hannan and Freeman 1977, 1984). We conclude with a brief discussion of the scientific and pragmatic significance of the findings and highlight opportunities for future research. Method To understand the challenges of implementing MC within an existing MP operating environment, we deployed a longitudinal, qualitative research design involving a single case, the primary plant ("Plant X") of a division belonging to a Fortune 1000 discrete manufacturing firm as it embarks on an MP-to-MC transition (see Table 1 for a profile of Plant X). We chose a longitudinal research design because the phenomenon of interest involves a transition or change in state across time periods and because the purpose of the research is to better understand the persistence of factors that hinder transition success (Pettigrew 1990). The choice of a qualitative research design is, moreover, motivated by our interest to build theoretical insights on a topic which has had little coverage in the literature (Strauss and Corbin 1998). We collected interview data and archival records from Plant X in multiple discrete rounds between May 2002 and August 2004. Interviews were conducted with 12 key informants: one plant manager, one product engineering supervisor, two product design engineers, two manufacturing


23

engineering supervisors, two manufacturing engineers, two strategic component sourcing managers, and two supplier flexibility and integration managers. Table 1: Profile of Plant X. Overview

Plant X is the primary manufacturing facility for a major business division of a Fortune 1000 company completing globally in diverse industries. This business division manufactures both premium and mass-produced products to serve a highly competitive outdoors market.

Products

The products assembled at Plant X are electro-mechanical units, whose product architecture comprises a few hundred elementary parts clustered into 20 major modules and subsystems. Demand for these products is highly seasonal, with a typical, peak-demand to non-peak demand ratio of 1:7. 64% of annual sales typically occur within a four-month span.

Manufacturing

Plant X, designed and built in the early 1980s, was initially split into two plants-within-plant (or PWPs), each having its own welding booths, stockrooms, docking bays, material handling, support personnel, and assembly line (Skinner, 1974: p. 121). The paint shop, given the large investment cost, was the only exception and was shared across the two PWPs. By 2001, a total of five PWPs, matching exactly the number of product families, and one paint shop, would exist within Plant X.

Production control

To meet demand, products are assembled following a level production strategy, with excess finished goods inventory being stored in a central distribution center during non-peak periods.

Distribution

Finished products are distributed primarily through a network of specialized dealers who entertained stable relationships with both the manufacturer and the customer base. Dealers are fed through a central distribution center.

Sourcing

Located nearby to large overseas suppliers of high-value modules and subsystems. Stamped metal parts, the exception, are produced by a stamping facility owned by the same discrete manufacturing firm and conveniently located a very short distance from Plant X.

During each round of interviews, we followed a semi-structured interview protocol using open-ended questions tailored to the specific responsibility of each key informant. For example, product design engineers were asked to provide concise descriptions of the architecture for the product family to which they had been assigned and how the product family architecture supported or failed to support the pursuit of MC within Plant X. Interviews lasted approximately 1.5 hours, were tape-recorded, and were transcribed, yielding over 600 pages of textual data. Transcriptions were checked by a third party and a research team member to ensure accuracy prior to qualitative data analysis. We also collected approximately 1000 pages of archival records (e.g., internal meeting notes,

24


PowerPoint presentations, and research reports) for triangulation with the interview data (Jick 1979). The importance of data triangulation, especially for empirical research in the production and operations management discipline, has been recently highlighted in Gupta et al. (2006). In performing qualitative data analysis, we applied the classic coding techniques of grounded research (Strauss and Corbin 1998) to the interview data. We then followed a four-step induction process to derive generalizations pertaining to hindrance factors in the context of an MP-to-MC transition. Research Site: History and Context for Empirical Investigation Plant X was designed and built in the early 1980s to mass-manufacture a novel type of product for outdoor use. The market, at the time, was segmented into two niches (residential versus professional use), with each niche being served by its own product model (and associated attachments). Plant X was accordingly split into two focused factories. In the two decades since Plant X began its operations, the external environment has undergone profound changes. The market success of the products assembled at Plant X led to a significant increase in competition, both from divisions of other large multinational firms with strong brands and market power and from new specialist entrants with product offerings competing directly with those available from Plant X. The distribution channel has also evolved with the establishment of large, relatively-powerful "big-box" retailers (e.g., Home Depot, Wal-Mart, and Sears) competing side-by-side with the traditional dealership structure, demanding product offerings differentiated from those available at dealers and exerting additional downward pressures on pricing. The cumulative response to these changes within the industry had been the proliferation of substitutable products in the market. During this two-decade period, Plant X increased not only the breadth of its total product offerings (from two to five product families) but also the depth of its total product offerings. By 2001, Plant X was building a few hundred product variants for a total of 42 product models across five product families. While Plant X may appear to have been successful in pursuing MC, the results of qualitative analysis suggest otherwise. Personnel within Plant X were, in fact, well aware that its product proliferation response was inefficient; for many years, they attempted to correct this but with no apparent success.


25

Product proliferation at Plant X In retrospect, it was clear that product proliferation at Plant X had been achieved at the cost of excessive proliferation of components. Inventory records for 1994 show, for example, that more than 1300 different stock-keeping units of stamped steel parts were being tracked in storage. Sometime afterward, Plant X instituted a "component development group" to try to standardize components across product families. Seven years later (in 2001), an internal assessment study found that common parts across the five product families still ranged from a low of 5% to a high of more than 35% and that by increasing component commonality up to 20% Plant X could potentially save $5.2 million annually. Nonetheless, despite the continued existence of the "component development group," no significant progress toward component commonality could be discerned by year-end 2003. Besides proliferating components, Plant X had also acquiesced to the undisciplined release of product changes throughout the year. These changes were often approved by engineering or mandated by marketing because of the perceived necessity either to counter an actual or an imminent competitive product offering or to incorporate some innovative product feature. The result of these changes was a constant interference with regular operational activities performed not only within Plant X but also by its supply base. Such interference was particularly harmful during peak-demand periods when Plant X and its suppliers were already at full capacity. To tame the untimely release of new or updated product variants during peak-demand months, Plant X issued a "model year" policy so that new models and non-safety-related engineering changes would only be released at certain times of the year. However, while there was agreement to this policy in principle, design engineers and marketing personnel, in reality, continued to violate this policy. Finally, because of the partitioning of Plant X into five plant-within-plants (PWPs), one for each product family, it failed to capture synergies in manufacturing across different product families. During its early years, when the product family mix was limited to two, Plant X was able to exploit repetition and simplicity using two PWPs to serve its two major, stable and homogeneous market niches. As greater product variety ensued in response to competitive pressures, the total number of PWPs within Plant X also grew from two to five. Over time, operating these five PWPs became increasingly burdensome since production mix could not be readily allocated across the five independent PWPs. An internal report completed in early 2000 urged Plant X to reduce the number of PWPs from five to three, arguing that this would reduce field inventory by 5%, decrease overtime penalties by 50%, and save $4 million in duplicate manufacturing equipment. Despite prompting extensive

26


debate as to the possibility and the pros and cons of this suggestion, the internal report, however, did not result in any consolidation. At the end of 2003, the number of PWPs within Plant X remained at five. Context and timing of data collection Interviews and archival data collection began shortly after a mandated 5-week production shutdown at Plant X. This drastic decision, unprecedented in the history of the plant, had been precipitated by the disastrous operational and financial results that Plant X faced at the end of the 2001 season. Because of extreme market turbulence and incorrect demand forecasts, Plant X saw its 2001 order fill rate plummet to an all-time low of 61%, leaving over $150 millions of unsold inventories within the distribution channel at year-end. The trauma of the mandated production shutdown set off a plant-wide reflection as to how and why Plant X had, for such a long time, failed to adapt appropriately to the needs and changes in the external environment. The timing of our research investigation was, therefore, opportune and allowed us to use observations and interviews to unobtrusively build a retrospective understanding of the factors that hindered the MP-to-MC transition for Plant X. Factors Hindering the MP-to-MC Transition Table 2 summarizes the results from a four-step induction process. We first highlight and describe a hindrance factor and, based on interview and archival evidence, illustrate how the factor hindered the MP-to-MC transition in Plant X. We then show, more generally, that the hindrance factor supports the business model for, and is likely to be present within, any Mass Producer seeking to effect an MP-to-MC transition. We demonstrate next that the hindrance factor, while appropriate for a Mass Producer, is incompatible with the requirements of MC and should, therefore, be removed if a Mass Producer wants to successfully implement MC. Finally, we explain how and why the hindrance factor inhibits an MP-to-MC transition. Marketing Approach Towards Product Specification Product planners for Plant X viewed the market as comprising of different niches defined by price points and of competitors competing by positioning products along these price points and, therefore, considered price to be the primary differentiator among products offerings. Product specifications, as a result, tended to include features to justify different product prices in the eyes of customers but

27


that did not necessarily enhance customer utility. This observation was confirmed by the findings of an independent market research that Plant X had commissioned in 2001. Over time, this view of the market and approach to defining product specifications became institutionalized and influenced the continual introduction of new product variants, often with new or modified components and regardless of their utility to customers. Designing, manufacturing, and delivering products to meet price points rather than to accommodate different customer needs, therefore, became the norm for Plant X. Table 2: Hindrance factors and MP-to-MC transition hazard: Nominal definitions and rationale for generalizations.

Factors hindering the MP-to-MC transition

Marketing Approach towards Product Specification: Marketing focus on identifying and exploiting needs that are maximally similar within large market segments

Why should the factor be present in an MP context? How and why does the hindrance factor increase MP-to-MC transition hazard? Why is the factor not appropriate for an MC context? Marketing in an MP context has to identify central tendencies among customer needs and target these "average needs" with one or a few appropriate standardized products. Marketing in an MC context has to identify and exploit real differences among customer needs and specify a "solutions space" to profitably serve these heterogeneous needs.

Component proliferation is generally not an issue in an MP Accounting Procedures context, reducing the need for accounting procedures to for Computing Direct accurately quantify and allocate the Product Costs: Accounting procedures "variety-related" manufacturing overhead to specific product that cannot accurately compute and allocate the offerings. portion of manufacturBeing able to compute precisely ing overhead resulting from parts proliferation how "variety-related" portion of manufacturing overhead would to specific product in/decrease with the addition offerings (subtraction) of a product variant within the existing portfolio is crucial in an MC context.

Lacking the knowledge and tools to identify varieties that are meaningful to customers, marketing is likely to rely on product differentiation criteria that had been successful in the past or to engage in mimicking competitors, therefore failing to uniquely tap unexploited customers' heterogeneities. Ineffective product extensions are likely to yield unsatisfactory returns, prompting the organization to challenge the validity of, and eventually abandon, the pursuit of MC. When the real cost of adding a new part or removing an existing part is not known, a Mass Producer is likely to adapt to the requirements for high requisite variety by hastily adding variants without fully realizing the cost implications of doing so. Even though a Mass Producer may notice the effect of wrong product extensions on its bottom line, its cost accounting system is likely to deter component commonality initiatives needed for MC, since the cost accounting system is not capable of quantifying the associated "varietyrelated" manufacturing overhead reduction.

28

HANDBOOK OF RESEARCH IN MASS CUSTOMIZATION & PERSONALIZATION Table 2 (Continued )

Factors hindering the MP-to-MC transition

Priorities of the Engineering Design Cultures: Priorities of the engineering design culture that overemphasize design uniqueness or variable cost minimization

Investment Criteria for Manufacturing Assets: Investment criteria for manufacturing assets tied to the quest for economies of scale

Why should the factor be present in an MP context? How and why does the hindrance factor increase MP-to-MC transition hazard? Why is the factor not appropriate for an MC context? The priorities for design engineers in an MP context are to create products that are cheaper or have superior features than available substitutes from competitors.

The priority for design engineers in an MC context is to create product families embedding component commonality so as to efficiently offer product variety.

In an MP context, high volumes and relatively stable demand call for manufacturing assets that deliver on efficiency and economies of scale.

In an MC context, product variety calls for manufacturing assets that also deliver on flexibility and economies of scope.

A supply chain to support MP needs to be designed to cope with volume uncertainty and sourcing

Established Structural Constraints: Structural constraints resulting from the need for a supply chain configured to primarily A supply chain to support MC cope with volume needs to be designed to deliver uncertainty and sourcing product mix flexibility uncertainty

Design engineers' focus on uniqueness or variable cost minimization is likely to relegate component commonality – a priority for MC – to a secondary goal. Altering the focus from design uniqueness or variable cost minimization towards component commonality is likely to be extremely difficult as this implies an engineering design culture change, one that replaces what design engineers had collectively considered to be "good design" with design priorities alien to their habits and biases. Since human beings are unwilling to challenge their beliefs, the design priorities supportive of MP are likely to be perpetuated at the expense of those supportive of MC. Past investments in inflexible, highly productive machinery and facilities are sunk costs when pursuing MC. Decision-makers tend to avoid divesting prior asset investments because of risk aversion, requisite responsibility, and the "psychology of sunk cost." Even if decision-makers are willing to contemplate the possibility of writing off past investments, they may be willing to postpone the decision itself on the basis that the payoff for pursuing MC is itself too uncertain. Reconfiguring the supply chain for product mix flexibility, as mandated by MC, is likely to be deterred by structural constraints within the organization and externally within supply chain constituents. Internal constraints can develop because supply chain reconfiguration decisions may be in the hands of other parts of the organization (e.g., corporate sourcing). External constraints can develop because supply chain constituents may, themselves, suffer from the same changeopposing factors and because of the trust and contractual issues associated with initiating changes across independent organizations.


29

Generalization. The fixation on price points, while possibly a peculiarity of Plant X, underscores more generally the marketing approach towards product specification as a factor hindering an MP-to-MC transition. The marketing approach towards product specification that is appropriate in an MP context does not support the pursuit of MC. For a Mass Producer, the marketing approach should ideally focus on the identification of central tendencies and commonalities among customer needs and, subsequently, on the targeting of these "average needs" with one or a few appropriate standardized products. In contrast, for a Mass Producer to effectively pursue MC, the marketing approach should focus first on the identification and exploitation of real differences among customer needs and then on the specification of a "solutions space" to profitably serve these heterogeneous needs (Tseng and Piller 2003). For a Mass Producer to effect an MP-to-MC transition, it must, therefore, change its marketing approach towards product specification and acquire the appropriate knowledge and tools to facilitate the identification and exploitation of real differences among customers (von Hippel and Katz 2002). Unfortunately, this is much easier said than done, especially in light of prior success in serving the differentiated needs of customers with mass-produced products. In the absence of a clear understanding of the heterogeneities among customers, a Mass Producer, if mandated to pursue MC, might simply opt to mimic competitors by replicating the same set of attributes within its own product offerings. Such mimicking would, of course, be unlikely to offer a sustainable competitive advantage (DiMaggio and Powell 1983). Alternatively, a Mass Producer might pursue MC by selecting an attribute that already has some embedded variety and add, to this attribute, even greater variety. Plant X, for example, did this by proliferating engine sizes which had historically been a useful price differentiation factor. However, simply pushing excessive final product variety further along a single attribute would likely yield marginally decreasing improvements in customer utility (Lancaster 1971), add unduly to choice complexity from the customer’s perspective, and ultimately deter a customer from placing an order (Schwartz 2000). In either case (mimicking competitors or adding variety to a single attribute), satisfactory financial returns would not likely be generated. Disappointed with the financial results, a Mass Producer may, therefore, erroneously conclude the market does not need or desire greater product variety, abandoning, as a result, its efforts to transition from MP to MC. Hence, for a Mass Producer:

30


Proposition 1: Reliance on a marketing approach towards product specification that focuses on identifying and exploiting maximally similar needs within large market segments increases the MP-to-MC transition hazard. Accounting procedures for computing direct product costs The cost accounting system for Plant X was not configured to capture and properly allocate the complexity cost resulting from parts proliferation. Whereas Plant X’s cost accounting system was able to capture the "variable costs" component of a part (e.g., direct materials cost and labor cost), it was not able to quantify precisely the "direct product cost" component tied to the additional system complexity generated by the part (e.g., overhead costs from part documentation, keeping the part in inventory, planning for production of the part, etc.) Instead, as documentations provided by the accounting department showed, the cost accounting system generally treated these variety-related "direct product cost" (including non volume-related production labor costs, engineering costs, and other indirect costs) as non-direct cost and simply pooled them into manufacturing overhead. Plant X’s cost accounting system, in essence, was not able to estimate the complexity cost due to parts proliferation. The cost accounting system appeared, moreover, to reinforce parts proliferation as acceptable behavior. Design engineers sought, for example, to minimize the "variable costs" component, which was precisely quantified, at the expense of the "direct product cost" component, which was not computed. As such, when requests for financial support to launch component commonality initiatives were made, Plant X was not able to quantify what the expected savings would be. The cost accounting system, therefore, did not help Plant X to reduce system complexity and was, in fact, directly responsible for increasing system complexity within Plant X. Generalization. The cost accounting system at Plant X identifies, more generally, the accounting procedures for computing direct product costs as a factor hindering an MP-to-MC transition. Accounting procedures for computing direct product costs that are appropriate in an MP context are inadequate and even misleading in an MC context (Carmona and Perez-Casanova 1993). A Mass Producer with limited product variety would have neither the incentive nor the need to devise accounting procedures to accurately quantify and allocate the "variety-related" manufacturing overhead (i.e., the portion of manufacturing overhead resulting from parts proliferation) to specific product offerings. However, as the number of product offerings expands to accommodate increasingly divergent market needs, as in the case of a Mass Producer pursuing MC, the


31

"variety-related" manufacturing overhead should increase as well (Banker et al. 1990) and would eventually become a non-negligible portion of the manufacturing overhead. In this context, being able to compute precisely how the "varietyrelated" portion of manufacturing overhead would increase (decrease) with the addition (subtraction) of a product variant within an existing portfolio of product offerings becomes critical. For a Mass Producer to effect an MP-to-MC transition, it must, therefore, change how it quantifies and allocates the "variety-related" manufacturing overhead to specific product offerings. Accounting procedures that cannot accurately allocate "variety-related" manufacturing overhead in computing direct product costs could, moreover, deter a Mass Producer from engaging in the activities necessary to support an MP-to-MC transition. Consider, for example, the decision to increase parts commonality by standardizing the chassis for a family of vehicles with different horsepower. To implement this decision, a stronger-than-needed, and so more expensive, chassis – one that meets the requirements of more powerful vehicle models – would have to be used with the less powerful vehicle models. While accounting procedures that support MP are likely to capture the increase in materials costs, they are not likely to capture the reduction in "variety-related" overhead due to component commonality. Component standardization projects that are proposed, given these circumstances, would likely yield a negative net present value and would, as a result, discourage decision-makers from committing to component standardization. Even more seriously, when such accounting procedures are tied to performance appraisals for design engineers, as they were at Plant X, they could discourage design engineers from pursuing component standardization in the design task. Design engineers would then likely to focus on cutting materials costs, since they would not receive any credit for the reduction in the "variety-related" manufacturing overhead. Hence, for a Mass Producer: Proposition 2: Reliance on accounting procedures for computing direct product costs that cannot accurately quantify the manufacturing overhead resulting from parts proliferation increases the MP-to-MC transition hazard. Priorities of the engineering design culture Rather than pursue component standardization across product families, the engineering design culture within Plant X embraced a view of the design task as one of maximizing design uniqueness, often through styling or the incorporation of the latest technological innovations. This engineering design culture was appropriate when Plant X was supplying, to the marketplace, premium products for which uniqueness and innovation were essential to market success. However,

32


as Plant X revised and expanded its portfolio of product offerings to include nonpremium products, the need for design uniqueness declined; yet the original engineering design culture continued to drive the execution of the design task. The continued focus on design uniqueness was, in fact, consistent with the pricedifferentiation logic inherent in the marketing strategy and with the variable-cost minimization logic embedded in the cost accounting system. The complementarities among the price-differentiation logic, the variable-cost minimization logic, and the design uniqueness logic not only led to the habitual design of unique components for specific use by specific product families but also encouraged the somewhat irrational proliferation of parts and product variants. Internal and publicly available documentations revealed, for example, that many product variants had evolved due to the inclusion of multiple and slightly different parts that reduced variable cost and added to aesthetic differentiation but often did nothing to functional differentiation. Generalization. The engineering design culture at Plant X identifies, more generally, the priorities of the engineering design culture as a factor hindering an MP-to-MC transition. Priorities of an engineering design culture that support MP do not support the pursuit of MC. A Mass Producer would not need to rely on product variety to win market orders. Rather, it would have to persuade customers to buy its products either because they are priced lower than or because they possess superior features relative to available substitutes from competitors. In the former case, design engineers are likely to pursue low cost by conceiving ad hoc parts for every mass-produced product variant, removing every ounce of material not needed to meet the exact specification for that particular product variant (e.g., steel thicknesses at Plant X). In the latter case, since style and technical innovativeness are highly desirable product attributes, design engineers are likely to continually update products to maximize perceived novelty and uniqueness. In contrast, because a Mass Producer pursuing MC wins customer orders by efficiently offering product customization, component commonality would become a critical priority in the design task. As such, for a Mass Producer to effect an MP-to-MC transition, it must alter its exclusive focus away from product design uniqueness or variable cost minimization and embrace component commonality as a primary design priority. Altering these design priorities would, however, be extremely difficult, especially when the design priorities have become institutionalized over time and embedded in what design engineers believe to be "good design" (Carrillo and Gromb 2007). Changing them would require altering fundamental beliefs about what constitutes good design. Since human beings are generally not able or willing to acknowledge


33

personal or collective certainties and beliefs as opinions, replacing an engineering design culture that equates product design uniqueness or variable cost minimization to good design with a new culture that considers, consistent with the pursuit of MC, component commonality as good design would, therefore, not be easy to achieve. Hence, for a Mass Producer: Proposition 3: Reliance on priorities of the engineering design culture that over-emphasize design uniqueness or variable cost minimization in product design increases the MP-to-MC transition hazard. Investments in manufacturing assets Investments in high-yield processes to achieve economies of scale (e.g., automated welding stations, automated painting shop) dominated the manufacturing strategy for Plant X. Emblematic of this investment logic was the stamping facility, tasked with the responsibility to supply numerous stamped steel parts for the five different product families. Initially, the stamping facility was cost competitive because the investment in each die was spread across many massproduced units. Likewise, the aesthetic quality of deep-drawn steel parts allowed Plant X to differentiate its products from those of competitors. However, with the advent of plastic composite technology, the economics of manufacturing these components changed. Small batch production of components made from plastic instead of steel became feasible and, more importantly, efficient without the need to sacrifice quality. Yet, rather than divesting itself of the stamping facility, Plant X maintained a continued focus on maximizing the utilization of the stamping facility. Analyses of part numbers show that the stamping facility not only produced and tracked more than 1300 stamped steel parts for use by the five product families but also produced and tracked over 2100 service parts to support several phased-out products variants that had once been manufactured by Plant X. Because of this need to absorb the large overhead cost tied to the stamping facility, Plant X was not concerned with, and even encouraged, the proliferation of new stamped steel parts, as long as they were being manufactured at the stamping facility. Generalization. The stamping plant at Plant X highlights, more generally, the investments in manufacturing assets as a factor hindering an MP-to-MC transition. Investments in manufacturing assets that support MP are not appropriate for, and become a burden in, the pursuit of MC. Because a Mass Producer should be operating in a relatively stable and highvolume environment, substantial investments in equipment to deliver on economies of scale (e.g., automation or specialized machinery) and to maximize

34


productivity would be expected. In contrast, in a MC context, substantial investments in flexible, modular production technologies (Gupta and Roth 2007; Starr 1965) would be needed to achieve economies of scope (Salvador 2007). For a Mass Producer to effect an MP-to-MC transition, it must, therefore, replace extant manufacturing assets that deliver on economies of scale with those that deliver on economies of scope. A Mass Producer would, nonetheless, find divestiture and replacement of such manufacturing assets to be a non-trivial challenge for several reasons. First, a Mass Producer (or, more specifically, a decision-maker within a Mass Producer) would likely find it difficult to ignore the original acquisition cost of an asset, especially when the residual value of the asset in question is much lower than its initial investment (Arkes and Blumer 1985). Second, because of the tendency to revaluate upwards those assets for which onerous effort has been expended (Festinger 1957) – not unlike what Plant X did with respect to the stamping facility – a Mass Producer would likely understate the business case for divesting existing assets. Third, mired in a past investment decision, a Mass Producer could even be willing to incur small repeated losses, waiting hopefully for an eventual positive outcome (Brockner et al. 1979). Finally, when the likelihood of a successful MP-to-MC transition is ambiguous, a Mass Producer would have an incentive to postpone investments to support this transition in the hope of gaining access to new information that might reduce the associated uncertainties (Dixit and Pindyck 1995). Hence, for a Mass Producer: Proposition 4: Reliance on investments in manufacturing assets tied to the quest for economies of scale increases the MP-to-MC transition hazard. Established structural constraints Plant X, as well as its supply chain, followed a level production approach and deployed buffer inventories to decouple one supply chain entity from another. By doing so, Plant X was able to source components globally and to pool similar purchases with those of sister facilities within the division or the enterprise so as to capture volume discounts. Over time, this sourcing approach evolved into an enterprise-wide purchasing policy wherein sourcing decisions for certain parts were left to Plant X (e.g., forged metal parts) while, for many other critical parts, sourcing decisions were centralized either at the divisional or at the enterprise level (e.g., engines and transmissions). Although the centralized sourcing strategy for these latter parts optimized purchasing costs for the entire enterprise, it actively prevented Plant X from seeking out suppliers who were located in close proximity to the assembly facility


35

and who, more importantly, could provide, quickly and flexibly, components that were needed in multiple variants as Plant X responded to increased product variety. Plant X had little choice but to source these latter parts from an a priori list of "preferred" strategic suppliers. Compounding this situation was the fact that the existing suppliers in this industry, irrespective of distance, were themselves not adequately equipped to deliver a variety of components in a flexible manner, since they too faced the same internal problems as Plant X did. Likewise, Plant X was not able to push product differentiation activities into the distribution chain, since this created a significant amount of complexity for the central distribution center. Postponing product differentiation activities to dealers further down the distribution chain was also problematic, since this mandated a renegotiation of legal contracts, an enterprise-level responsibility over which Plant X was unable to exert much influence. Plant X, therefore, operated under a set of structural constraints, both between Plant X and its parent organization and between Plant X and other upstream and downstream entities within its supply chain. These structural constraints allowed Plant X little control over the configuration of its supply chain which, in turn, limited its ability to achieve sufficient product mix flexibility in support of pursuing MC. Generalization. The difficulties that Plant X faced with respect to reconfiguring its upstream and downstream supply chain recognize, more generally, established structural constraints as a factor hindering an MP-to-MC transition. Structural constraints established by a supply chain configuration for supporting MP are inappropriate for MC. Since a Mass Producer offers only a few, relatively stable products, it would be able to level its production and to have its supply chain build inventory in advance of expected peaks and drain such inventory during non-peak periods. With relatively long product life cycles, production leveling and anticipatory inventory would make sense given low inventory obsolescence risk. The surplus inventory could then buffer the operations of the various value chain constituents against volume fluctuations and reduce the need for tight supply chain coordination. In contrast, the pursuit of MC would mandate the offering of a wide assortment of products characterized by high demand volatility. The entire supply chain, consequently, would have to be configured to cope not only with volume flexibility but also with product mix flexibility (Salvador et al. 2004). No longer would it make sense to carry and position vast quantities of decoupling inventory along the supply chain. By drastically reducing decoupling inventory in the supply chain, the need for increased supply chain coordination among supply chain entities would become more critical. For a Mass Producer to effect an MPto-MC transition, it must, therefore, remove the structural constraints that had

36


been previously established to primarily mitigate volume uncertainty and sourcing uncertainty. Like other hindrance factors, reconfiguring the supply chain of a Mass Producer to support an MP-to-MC transition would also be a daunting task. External suppliers to a Mass Producer are, of course, not internal departments that could be commanded to change so as to support a Mass Producer’s efforts to effect an MO-toMC transition. The negotiation of contracts and the building of trust are timeconsuming but necessary concerns that could slow or even paralyze the change process. Supply chain entities, moreover, could themselves be operating as Mass Producers and, therefore, face the same impeding factors as those discussed above; even if they were willing to engage in an MP-to-MC transition, they could themselves not be successful. Last, the specific unit of a Mass Producer interested in transitioning to MC could lack the authority to mandate a supply chain reconfiguration, simply because such decisions could reside at levels beyond the unit – for example, the home division or the corporation itself. Hence, for a Mass Producer: Proposition 5: Reliance on established structural constraints resulting from the need for a supply chain to primarily cope with volume uncertainty and sourcing uncertainty increases the MP-to-MC transition hazard. Theoretical Validation To theoretically validate the five hindrance factors and corresponding generalizations, we map them onto the antecedents and tenets of Structural Inertia Theory (Kelly and Amburgey 1991). This theoretical mapping complements the induction process (Figure 1) and (a) verifies the extent to which predictions about "why it is difficult for a Mass Producer to effect an MP-to-MC transition" are compatible with the tenets of Structural Inertia Theory concerning "why it is difficult for organizations to change," (b) establishes the five propositions as a middle-range theory of Structural Inertia Theory (Bourgeois 1979), and (c) provides an opportunity to identify errors in the induction process used to generalize contextspecific insights. Structural inertia theory Structural Inertia Theory is a grand theory explaining why organizations find it difficult to change. A central tenet of Structural Inertia Theory is the claim that organizations generally resist change, with this resistance being amplified when the change required is with the organization core (Hannan and Freeman 1977, 1984) – i.e., with the parts of the organization that define its raison d'être, which


37

typically includes the organization’s stated goals, its forms of authority, its core technology, and its business strategy. Any organization that is unable to adapt fast enough to accommodate the rate of change in the external environment is said to be beset by Structural Inertia. Such an organization, according to the structural contingency perspective of Structural Inertia, is bound to fail in the long run (Scott 1987; Nickerson and Silverman 2003).

STRUCTURAL INERTIA THEORY:

Why Is It Difficult for Organizations to Change? (GRAND THEORY)

2

THEORETICAL VALIDATION

Are the predictions of the proposed middle-range theory compatible with those of a relevant grand theory that has been successfully subjected to empirically testing?

a.

b.

c.

We show that the proposed middle-range theory of hindrance factors and MP-toMC transition hazard is a special instance of Structural Inertia Theory, a grand theory informally, more generally, of the difficulty of organization change. We show that the constituent elements of Structural Inertia Theory (i.e., antecedents) subsume the hindrance factors in the proposed middle-range theory. We demonstrate that the theoretical arguments relating antecedents to Structural Inertia are consistent with the logic relating hindrance factors to the likelihood of a successful MP-to-MC transition.

THEORY OF HINDRANCE FACTORS AND MP-to-MC TRANSITION HAZARD

Why Is It Difficult for a Mass Producer to Effect an MP-to-MC Transition? (MIDDLE-RANGE THEORY)

1

ANALYTICAL GENERALIZATION

Are the empirical facts observed from Plant X generalizable beyond Plant X to other organizational entities similar to Plant X?

a. We identify and describe the MP-to-MC hindrance factors and demonstrate how and why they hinder the MP-to-MC transition at Plant X. b. We show that the hindrance factors are likely to be present not only in Plant X but also within any Mass Producer. c. We demonstrate that the hindrance factors are incompatible with the requirements of MC and should be removed in order to implement MC successfully. d. We illustrate how and why the hindrance factors increase transition hazard (i.e., reduce the likelihood of a successful MP-to-MC transition).

EMPIRICAL OBSERVATIONS

Why Was It Difficult for Plant X to Effect an MP-to-MC Transition?

Figure 1: Analytical generalization and theoretical validation.

The relevant literature on Structural Inertia Theory has identified a number of antecedents of Structural Inertia, including organization age, organization size, organization complexity, routinization, past success, past failure, sunk costs, dependency, influence activity of organizational constituents, etc. (e.g., Hannan and Freeman 1977, 1984; Kelley 1990). These antecedents are, however, not intended to be mutually exclusive (i.e., the presence of one antecedent of Structural Inertia does not preclude the presence of another) nor are they allencompassing (i.e., all antecedents need not be present for Structural Inertia to result). Because a Mass Producer transitioning from MP to MC has to effect

38


numerous changes to the organization core, as evidenced for Plant X, such a transition is likely to be affected by Structural Inertia. Structural Inertia Theory is arguably a relevant theoretical lens onto which the hindrance factors and corresponding generalizations can be mapped. Theoretical mapping In Figure 2, we summarize this theoretical mapping. For two of the five hindrance factors, this theoretical mapping is relatively straightforward. For example, the hindrance factor pertaining to accounting procedures for computing direct product costs and the reasons that these accounting procedures are difficult to change relate to the routinization antecedent of Structural Inertia. Likewise, the hindrance factor pertaining to investment criteria for manufacturing assets, as well as the logic as to why assets acquired or developed based on the economies of scale investment criteria cannot be readily changed to support MC, relates to the sunk costs antecedent. For the remaining three hindrance factors, the theoretical mapping is relatively more complex but nonetheless consistent and congruent with Structural Inertia Theory. For example, relating the marketing approach towards product specification hindrance factor to past success bolsters explanations as to why a Mass Producer would stick to a focus on the identification of "average customer needs," even though this focus is logically incompatible with the pursuit of MC. On the other hand, mapping this particular hindrance factor onto organization age helps to explain how and why a marketing approach towards product specification supportive of MP can become institutionalized and the consequent difficulty in replacing it with a marketing approach supportive of MC. The persistence of the priorities of the engineering design culture supporting a Mass Producer can be similarly explained when this hindrance factor is related to past success, organization age, and rountinization. Likewise, arguments as to why it is difficult for a Mass Producer to reconfigure its supply chain in order to overcome established structural constraints that were created to support MP can be mapped onto explanations pertaining to the dependency antecedent and to the organization age antecedent.


39

Figure 2: Theoretical mapping of hindrance factors and propositions onto structural inertia.

40


Conclusions The extant literature has identified product, process, and supply chain redesigns as crucial to the implementation of MC. By articulating the five hindrance factors and their corresponding generalizations pertaining to MP-to-MC transition hazard, we broaden the breadth of critical concerns to also include the necessity of redesigning the cost accounting system, changing the organizational culture, reengineering the marketing-sales process, revamping policies pertaining to capital investments and budgeting, and overcoming barriers to reconfiguring the supply chain. Pragmatically, for a Mass Producer wishing to engage in an MP-to-MC transition, a preliminary evaluation of the strength of the five diverse hindrance factors can be conducted. Such an assessment can highlight whether or not to postpone the initiation of technocratic solutions for implementing MC until specific hindrance factors have been removed or mitigated. For those already in the midst of an MPto-MC transition, ongoing evaluations regarding the prevalence of these hindrance factors can facilitate the prioritization of attention and efforts aimed at overcoming roadblocks to changing the organization core. Moving forward, further research seeking to replicate, confirm, or augment these findings in a different research setting, across multiple settings simultaneously, or within larger statistical samples would be welcome and would strengthen the scientific knowledge base pertaining to this issue. Research delving into specific mechanisms and initiatives for overcoming the hindrance factors (e.g., forms of compensation) would likewise be worthy of consideration. Lastly, research that extends the focus beyond the manufacturing environment to other functional areas (e.g., to accounting or marketing) may reveal additional function-specific hindrance factors that impede changes within the respective functions and reduce the likelihood of a successful MP-to-MC transition for the entire firm. Acknowledgments We wish to thank many colleagues including C. Forza from Università degli Studi di Padova (Italy) and P. M. de Holan from Instituto de Empresa (Spain) for their comments and support. Finally, we acknowledge the financial support in the form of a 2003 Senior Research Fellowship from the Institute for Supply Management for Rungtusanatham and in the form of a research grant for Salvador from the Spanish Ministry of Science and Education (Project FIT-400000-2005-68).


41

References Amburgey, T. L., D. Kelly. (1993). Resetting the clock: The dynamics of organizational change and failure. Administrative Science Quarterly. 38(1): 51–73. Arkes, H. R., C. Blumer. (1985). The psychology of sunk cost. Organizational Behavior and Human Decision Processes. 35(1): 124–140. Banker, R. D., S. M. Datar, S. Kekre, T. Mukhopadhyay. (1990). Costs of product and process complexity. R. S. Kaplan, ed. Measures for Manufacturing Excellence. Harvard Business School Press, Cambridge, MA, 269–290. Bourgeois, L. J. (1979). Toward a method of middle-range theorizing. Academy of Management Review. 4(3): 443–447. Brockner, J., M. C. Shaw, J. Z. Rubin. (1979). Factors affecting withdrawal from an escalating conflict: Quitting before it is too late. Journal of Experimental Social Psychology. 15(5): 492–503. Carmona, S., G. Perez-Casanova. (1993). Organizational forgetting and information systems. Scandinavian Journal of Management. 9(1): 29–44. Carrillo, J. D., D. Gromb. (2007). Cultural inertia and uniformity in organizations. Journal of Law Economics & Organizations. 23(3): 743–771. Child, J., A. Kieser. (1981). Development of organizations over time. P. C. Nystrom and W. H. Starbuck, eds. Handbook of Organizational Design. Oxford University Press, Oxford, United Kingdom, 28–63. Colombo, M. G., M. M. Delmastro. (2002). The determinants of organizational change and structural inertia: Technological and organizational factors. Journal of Economics & Management Strategy. 11(4): 595–635. Cyert, R., J. March. (1963). A Behavioral Theory of the Firm. Prentice Hall, Englewood Cliffs, NJ. DiMaggio, P. J., W. W. Powell. (1983). The iron cage revisited: Institutional isomorphism and collective rationality in organizational fields. American Sociological Review. 48(2): 147–160. Dixit, A. K., R. S. Pindyck. (1995). The options approach to capital investment. Harvard Business Review. 73(3): 105–115. Duray, R. (2002). Mass customization origins: mass or custom manufacturing? International Journal of Operations and Production Management. 22(3): 314–328. Festinger, L. (1957). A Theory of Cognitive Dissonance. Stanford University Press, Stanford, CA. Sushil, G., A. V. Roth. (2007). Martin K. Starr: A visionary proponent for system integration, modular production, and catastrophe avoidance. Production and Operations Management. 26(1): 1–12. Hannan, M. T., J. Freeman. (1984). Structural inertia and organizational change. American Sociological Review. 49(2): 149–164. Hannan, T. M., J. Freeman. (1977). The population ecology of organizations. American Journal of Sociology. 82(5): 929–964. Huang, C., A. Kusiak. (1998). Modularity in design of products and systems. IEEE Transactions on Systems, Man and Cybernetics. 28(1): 66–77. Jick, T. D. (1979). Mixing qualitative and quantitative methods: Triangulation in action. Administrative Science Quarterly. 24(4): 602–611. Kelley, M. R. (1990). New process technology, job design and work organization: A contingency model. American Sociological Review. 55(2): 191–208.

42


Kelly, D., T. L. Amburgey. (1991). Organizational inertia and momentum: A dynamic model of strategic change. Academy of Management Journal. 34(3): 591–612. Krishnan, V., R. Singh, D. Tirupati. (1999). A model-based approach for planning and developing a family of technology-based products. Manufacturing & Service Operations Management. 1(2): 132–156. Lancaster, K. J. (1971). Variety, Equity, and Efficiency: Product Variety in an Industrial Society. Columbia University Press, New York, NY. Levinthal, D., J. G. March. (1993). The myopia of learning. Strategic Management Journal. 14(8): 95–112. Milliken, F. J., T. K Lant. (1991). The effects of an organization’s recent performance history on strategic persistence and change. J. Dutton, A. Huff, and P. Shrivastava, eds. Advances in Strategic Management, 7. JAI Press, Greenwich, CT, 129–156. Nickerson, J. A., B. S. Silverman. (2003). Why firms want to organize efficiently and what keeps them from doing so. Administrative Science Quarterly. 48(3): 433–465. Pettigrew, A. M. (1990). Longitudinal field research on change: Theory and practice. Organization Science. 1(3): 267–292. Pine, B. J. (1993). Mass Customization. Harvard Business School Press, Boston, MA. Romanelli, E., L. M. Tushman. (1985). Organizational transformation as punctuated equilibrium: An empirical test. Management Science. 37(5): 1141–1166. Salvador, F. (2007). Toward a product system modularity construct: Literature review and reconceptualization. IEEE Transactions on Engineering Management. 54(2): 219–240. Salvador, F., C. Forza, M. Rungtusanatham. (2002). Product variety, modularity, and component sourcing decisions: Theorizing beyond generic prescriptions. Journal of Operations Management. 20(5): 549–575. Salvador, F., M. Rungtusanatham, C. Forza. (2004). Supply chain configurations for mass customization. Production Planning and Control. 15(4): 381–397. Schwartz, B. (2000). Self determination: The tyranny of freedom. American Psychologist. 55(1): 79–88. Scott, W. (1987). Adolescence of institutional theory. Administrative Science Quarterly. 32(4): 493–511. Skinner, W. (1974). The focused factory. Harvard Business Review. 52(3): 113–121. Squire, B., S. Brown, J. Readman, J. Bessant. (2006). The impact of mass customization on manufacturing trade-offs. Production and Operations Management. 15(1): 10–21. Starr, M. K. (1965). Modular production – A new concept. Harvard Business Review. 43(6): 131–142. Strauss, A. L., J. Corbin. (1998). Basics of Qualitative Research: Techniques and Procedures for Developing Grounded Theory. Sage Publications, Thousand Oaks, CA. Su, J. C. P., Y. Chang, M. Ferguson. (2005). Evaluation of postponement structures to accommodate mass customization. Journal of Operations Management. 23(3-4): 305–318. Trovinger, S. C., R. E. Bohn. (2005). Setup time reduction for electronics assembly: Combining Simple (SMED) and IT-Based Methods. Production and Operations Management. 14(2): 205–217. Tseng, M. M., F. T. Piller (Eds.). (2003). The Customer Centric Enterprise: Advances in Mass Customization and Personalization. Springer, New York, NY. Von Hippel, E., R. Katz. (2002). Shifting innovation to users via toolkits. Management Science. 48(7): 821–833.


43

Author Biographies Fabrizio Salvador is Professor of Operations Management at Instituto de Empresa Business School, Adjunct Professor at the MIT-Zaragoza Logistics Program and Research Affiliate at the Massachusetts Institute of Technology. He has been Faculty Resarch Associate at Arizona State University. He received a Ph.D in Operations Management from the University of Padova, where he also graduated in Industrial Engineering. Dr. Salvador’s research focuses on operation strategy in uncertain environments and customer-centric organization design. He has been researching such topics as mass customization, concurrent product-process-supply chain design and organization design for efficient product configuration. His research has been published in many prestigious academic journals, and he has co-authored the book "Information Management for Mass Customization" Contact: http://fabrizio.salvador.profesores.ie.edu | [email protected] M. Johnny Rungtusanatham joined the faculty of the Carlson School of Management at the University of Minnesota-Twin Cities in 2006. Prior to this, he was a faculty member of the W. P. Carey School of Business at Arizona State University (ASU). His research addresses topics like implementing TQM across countries, implementing mass customization, defining and qualifying the impact of supply chain disruptions, managing purchasing items inventory, and optimizing Internet retailing operations. He has conducted research with, consulted with, and provided executive training for Arizona Public Services, Chevron Corporation, Deere & Company, e-Bags.com, E-Source, Honeywell, Intel, LG Electronics, Medtronics, ON Semiconductor, Phelps Dodge, Seaquist Closures, United Technologies, and Zytec. Contact: www.csom.umn.edu | [email protected]

1.2

How to Implement a Mass Customization Strategy: Guidelines for Manufacturing Companies Erlend Alfnes Department of Production and Quality Engineering, Norwegian University of Science and Technology, NTNU, Norway Lars Skjelstad Department of Operations Management, SINTEF Technology and Society, Norway

The attention on Mass Customization as a viable manufacturing strategy is increasing in academia. Also, more and more companies report from successful implementations. However, the transformation process necessary to become a mass customizer (from the company’s outset as a mass producer or a handcraft type industry) is still not fully developed, and research on practical implementations is needed to gain experience on how to proceed. The research presented in this paper is based on two case studies in the Norwegian furniture industry. Efforts towards the new strategy in both companies are analyzed related to central decision areas when implementing mass customization. The decision areas are extracted from literature. Three performance objectives; low cost, short delivery time and degree of customization are considered to be the order winning criteria’s, and it is argued that enterprises need to balance these performance objectives in their effort to realize mass customization. The lessons learned from the cases are structured in a set of guidelines for mass customization, which propose the necessary changes to undergo for a mass-producer as well as for a craft manufacturer.

Introduction The realization of a Mass Customization (MC) strategy is challenging for manufacturing companies. Even though the opportunity to mass customize has significantly increased with the latest innovations in information and communication technology, there is a limited number of studies that explicate how to design and operate manufacturing systems in order to utilize this opportunity. Today, "there is a relative dearth of research on how to design and operate a manufacturing system capable of mass customization" (McCarthy 2004). The purpose of this research is to present two successful implementations of the mass customization strategy in manufacturing companies, and based on these case studies to suggest a 44


45

set of guidelines to support mass customization. The case companies have shifted their strategy to mass customization from two typical outsets, one with background in mass production and one from the handcraft industry. These outsets are reflected in the suggested guidelines, which propose the necessary changes to undergo for a mass-producer as well as for a craft manufacturer. The paper is structured as follows. First, the mass customization research field is briefly reviewed from a manufacturing strategy perspective. Based on this review, a set of performance objectives and a set of decision areas are suggested for MC manufacturing. Second, the suggested performance objectives and decision areas are used to structure case studies of two successful mass customization transformations. Third, the findings from the case studies are used to suggest a set of guidelines for mass customization implementation. MC Manufacturing Strategy The problem of how to implement and operate new manufacturing strategies is a recurring and important theme in operations management (Skinner 1996). The mass customization strategy, as any manufacturing strategy, must aim to provide the performance objectives that are required for a certain market position. Furthermore, the strategic decisions that are made to achieve these objectives can be grouped together under a number of headings. In the field of operations strategy, these are usually referred to as decision areas or decision categories (Spring and Boaden 1997; Alfnes 2005; Beckman and Rosenfield 2008). In this section, the mass customization research field is briefly reviewed in order to suggest a set of performance objectives and a set of decision areas that can be used to structure and analyse the implementation of a mass customization strategy. MC Performance Objectives An enterprise may seek competitive advantage through generic strategies of cost leadership, differentiation, and focus (Porter 1980). The operations activity translates these advantages into at least four groups of performance objectives; flexibility, quality, cost and time (Skinner 1969; Hayes and Wheelwright 1984; Fine and Hax 1985, among others). These are the competitive criteria (or performance levels) that enable the products to qualify and win orders in the marketplace (Hill 2000). Several different performance objectives are suggested in the mass customization literature. There seems to be a general agreement that mass (efficiency, reliability, low costs etc.) and customization (variety, individualization, flexibility etc.) are order winning criteria for mass customization. For

46


example, Pine II (1999) defines mass customization as the "the ability to provide variety, and individual customization, at prices comparable to standard goods and services". This view, i.e. that mass customization mainly is about providing custom products at low cost, is also supported in publications such as (Haug et al. 2007; Piller et al. 2004; Duray 2004; Da Silveira et al. 2001; Reichwald et al. 2000, Alfnes and Strandhagen 2000). In this paper, quick response/rapid delivery is regarded as a third (order winning) criterion for mass customization companies. This view is in line with authors like Pine II et al. (1993); Feitzinger and Lee (1997); or Steger-Jensen and Svensson (2004). Mass customization must be able to deliver customized products quickly and at a low cost. Several other performance objectives are also important for mass customization. Leading mass customization companies are achieving low costs, high quality, and customized products (Pine II et al. 1993; Da Silveira et al. (2001)). Thus, quality can be viewed as another performance objective for mass customization. Precise delivery is also critical for the customer of customized products (Steger-Jensen and Svensson 2004; Alfnes and Strandhagen 2000). Other, more marketing oriented objectives for mass customizers are customer loyalty, rapid product development, individualized services, brand name etc. Based on this brief review of performance objectives, the authors suggest the following list of order winning and order qualifying performance objectives for mass customization. Order winning criteria include cost, customization, and responsiveness. Order qualifying criteria include quality, delivery precision, customer loyalty, rapid product development, individualized services, and brand name. All mass customization manufacturers should especially strive to improve their performance regarding cost, customization, and responsiveness. Figure 1 illustrates that the level of customization should be balanced with the level of cost and responsiveness. The level of customization will impact on cost and responsiveness, and should not exceed what is required in the market. Mass Customization Decision Areas Several authors have suggested characteristics, building blocks, conditions, enablers, or success factors that will enable a mass customization strategy. The topics that are highlighted regarding manufacturing are briefly reviewed below. Many authors (Piller et al. 2004; Silveira et al. 2001; Zipkin 2001; Tseng and Piller 2003; Reichwald et al. 2000; Kotha 1996; Pine II 1993, among others) seem to agree that mass customization is based on flexible manufacturing technology and information technology that enable manufacturing systems to deliver high variety products at low costs. Advanced manufacturing technology


47

(such as computer numeric control (CNC) machines, robots, flexible manufacturing systems (FMS), and computer aided design and manufacturing (CAD/CAM) systems) makes it possible to fabricate any type of product within a fixed solution space (Tseng and Piller 2003). This reduces the traditional trade-off between customization (variety/flexibility) and efficiency. Advanced information and communication technology (such as enterprise resource planning (ERP) systems, customer relation management (CRM) systems, product configurators, manufacturing execution systems, advanced planning and scheduling (APS), internet, and XML/EDI) make it possible to transform customer needs and physical measurements into a customer order that are processed individually through the entire order-cycle (order handling and planning, design, production, and distribution). Furthermore, the flexibility for mass customization is not only achieved through flexible technology that fabricates the product according to customer information, it is also achieved through a flexible work organization (Hart 1995; Alfnes and Strandhagen 2000; Pine II 1993). Especially in customer-specific operations, the workforce should be educated and reorganised to handle dynamic and changing environments (Alfnes and Strandhagen 2000).

Figure 1: The order winning criteria for MC manufacturers.

Several authors also seem to agree that, due to efficiency requirements, mass customized products are limited in their variety compared with purely customized products (Piller et al. 2004; Duray 2004; Swaminathan 2001; Gilmore and Pine 1997, among others). A limited customization of products is, according to (Duray

48


2004; Feitzinger and Lee 1997), achieved through a modular design based on standardized design elements or components that easily fit into a wide range of products. Such a streamlining of the product program is an important enabler to achieve sufficiently stable and predictable processes for mass customization (Tseng and Piller 2003; Duray 2004; Thoben 2003). Process performance can be further improved by streamlining and modularizing the total manufacturing system. Modular processes enables the company to store inventory in semifinished form (before full information about demand is realised) and later customize the product according to requirements. This enables manufacturers to achieve efficiency and shorten delivery lead times (Duray 2004; Swaminathan 2001; Alfnes and Strandhagen 2000; Feitzinger and Lee 1997; Pine II 1993; Mertanen and Sievänen 2007). The market interaction strategy for a group of products (also termed point of customer involvement, or customer order decoupling point) is a critical condition for mass customization (Duray 2004; Thoben 2003; Alfnes and Strandhagen 2000). The market interaction strategy decouples the manufacturing system into a customer-specific and customer-neutral part. Customer-specific processes are often time-critical and characterised by uncertainty and demand variety. Customer-neutral processes, on the other hand, allow a focus on stability and efficiency (Alfnes and Strandhagen 2000). The key to mass customization is therefore to postpone the task of differentiating a product for a specific customer until the latest possible point in the supply system (Feitzinger and Lee 1997). The market interaction strategy also plays a critical role in the selection of manufacturing planning and control methods (Berry and Hill 1992). Customer-specific processes are usually controlled by customer order-driven planning and control, while customer-neutral processes are controlled by forecast-driven planning and control. The role of planning and control in mass customization is discussed by several authors (Duray 2004; Reichwald et al. 2000; Alfnes and Strandhagen 2000). Some authors (Steger-Jensen et al. 2003; Swaminathan 2001, and others) argue that the increased variety associated with mass customization makes planning and control more difficult. Piller et al. 2004, on the other hand, argues that the shift from a traditional Make-To-Stock strategy to a situation where manufacturing and assembly is performed on-demand, simplifies planning and control. Mass customization is also linked to supply chain management. Manufacturers, retailers, and other value chain entities should collaborate efficiently through integrated supply chains (De Silveira et al. 2001; Margretha 1998; Reichwald et al. 2000; Kotha 1994; Feitzinger and Lee 1997; Tseng and Piller 2003; Zipkin 2001). Such integration of supply chains should be based on an interconnected


49

information network (Magretha 1998), and designed for cost-effective supply of components, and fast delivery of the finished, customized product (Feitzinger and Lee 1997; Reichwald et al. 2000). Based on this brief review, the authors propose a list of decision areas, which companies aiming for a mass customization manufacturing strategy need to address the following decisions areas: market interaction, product, ict, manufacturing technology, processes, manufacturing planning and control, supply chain integration, and work organization. In the next section, two cases of successful Mass Customization Manufacturing implementations are presented. The cases are structured according to the decision areas listed above. For both companies, the change process has been carried out over several years. Much of the success was based on innovative solutions that were developed and realised in research projects with NTNU/SINTEF. Based on the cases, a set of guidelines are proposed for each decision area. Case A: HÅG, a Swivel Office Chair Manufacturer About the company HÅG is one of the leading manufacturers of office chairs in Scandinavia. Most of their customers are located in Europe and US, and 80 percent of total sales is from exports. HÅG’s products (seating solutions for work, visit, and conference) encourage movement and variation, and are easily adjusted to individual needs. The strategy is to offer configurable and ergonomic products, with a distinctive and attnractive visual appearance. Important order qualifiers are short delivery time and high delivery precision. Motivation for choosing the MC strategy The demand for standard chairs with limited variation was declining. The new trend among the customers was to specify details and colors on their chairs, making it match their specific needs and office environment. HÅG’s manufacturing strategy however, was Make-To-Stock (MTS) with warehouses in many countries. This was an increasingly unsuitable strategy for meeting the new market demands. The finished goods inventory was large, and yet HÅG experienced many stock outs due to increased variations in demand. The situation generated extra costs and resulted in long lead times and low delivery precision. Compared to their competitors, HÅG had weak profitability and delivery performance. It was obvious for the company that changes were required in order to be competitive.

50


Changes In the following section, the transformation from mass production to mass customization manufacturing at HÅG is discussed according to the decision areas identified. 1. Market interaction HÅG’s MTS strategy became disadvantageous in the new market situation. Introducing an Assembly-to-Order (ATO) strategy was the first and most powerful change to become a mass customization manufacturing enterprise. By moving the customer order decoupling point (CODP) upstream in the value creating chain, the customer influence was increased and uncertainty reduced. The response time to customer orders was improved. 2. Products HÅG’s innovative design increasingly attracted customers that wanted to specify the fabric, color, casters, and other features on their chairs. An office chair consists of 9 major modules (foot, seat, back, headrest etc.) with numerous options of customer choice. A redesign of the product portfolio was carried out in order to improve customer options and reduce manufacturing complexity. The variant profile before and now is shown in Figure 2. The textile cover is the most visible component on the product, and therefore a component that customers want to personalize. It is also one of the easiest for HÅG to manufacture in a wide range, and hence the solution space for this component was increased dramatically, as shown in Figure 4. Possible choices for wheels, foot, seat base, lift, armrests and mechanisms were reduced to allow a more rational production, and to simplify the configuration process for customers. 3. Information and communication technology A new ERP-system and product configurator was implemented. Customer information (address etc.), order content (product specification) and delivery information (dates) are now directly entered into the web-based ordering and configurator system. Dealers and selected customers configure the product and place the order automatically. The information is then made available to HÅGs planners and suppliers in real time, and enables a synchronised production without unnecessary delay. The textile supplier is integrated in HÅGs ERP-system and cuts textiles based on HÅGs distribution plan for finished goods. On each set of textile (including covers for seats, backs, armrests and headrests) the supplier attaches a tag with a bar code generated from the ERP-system. This bar code is


51

subsequently used by operators at HÅG in the final assembly to print a productspecific picking, assembly and packaging list.

Figure 2: Number of variants per component of a swivel office chair.

4. Manufacturing technology The transition to mass customization did not imply heavy investments in new machinery. HÅG implemented a semi automatic conveyor system from the picking zone to the distribution area. The system was customized to their products, allowing every component a predefined position in the transportation system. The conveyor system ensures a batch size of one (singular chairs), a fast streamlined flow, and automatic feeding of assembly stations.

52


5. Processes A product oriented layout was established to improve space utilisation and flow through the factory. The manufacturing process was segmented in productfocused and coupled operations areas. Each operations area was a physically defined area with dedicated processes and operators, and served one or a few product families. The factory was divided into separate areas for component production, module assembly and final assembly. The sub-assembly operations were segmented in two areas, one for high volume products and another for lowvolume products. In the final-assembly, dedicated assembly lines were established for each product family. 6. Manufacturing planning and control Traditionally, all chairs were made to stock and controlled by MRP and forecasts. This kind of control made it difficult to cope with the variety in demand and the variety of configurations necessary to match customer demand. With the new solution, the chairs were assembled to customer orders and part production was mainly controlled by kanban cards. Figure 3 illustrates the different processes and the control principles used for different control areas.

Figure 3: Main processes and their control principles.


53

Final assembly is based on customer orders, where the sequence is roughly generated in the order registration phase. The sub-assembly processes are controlled by a kanban system. The task of operations in this area is to refill the component buffer in the order pick-zone, and work is performed independently of customer orders. Decentralised responsibility for prioritising jobs improves flexibility and responsiveness. Some of the early mechanical processes (cutting and stamping) as well as procurement of raw materials are still controlled by MRP and forecast. 7. Supply chain integration The supplier base was reduced from 120 to 80 companies, and JIT partnerships were established with most of the remaining suppliers. The supplier of textile covers was located in the same region as HÅG and was given a key role in the new solution. The supplier had online access to HÅG’s order system, and delivered customer specific textiles to the assembly lines at HÅG twice a day. The new solution also implied radical changes for distribution. All the distribution tasks and management were sourced from a single distributor. A time guarantee scheme defined (short) transport times and freight costs for Europe. The scheme should enable dealers to give an exact price and delivery date at early point of time. 8. Work organization The Assembly-To-Order strategy required that volume and mix variations were handled on a daily basis. Hence, the need for capacity in different assembly lines would fluctuate from day to day. Operators were trained to handle assembly of all product-families and to rotate between assembly lines. Effects The effect of the implemented mass customization strategy is illustrated in Figure 4. HÅG achieved radical improvement in several competitive dimensions. By improved product quality, increased flexibility, shorter delivery time, higher delivery precision, and reduced costs in production and distribution, the realisation of mass customization has been very successful for HÅG.

54


Figure 4: HÅG effects of implementing mass customization.

Case B: Hagen, a Wooden Staircase Manufacturer About the company Hagen is one of the ten largest wooden staircase manufacturers in Europe. The company serves mainly the domestic market where they have about 35 % of the market share. They are also present in the German and Danish market, and 9 % of the company’s total turnover is related to exports. The products are high-end staircases with numerous choices of wood, surface treatment, style and accessories, but also low-cost products with fewer options and more standardized solutions. For all products, physical dimensions must fit exactly to the specific staircase measures of the house. Each year, Hagen produces approximately 6,000 tailored staircases. The customers are both end-users (30% by volume) and construction firms (70% by volume). Hagen’s distribution is direct to the domestic market and via resellers to the international market.


55

Motivation for choosing the MC strategy Traditionally the staircase industry has been characterised by full customization, a high level of craft-work, and many manual operations. In the late 90’s, Hagen’s international sales of traditional staircases were declining due to increased competition from low-cost manufacturers. Hagen order-, production-, and delivery-processes were too cumbersome and work-intensive to compete with manufactures in low-wage countries such as Poland. At the same time, staircase customers were increasingly more concerned with style, aesthetics, and quality. The service they were offered regarding staircase specification, delivery, and installation also became more important for customers. These market trends forced Hagen to alter their manufacturing strategy to improve competitiveness. A shift was needed towards more automated specification and production processes, and shorter delivery time. The improvement activities that enabled Hagen to mass customize staircases with reduced costs and improved service are described below. Changes 1. Market interaction The majority of customers prefer staircases that are engineered and made specifically to their wishes and room dimensions. There are some examples of prefabricated staircases that can be adjusted to building dimensions, but this is still a very small market segment. Hagen therefore maintains their MTO strategy, but for several components the customer order decoupling point (CODP) is moved downstream in the value chain. The CODP has traditionally been positioned at the raw material stock. Today laths, poles and handrails are prefabricated in standard lengths, waiting for customer specific information to be cut in correct lengths and angles. This enables them to improve lead-time and reliability. 2. Products The product portfolio was streamlined to reduce complexity, and the different models and options were specified in an advanced product configurator. In the new portfolio, some traditional models were eliminated, and models with more advanced and functional design were introduced. However, the most important change was the rationalisation of components. A wooden staircase typically consists of laths, handrails, poles, sidewall strings and treads. A profile of the variation of the five main components before and after streamlining is presented in Figure 5.

56


Figure 5: Number of variants per component type for a staircase.

Each main component can be delivered with different style, type of wood and surface treatment, and with customer specific measures. Some variants for laths and poles only represented small nuances in design, and caused more confusion than real choice. The number of variants for these components was therefore reduced. Furthermore, the introduction of new modern models required that some components were added. The number of variants for the hand-rail component was therefore increased. Hagen’s staircases are now based on standard design elements and standardized rules. Through new constructions and designs, the number of components was reduced from 90 to approximately 70, reducing costs, need for machine capacity and overhead. 3. Information and communication technology Hagen has implemented a new ERP system, integrating the information flow from order handling, via procurement and construction, to manufacturing and invoicing. The different departments used to have stand alone tools for this, resulting in comprehensive data transactions with multiple sources of errors. The procurement process was partly automated through EDI transactions to central suppliers.


57

The company also implemented a product configurator and a web portal. These were integrated with the ERP-system and their existing CAD (Computer Aided Design) program, and made the specification and ordering process much more efficient. The customer builds a "prototype" of the staircase from a library of predefined design elements. Based on the prototype, the order handler constructs a rough CAD-design that can be quickly communicated back to the customer for acceptance. After acceptance, a detailed design with CNC programs and cut-lists is developed for production. 4. Manufacturing technology Hagen invested in CNC (Computer Numerical Control) milling machines for effective manufacturing of customized threads, sidewalls, and poles. Critical components are now fabricated fast and precisely according to customer approved sizes and angles. A new automatic surfacing line allows efficient painting of small series with different colors. There is still a lot of manual work in the factory, and further investments are expected to automate material handling, packing and shipping. 5. Processes Although new technology has been introduced in some of the operations, the overall manufacturing process was not significantly altered. The layout however, was changed from traditional job shop to a flow shop with "production lines" for every component family. CNC machines retain the necessary flexibility, and the same process route is kept regardless of type of wood and component shape. The logical and shortened material flow has increased the efficiency and reliability of the manufacturing process. 6. Manufacturing planning and control Traditionally, all components were made to customers' orders (craft production) in a rather unsynchronized way. A new control model was developed for Hagen in order to co-ordinate production lines, and to shorten the overall throughput time (Figure 6). The new control model was based on a decentralised push-pull scheduling system to synchronize the production of different components of a customer order, and to roughly keep the pace of the bottleneck. The bottleneck was usually the milling machines, but it might occasionally be the coating line. Since the product mix varies, criteria/rules used in the scheduling included delivery date, capacity utilisation, and serial effects on changeover time. By synchronising the flow, the

58


different staircase components of a customer order reached the packing zone simultaneously.

Figure 6: Main processes and their control principles.

7. Supply chain integration Hagen is involved in the entire process of supplying the customer with a new staircase. This includes initial measuring of the staircase dimensions in the house and the final installation process. Strategic partnerships with installers in all markets were established. These independent installers play a key role, being the prime contact link to the end-customer, both in doing the measurement of the staircase room and later, in doing the installation. In addition, an EDI-based partnership was established with local suppliers for frequent supply of baluster and steel components (standard components are replenished just-in-time and customized components are ordered).


59

8. Work organization Major changes were made to the order-handling and construction department. The existing order and construction process at Hagen was work intensive and involved several responsibility changes. Several employees had to involve themselves sequentially in the same order, and the lead time for an order was approximately two weeks. The process were reengineered and supported by IT. Now, personnel with different responsibilities are located together in teams and are all crosstrained to carry out a larger share of the process. One person has the total responsibility for a customer order and provides a single-point-of-contact for the customer. Effects Hagen set an example for handcraft companies that want to improve their competitiveness through an implementation of a Mass Customization strategy. The results are summarized in Figure 7, which illustrates how cost and time is improved and variability of components are reduced.

Figure 7: Hagens effects of implementing mass customization.

60


Implementation Guidelines for Mass Customization Manufacturing The two cases have highlighted some of the major changes or building blocks that are necessary to become a successful mass customization manufacturer. The necessity and appropriateness of these building blocks in a mass customization strategy is also confirmed in literature. Several authors have suggested characteristics, conditions, enablers, or success factors that are in line with the changes carried out by Håg and Hagen. In his study of National Bicycle Company of Japan, Kotha (1996) proposed a set of necessary conditions for success of Mass Customization. The conditions at the firm level can be summarized as; access to a supplier network in close proximity, increased product proliferation and new product introductions, interconnected information network, investment in advanced manufacturing technologies and human resource development, access to substantial in-house engineering expertise and a culture that focuses on knowledge creation. Reichwald et al. (2000) focuses on the information cycle of Mass Customization, and proposes a five step model to create interconnected integrated information flows. The model consists of five steps; listen to your customers, configuration, manufacturing planning, production and supply chain integration, and relationship management. In the article "The limits of Mass Customization", Zipkin (2001) states that there are three main capabilities that should be developed: elicitation (integrated information flows), flexible processes, and logistics. These three elements must operate well individually, but also as a "seamless whole". Piller et al. (2004) review various approaches to counterbalance the costs of mass customization. These are: an appropriate design for variety, product line planning, the use of a modular product family architecture, stable processes, high variety production planning and control, postponement strategies, and also specialised information systems for configuration, manufacturing planning, order tracking, and relationship management. Based on their study of more than 200 cases, they propose some additional approaches. These are: decoupling the value chain into a order-specific and a customer-neutral part, efficiency in forecasting and generation of customer knowledge, utilisation of the customer base to increase switching costs for the customer. Although the building blocks proposed in literature provide a general understanding of how mass customization can be achieved, there is still the need for a structured framework that explains how a mass customization strategy can be implemented. Based on the authors experience from the two case studies, a set of guidelines are therefore suggested for each of the decisions areas in a mass customization strategy (Some guidelines are only valid for (a) mass producers or (b) craft producers that aim to implement the mass customization strategy). The guidelines are given in Table 1.


61

Table 1: Set of guidelines suggested for key decisions areas in mass customization. 1. Market interaction Change the market interaction strategy to Make-To-Order or Assembly-To-Order (a) Aim to position CODP upstream in the value chain (b) Aim to position CODP downstream in the value chain 2. Products Offer high level of customization on components/modules that represent the highest added value to customers Make a product program based on similar design elements for all product families (a) Modularise components to enhance the variability for the customers (b) Standardise components to reduce the complexity for the manufacturing 3. ICT Establish online order registration Establish a product configurator Guide the customer through the order process and visualise the choices Strive for seamless integration of all information system (CAD/CAM, product configurator, ERP, order tracking, etc.) 4. Manufacturing technology Strive for automation in manufacturing, but balance it towards the flexibility obtained by human resources Utilize efficient technology in processes upstream of CODP Utilize responsive and flexible technology (FMS) in customer specific processes 5. Processes Establish a product oriented material flow Design a layout that reduces non value added processes Manufacturing processes should perform operations based on digitally transferred information about customer specifications 6. Manufacturing planning and control Introduce demand driven replenishment of standard components and modules Define and prioritise criteria for sequencing of orders in customer specific processes Aim to introduce pull-principle in processes upstream of CODP Aim to introduce push-principle (FIFO) downstream of CODP 7. Supply chain integration Establish JIT partnership with suppliers of standard components/modules Allow key suppliers of customer specific components online access to the order system Establish rapid distribution channels to all the markets areas 8. Work organization Train operators to be multi-skilled Educate operators in multiple tasks Develop a flexible job-rotation and job-allocation system

62


Conclusions This paper highlights that in order to achieve proper mass customization, it is not sufficient to offer customized products. The order winning criteria for customers also include delivery time and costs. Hence a balanced improvement of customization, cost-efficiency, and responsiveness is the only appropriate approach to mass customization. The mass customization challenges for mass producers and handcraft producers are to a large extent the same. However, the performance objectives will be different for the two outsets. Mass producers must increase the number of variants on selected components and allow customers specification, and at the same time maintain cost efficiency and lead times. Handcraft producers must restrict customer choice to a set of predefined design elements, and improve cost-efficiency and responsiveness. For both type of outsets, the realisation of mass customization is a strategic and comprehensive process that encompass all areas of the manufacturing system. The suggested MC implementation guidelines will support manufacturing companies in such a process. Acknowledgments We would like to thank logistics manager Ingvar Hagen at Hagen Treindustrier AS for his contribution to this paper.

References Alfnes, E. (2005). Enterprise reengineering: A strategic framework and methodology. Faculty of Engineering Science and Technology, Department of Production and Quality Engineering. Trondheim, Norwegian University of Science and Technology, NTNU. Doctoral thesis, 2005:153. Alfnes, E., Strandhagen, J. O. (2000). Enterprise Design for Mass Customization; The Control Model Methodology. International Journal of Logistics. 3(2): 111–125. Beckman, S.L., Rosenfield, D.B. (2008). Operations strategy: Competing in the 21st Century. McGrawHill. Berry W. L., Hill T. (1992). Linking systems to strategy. International Journal of Operations and Production Management. 12: 3–15. Duray, R. (2004). Mass customizers' use of inventory, planning techniques and channel management. Production planning and Control. 15(4): 412–421. Feitzinger, E., Lee, H.L. (1997). Mass customization at Hewlett-Packard: The power of postponement. Harvard Business Review. January-February, 116–121. Fine, C.H., Hax, A.C. (1985). Manufacturing strategy: a methodology and an illustration. Interfaces. 15(6): 28–46.


63

Gilmore, J., Pine, J. (1993). The four faces of mass customization. Harvard Business Review. 75(1): 91–101. Hart, C. (1995). Mass customization: conceptual underpinnings, opportunities and limits. International Journal of Service Industry Management. 6(2): 36–45. Hayes, R.H., Wheelwright, S.C. (1984). Restoring our competitive edge: competing through manufacturing. New York: Wiley. Haug, A., Ladeby, K., Edwards, K. (2007). Reflections on the transition from ETO to Mass Customization. Proceedings from the 2007 World Conference on Mass Customization and Personalization, October 7–10, MIT Cambridge: Boston. Hill, T. (2000). Manufacturing Strategy. Palgrave. Kotha, S. (1994). A Book Review of Mass Customization: The New Frontier in Business Competition by B.J. Pine II. Academy of Management Review. 19(3): 588–592. Kotha, S. (1996) From Mass Production to Mass Customization: The case of the National Industrial Bicycle Company of Japan. European Management Journal. 14(5): 442–450. McCarthy, I. P. (2004). Special issue editorial: the what, why and how of mass customization. Production Planning and Control. 15(4): 347–351. Mertanen, M., Sievänen, M. (2007). A Practical Approach to Mass Customization – Lessons learned from Finish Machine Construction. Proceedings from the 2007 World Conference on Mass Customization and Personalization, October 7–10, MIT Cambridge: Boston. Pine II, J. B. (1993a). Mass Customization; The New Frontier in Business Competition. Boston: Harvard Business School Press. Pine II, J. B., (1993b). Mass customizing products and services. Planning Review. 21(4): 6–13. Pine II, J. B., Viktor, B., Boynton, A. (1993). Making mass customization work. Harvard Business Review. 71, September–October, 108–119. Pine II, B.J. (1999). Mass Customization: The new frontier in business competition. Boston: Harward Business School Press. Piller, F.T., Moeslein, K., Stotko, C.M. (2004). Does mass customization pay? An economic approach to evaluate customer integration. Production Planning and Control. 15(4): 435–444. Porter M.E. (1980). Competitive Strategy: techniques for analysing Industries and Competitors. New York: Free Press. Reichwald, R., Piller, F.T., Möslein, K. (2000). Information as a critical success factor for mass customization or: why even a customized shoe not always fits. Conference proceedings, ASAC-IFSAM 2000 Montreal, Quebec, Canada. Silveira, G. D., Borenstein, D. & Fogliatto, F. (2001). Mass Customization: Literature review and research directions. International Journal of production Economics. 72(1): 1–13. Spring, M., Boaden, R. (1997). One more time: how do you win orders?: a critical reappraisal of the Hill manufacturing strategy framework. Int. Journal of Operations & Production Management. 17(8): 757– 779. Skinner, W. (1969). Manufacturing – missing link in corporate strategy. Harvard Business Review. 47(3): 136–145. Skinner, W. (1996). Manufacturing strategy on the S-curve. Journal of Production Operations Management. 5(1): 3–14.

64


Steger-Jensen, K., Svensson, C. (2004). Issues of mass customization and supporting IT-solutions. Computers in Industry. 54: 83–103. Swaminathan, J.M. (2001). Enabling customization using standardized operations. California Management Review. 43(3): 125–135 (Spring). Thoben, K.D. (2003). Customer driven manufacturing versus mass customization: Comparing system design principles for mass customization and (traditional) customer driven manufacturing. In: The customer centric enterprise Tseng, M.M., Piller, F.T. (eds) New York, Berlin: Springer. Ch. 5. Tseng, M.M., Piller, F.T. (2003). The customer centric enterprise. In: The customer centric enterprise. Tseng, M.M., Piller, F.T. (eds) New York, Berlin: Springer. Ch. 1. Zipkin, P. (2001). The Limits of mass Customization. MIT Sloan Management Review. 42(3): 81–87.

Author Biographies Dr. Erlend Alfnes is an adjunct associate professor at the Department of Production and Quality Engineering at Norwegian University of Technology and Society (NTNU). He is also a Senior Research Associate at SINTEF Technology and Society, Department of Operations Management. Alfnes has 10 years experience as a project leader and work package leader of national and international research projects. His research focuses on manufacturing planning and control, enterprise resource planning systems, and manufacturing strategy. Contact: www.smartlog.no | [email protected] Dr. Lars Skjelstad is a Senior Research Associate at SINTEF Technology and Society, Department of Operations Management. His dissertation was titled "Model and guidelines for realisation of mass customization". Skjelstad has 16 years experience of managing research and consultancy projects in the Norwegian manufacturing industry. He has also been the general manager of Intrapoint, a worldwide supplier of incident and crisis management software solutions. His research focuses on mass customization, lean production, socio technical systems, and the use of RFID in manufacturing. Contact: www.sintef.no | [email protected]

1.3

Media Market Inertia: A Potential Threat to Success of Mass Customization Detlef Schoder Department of Information Systems and Information Management, University of Cologne, Germany Johannes Putzke Department of Information Systems and Information Management, University of Cologne, Germany Kai Fischbach Department of Information Systems and Information Management, University of Cologne, Germany

Several empirical studies examine the consumer acceptance of mass-customized (MC) media. Although these findings overall suggest promising prospects for mass-customized media, entrepreneurial experience still contrasts these prospects with missing market take off. This paper sheds light on this antagonism. In the first part of this paper, we will discuss insights from studies examining consumers' acceptance of mass-customized products (particularly individualized printed newspapers). Based on entrepreneurial experience by the authors in setting up an individualized printed newspaper, the rich prospects of mass customization in content-related industries are contrasted with the lack of market take off. One of the authors holds the patent for an individualized, printed newspaper (WO03052648) and developed a prototype of an individualized printed daily newspaper that is ready for market launch. However, several factors impede market introduction. We tell the inside story and give some preliminary explanations. Conclusions will highlight scenarios of mass customization blended with Web 2.0 in content related industries.

Consumer Acceptance of Mass-Customized Products Several studies examine the consumer acceptance of mass-customized products. The first part of this chapter will give an overview over these studies. In the first section, the authors will highlight studies examining the consumer acceptance of mass-customized products in general (such as apparel, cars and personal computers). In the second section they will focus on the consumer acceptance of masscustomized media.

65

66


Apparel, cars and personal computers A number of studies examined the consumer acceptance of mass-customized products (see Piller (2003) and Franke and Piller (2003) for an extensive literature review of early empirical studies). Apart from a few exceptions (e.g. Huffman and Kahn 1998), these early studies generally suggest promising prospects for masscustomized products. Recent studies also indicate a large consumer acceptance of MC products (see the literature review below). For example, Franke and Piller (2004) conducted four experiments using students (n=717) as respondents. They elicited their willingness to pay (WTP) a price premium for mass-customized watches in comparison to comparable standard watches. In the first experiment, 165 students had to design their own watches and were asked about their WTP for their own MC watch as well as for several standard watches using contingent valuation methods. In the second experiment (n=248), some other students were asked about their WTP for the watches designed by the students in experiment 1 as well as several standard watches using contingent valuation methods. Experiments three (n=102) and four (n=202) repeated the same experimental design, but used Vickrey auctions to measure WTP (In a Vickrey auction, each bidder has to submit a sealed bid that is not known to the other bidders participating in the auction. The highest bidder wins the auction, but has to pay the price of the second highest bid only, which is an incentive to submit true value during the bidding stage.). Results indicate that a student’s WTP for his/her individual mass-customized watch is on average 100 percent higher than for a comparable standard watch. Bardakci and Whitelock (2004) questioned (1) 58 customers of car dealers in Manchester, UK, and (2) 97 persons who work full time at the University of Salford about their "readiness" for MC in context of the new car market. In a descriptive analysis, they find that almost 60% of the respondents are willing to pay a price premium for MC cars, 85% are willing to wait to receive a customized car, and about 85% are willing to spend time for designing a MC car. Piller and Müller (2004) report the results of two studies examining the consumer acceptance of mass customized footwear. In the first study, 420 consumers in Northern Europe (Germany, UK) and Southern Europe (Spain, Italy) were asked about their interest in MC footwear, as well as their WTP. Results indicate that people in Northern Europe generally have a larger interest in customized shoes than people in Southern Europe. Furthermore, women are found to be more interested in the customization of footwear than men (except in Spain). In addition, people either appreciate the concept of MC very much or completely reject it. For example, 41 percent (31 percent) of women (men) are very much


67

interested in customized shoes, whereas 33 percent (28 percent) are not at all. Concerning the WTP, 46 percent (42 percent) of men (women) are found to be willing to pay a price premium of 10–30 percent; 40 percent (35 percent) are willing to pay a price premium of 10 percent or less; and 12 percent (18 percent) of women are willing to pay a price premium of more than 30 percent. In the second study, 155 customers of a German (online) retailer of customized ladies shoes, and 213 female customers at the traditional point of sale (who were introduced to the online retailer) were asked about the reasons to buy customized shoes. Study findings indicate that design (style, color and heel) and custom fit are equally important for the purchasing decision. Dellaert and Stemersch (2005) examine the consumer acceptance of masscustomized personal computers. They conducted an internet-based experiment with 409 people out of a consumer panel at Tilburg University. The total population of the regular consumer panel is approx. 2000. However, Dellaert and Stremersch (2005) selected respondents that are older than 16 years that have interest in purchasing a PC in the next two year or have not purchased a PC in the past four years only. Respondents without internet access were provided with internet access by the general panel management if necessary The average age of the respondents is 43.7 years, and 37.2 percent of the respondents are female. 52.6 percent of the respondents hold a bachelor’s degree or higher. Using choice (or no choice) of a certain mass customization configuration (i.e. the outline or arrangement of different product components that can be mass customized such as number and levels of product modules) as dependent variable, they estimate an extended logit model allowing for heterogeneity in consumers' expertise and other unobserved factors (through a random coefficient specification). In the model, the utility of a certain mass customization configuration is expressed in latent variable equations as a function of product utility (that consumer can achieve by using the mass customization configuration) and complexity (i.e. perceived complexity of composing the product when using the mass customization configuration). Results indicate that mass customization configuration affects product utility and perceived complexity. Furthermore, product utility positively affects the utility that consumers derive from using a certain mass configuration, and complexity affects the utility that consumers derive from using a certain mass customization configuration, as well as product utility negatively. Furthermore, novices consider mass customization configurations more complex than do consumers with higher levels of product expertise. Schreier (2006) conducted three experiments with 187 students about their WTP for (1) cell phone covers (n=60), (2) T-shirts (n=63), and (3) scarves (n=64). The students (mean age: 23 years; 49 percent female) used different toolkits on a PC to

68


mass customize their products (without any time constraints). Afterwards, the students had to compare their individually designed product with the best standard product available (from a predefined list of 10 products). WTP was measured via a Vickrey Auction. Results indicate no differences in WTP for the standard products between the three subsamples, whereas the mean WTP for the masscustomized products was 134 percent higher for the total sample. In the case of (1) cell phone covers, mean WTP a premium for the mass customized product was 207 percent, (2) self-designed T-shirts mean WTP extra was 113 percent, and (3) self-designed scarves WTP extra was 107 percent. However, mass customization did not offer any value (WTP≤ 0) for 12 percent of the sample. In a recent study about consumer satisfaction with Dell laptop computers, Randall, Terwiesch and Ulrich (2007) examine which kind of toolkits (parameter-based vs. need-based systems) lead to a higher consumer utility (measured, amongst others, as perceived comfort and speed of the customization process, as well as fit of final product). In parameter-based systems, users specify the values of design parameters of the product directly (e.g. microprocessor type, microprocessor speed, memory size, hard drive size, and video processor type.). In contrast, in needbased systems they specify the relative importance of their needs (e.g. "I would like to do video processing", "My computer is light enough to easily hold it in one hand"). Afterwards, a recommendation algorithm proposes a product out of a combination of design parameters that maximizes customers' utility. In the study, 164 students (1) answered an initial survey about their demographics, computer expertise and experience purchasing computers, (2) virtually designed and "bought" a computer in an experiment with random assignment to the parameterbased system or need-based system, (3) filled in a second survey about user satisfaction with the toolkit as well with the configuration selected, and (4) were guided to a simulated showroom where they could revise their choices. In the virtual showroom, the users were shown 10 computers that could cover all physical dimensions of Dell laptop computers (different screen sizes, video cards etc.), posters about memory performance etc. and alternative configurations (cheaper / more expensive). The changes done by the users (measured as well in price change) were used as a measure of fit in the further analysis. Regression analysis and multivariate analysis of variance were applied to the data obtained through the experiment (130 useable cases; median age: 24 years; 36.7 percent female). Results indicate that (1) in the parameter-based systems, a higher expertise of the users leads to higher comfort, greater speed, and better fit of the final product. Furthermore, the advantage of the need-based system relative to the parameter-based system decreases in the user’s experience (for comfort and speed, less evidence for fit), (2) for non computer experts, the need-based system results


69

in better comfort and fit (measured as number and amount of changes in the showroom), but not in perceived fit or time spent during the customization process. Mass customized media Beside the recent studies suggesting promising prospects for mass-customized products in general (apparel, automotives, and personal computers), several authors examine the consumer acceptance of personalized media. Ihlström and Palmer (2002) question 153 readers of nine traditional Swedish newspapers in semi-structured interviews about their interest in personalized online news services, as well as about their WTP. They find that 22 percent of the readers are willing to pay for personalized news services, 31 percent are willing to pay for deepened news services, and 25 percent are willing to pay for coverage service of special events. In a large scale study, Franke and Steger (2006) analyze the consumers' WTP, purchase intention and attitude towards individually printed newspapers in comparison to segment-specific and one-size-fits all newspapers (containing 10 headlines each). The study consists of two samples (n= 1201 / 11,616; response rates= 70.5 percent / 13.7 percent). In the first sample, respondents expressed their preference for news by evaluating 90 headlines on a 5 point rating scale. Afterwards the authors conducted a latent class analysis and split the sample into 5 subsamples. Respondents of subsample A obtained a one size fits all newspaper. Respondents of subsample B obtained a segment specific newspaper (5 classes obtained in LCA). Respondents of subsample C obtained a (10-)segment specific newspaper, and respondents of subsample D obtained an individualized newspaper. Finally, respondents belonging to subsample E obtained all 4 newspapers. Afterwards, WTP was measured with dichotomous multiple contingent valuation, purchase intention was via a Juster-Scale, and attitude towards the product via a 3 item scale. Samples A-D were used for between-group analyses, whereas subsample E was used for within-group analysis. In the second sample, the experimental design was the same as for sample 1, apart from the segmentation via latent class analysis. In sample 2 respondents were assigned to the predefined clusters from sample 1 via minimizing the squared Euclidean distance. Results (using ANOVA and Fisher’s least square significant difference post-hoc tests) indicate that there is no difference between WTP for the standard product and the two segment-specific newspapers. However, WTP for the individualized solution is significantly higher. Furthermore, purchase intention (attitude toward the product) is found to be on average 17.2 (16) percent higher for the individualized

70


newspaper. Also OLS regressions using WTP, purchase intention and attitude as dependent variables and controlling for product involvement, age, sex, income and education indicate significant effects of preference match. During their own work, the authors conducted a large scale survey about the consumer acceptance of individually printed daily newspapers (see also Kaplan et al. 2007; Schoder et al. 2006). In the study, a panel of 2,114 respondents throughout Germany was questioned by several interviewers in one-to-one interviews. Results of a conjoint analysis (see Schoder et al. 2006) indicate that the type of newspaper (personalized / not personalized) is more important than the effort necessary for the customization process. However, only customers who are well-educated and belong to the upper socioeconomic strata are willing to pay extra for individualized newspapers. Furthermore, the results of a structural equation model estimated with componentbased techniques and a latent class analysis (see Kaplan et al. 2007) indicate that (1) base category consumption frequency has a positive effect on the likelihood to adopt MC products within this base category (2) base category need satisfaction has a positive effect on the likelihood to adoption, (3) base category consumption has a moderating effect in the context of theory of reasoned action and technology acceptance model (TAM). That means, the more frequently a subject consumes products out of the base category, the more important will be the impact of perceived ease of use mediated by perceived usefulness, and (4) different latent classes (with respect to unobserved heterogeneity regarding the latent variables need satisfaction and dissatisfaction) have different adoption behaviors (the higher the need dissatisfaction of the readers, the less interested they are in perceived ease of use in comparison to perceived usefulness). In a recent study basing on the same data set, the authors extended TAM by several antecedents to examine the underlying processes that drive the consumer’s adoption of mass-customized newspapers. Model findings (using covariance based structural equation modeling) suggest, amongst others, a moderating effect of gender, i.e. women might adopt mass-customized products if they perceive a relative advantage of the mass-customized product in comparison to the comparable standard product or substitution products available in the market, while men might adopt mass-customized products only if they help to fulfill a general yet unfulfilled need. Furthermore, the higher a consumer’s willingness to invest effort for a mass-customized product, the higher his / her perceived usefulness and perceived ease of use of an individualized newspaper. Finally, the greater a reader’s wish for "news serendipity" (i.e. the faculty of making happy and unexpected news discoveries by accident while browsing through a newspaper)


71

the lower his / her perceived ease of use. However, no effect was found from serendipity on perceived usefulness. That means that "serendipity" influences only the process of information retrieval (perceived ease of use), but not the expected outcome (perceived usefulness). In summary, from a theoretical perspective (see, for example, the extensive discussion about the feasibility of hybrid competitive strategies and the related literature in Piller (2003) and recent (game-)theoretical models such as Dewan, Bing and Seidmann (2003) and Syam and Kumar (2006)), as well as from empirical insights (see the literature above), one may conclude great prospects for mass-customized media products. However, the authors' entrepreneurial experience contrasts these rich prospects with still missing market take off. In the following case study, the authors tell the inside story and give some preliminary explanations. Case Study Medieninnovation.com GmbH In June 2003, one of the authors published a patent (WO03052648) for a printed product individualized for the customer (Figure 1) such as, for example, a daily newspaper. The process protected by the patent can be shortly described as follows: Customers transmit their (news) preferences via Internet, telephone, post, fax or other backward channels to the media enterprise. These preferences can be, for example, certain domains of interest, detail and proportion of reporting, follow-up reports, regional weather forecasts, personal daily agendas etc. In the media enterprise, the customer profiles are then matched automatically to news, ads and other content desired by the customers. A layout generator composes portable document format (pdf)-versions of the newspapers that are delivered electronically to local digital printing centers. After local printing, the paper versions of the newspapers are delivered directly to the readers' homes. A market survey conducted in cooperation with Allensbach Institute suggested promising prospects of an individually printed daily newspaper. More than 15% of the respondents stated that they were very likely to subscribe to such a type of newspaper. Hence, one of the authors and his co-inventor decided to commercialize the invention. Experiences from market-launch However, the acquisition of business partners at the supply side (content) was more problematic. Although the founders have a cooperation with the federal

72


association of German newspaper publishers (BDZV) and could convince some independent regional newspapers for collaboration, they could not convince a national operating, big publishing house that has a reputation of high quality, investigative journalism (like e.g. Springer or Bertelsmann). Neither, they could yet gain venture capital (to acceptable conditions) to set up an own editorial department that can compete with established publishing houses.

Figure 1: Patent for a printed product individualized for the customer (WO03052648).

During numerous presentations and negotiations with incumbents and venture capitalists, the authors disclosed several reasons that foster these stakeholders' scepticism towards an individually printed daily newspaper and impede market introduction.

First, the general trend of declining readership and diminishing advertising incomes let to a large need to economize, the consolidation of editorial departments and redundancies. Hence, the label "newspaper" evokes negative associations in most decision makers.

Second, many incumbents are afraid of cannibalization effects. The gains obtained from the provision of news for an individualized newspaper (perhaps even to a third party) probably might be lower than the diminishing returns due to a further declining readership of the traditional newspapers / magazines (particularly of special interest publications).

Third, the new technology is too innovative and a potentially disruptive technology making large parts of the knowledge obsolete that has been ac-


73

quired by the incumbents during a long period of time. Introducing an individualized newspaper into the market might even evoke new market entrants with more specialized knowledge imitating the idea.

Fourth, traditional (German) newspapers are not used to syndicate their content. Up to now, mainly news agencies and photo agencies took the role of content syndicators.

Fifth, the trialability of the product for the customers is low. A particularity of newspapers is their "dual market", i.e. newspapers have to fulfil the reader’s requirements as well as the requirements of their advertising customers. However, readers cannot "trial-read" the newspaper that is individually printed for another person, but have to go through the whole customization process for themselves. Also advertising customers do not yet know much about the efficiency of individualized advertising campaigns in daily newspapers.

Sixth, starting up an individualized newspaper is associated with high fixed costs before market introduction. These fixed costs are a high risk in the light of many uncertainties.

In summary, these six reasons make it difficult to communicate the value-added of an individually printed daily medium to decision makers. However, the next sections will illustrate the value-added of an individually newspaper and why it probably might be successful. Conceptual considerations Individually printed newspapers differ substantially from their traditional counterparts. The authors identified three main routes through which individually printed daily newspapers might differ from their traditional counterparts (1) combination of new content types through scripted questions (2) collaborative filtering and news production, and (3) self-booking and ad-placement. Combination of new content types through scripted questions In a traditional newspaper the content is selected by an editor. The content of an individually printed newspaper is selected by the reader from a variety of different conventional news sources as well as from new content formats. However, an individually printed newspaper is not just a re-compilation of given pdfs ("standard news"). Rather, content from conventional news sources (ticker, journals, editors, agencies) is enriched by content from the Internet (blogs, (RSS) feeds, newsletters, emails, alerts from, for example, google, ebay, amazon, IMDB), context information (calendar, geographical information), and targeted ads (cf. permission marketing). Furthermore, APIs (application programming interfaces)

74


allow the creation of new content types and value-added services through scripted questions. A nice example how such a content aggregation might look like are the services provided by yahoo pipes, netvibes, pageflakes and nilter on the Internet. For example, yahoo pipes has an introductory video on its webpage that illustrates how readers can combine internet pages in different formats (e.g. html, rss) to a pool of information (e.g. craigslist, boston globe etc.). Afterwards, the users can search and filter the information pool (e.g. apartments in Boston that cost less rent than $1000 a month). The extracted information can than be enriched by other contextual information (e.g. use Googlemaps API to display all suitable apartments in a map). Hence, readers can compose their individual newsclips and special interest compilations. However, many people do not want to read their "news" on a screen, but prefer a paper copy of their "newspaper". Hence, an individually printed daily newspaper might offer some value added for the readers through the possibility of combining various content types through scripted questions. Collaborative filtering and news production The possibility of scripted questions is not the only web 2.0 feature that might conquer print. Social Networking sites (e.g. facebook and linkedin), social bookmarking sites (e.g. del.icio.us) and social photo galleries (e.g. flickr) illustrated how users share content with their family, friends and all other internet users. An individualized newspaper can incorporate many of those features, of which (1) sharing user-generated content to specific peers, (2) collaborative filtering, and (3) collaborative entertainment might be the most important. First, sharing user-generated content to specific peers includes such features as, for example, sharing holiday pictures of family and friends, suggestions of new friends (i.e. friends of friends with similar interests), and telling your family and friends in which towns you will be during the next months for conferences etc. Second, collaborative filtering (e.g. Zan, Zeng and Hsinchun 2007) has been applied in an online context mainly to suggest the consumer products that are similar to consumers with similar consumption profiles. Amazon illustrates quite impressively how such recommendation mechanism might work. In the context of an individually printed daily newspaper, collaborative filtering might not only be employed for recommendations of books, music or films, but also for the selection of interesting news (after the users evaluated some articles). Third, collaborative entertainment such as gaming (for example playing chess against partners with a similar ELO), discussion boards, swapping interest forums, and collaborative story writing / song writing offer new ways of content development.


75

Self-booking and ad-placement The two last sections emphasized the benefits of an individually printed daily newspaper for the reader only. However, the last section also focuses on some additional value-added for the advertising customers. The detailed customer profiles collected during the customization process allow for very specific ad-placements. Similar to google adSense, a crawler can analyze the content of a newspaper and deliver ads that are matched to the reader’s interests. Hence, even the reader might find the advertisements useful. Furthermore, advertisers have the possibility to self-book advertisements that are printed as soon certain actions / queries are taken by the reader. For example, baby food and diaper manufacturers can take out ads as soon as some information about birth preparation is requested. In summary, the possibilities of self-booking and ad-placement allow better controlling, higher contact quality and less loss at campaigns. Lessons learned At the end of the case study the question arises which lessons can be learnt by other managers and inventors out of the authors' entrepreneurial experience. One of the most important lessons learnt during our negotiations with venture capitalists is that we should not label a printed mass-customized media product as "newspaper" (the same might be true for other mass-customized products such as customized CDs). The label "newspaper" evokes negative associations for most decision makers such as a declining readership trend, diminishing advertising income, need to economize, consolidation of editorial departments and redundancies. This statement is illustrated by the fact that the authors had much less problems in raising venture capital for their individualized information web portal nilter.com . However, a printed mass-customized medium is not necessarily affected by those trends. Rather, it is a disruptive print / hybrid product that overcomes with new features. First, the combination of different (traditional and new) content types through scripted questions, and collaborative filtering and news production develops new niche markets by serving "the long tail". Hence, the declining readership trend might be impeded. Second, detailed customer profiles allow for self-booking and ad-placements which probably translates into a much higher Cost Per Mille (CPM). Another important lesson learnt is that it is difficult to convince incumbents of the advantages of an individually printed daily newspaper. Many incumbents are afraid of cannibalization and channel conflicts, i.e. they fear not to gain new readers, but to lose old readers. However, this might not necessarily be the case.

76


Many families might want to continue having a common newspaper for the whole family and an individualized version only for some family members. Furthermore, many decision makers in newspaper publishing companies have a journalistic background. Shifting the editors' role from "agenda setters" (see also McCombs and Shaw 1972) to customer assistants, i.e. people that help readers that are not able to manage the customization process for themselves during the customization process, might contradict their self-concept of editors. Conclusions Many decision makers in the newspaper industry have not yet fully realized the value-added of individually printed daily newspapers for readers, advertising customers and media enterprises. Rather, they lapse into a common mindset and continue moaning about declining readership trends and diminishing advertising incomes. However, individualized newspaper may overcome many of those trends. The decision makers have to decide whether they prefer taking the risk of introducing a new (disruptive) technology into the market by themselves, or to take the risk to be put out of business by declining readership trends, diminishing advertising incomes and new market entrants realizing personalized media on the Internet. Nevertheless, understanding the factors that foster incumbents' scepticism toward mass customization is an interesting question for future research. We would suggest researchers to conduct large-scale empirical studies addressing this question quantitatively.

References Bardakci, Ahmet and Whitelock, Jeryl (2004). How "ready" are customers for mass customization? An exploratory investigation. European Journal of Marketing. 38(1112): 1396–1416. Dellaert, Benedict G. C. and Stremersch, Stefan (2005). Marketing Mass-Customized Products: Striking a Balance Between Utility and Complexity. Journal of Marketing Research (JMR). 42(2): 219–227. Dewan, Rajiv, Bing, Jing and Seidmann, Abraham (2003). Product Customization and Price Competition on the Internet. Management Science. 49(8): 1055–1070. Franke, Nikolaus and Piller, Frank (2004). Value Creation by Toolkits for User Innovation and Design: The Case of the Watch Market. Journal of Product Innovation Management. 21(6): 401–415. Franke, Nikolaus and Piller, Frank T. (2003). Key research issues in user interaction with user toolkits in a mass customization system. International Journal of Technology Management. 26(56): 578–599. Franke, Nikolaus and Steger, Christoph (2006). Segmentation or Individualization? An Empirical Analysis of Customer Value of Standard, Segment Specific and Individualized Products, paper presented to 4th International Workshop on User Innovation, Munich, July 13–14.


77

Huffman, Cynthia and Kahn, Barbara E. (1998). Variety for Sale: Mass Customization or Mass Confusion? Journal of Retailing. 74(4): 491–513. Ihlström, Carina and Palmer, Jonathan (2002). Revenues for Online Newspapers: Owner and User Perceptions. Electronic Markets. 12(4): 228–236. Kaplan, Andreas M., Schoder, Detlef and Haenlein, Michael (2007). Factors Influencing the Adoption of Mass Customization: The Impact of Base Category Consumption Frequency and Need Satisfaction. Journal of Product Innovation Management. 24(2): 101–116. McCombs, Maxwell E. and Shaw, Donald L. (1972). The Agenda-Setting Function of Mass Media. Public Opinion Quarterly. 36(2): 176–187. Piller, Frank T. and Müller, Melanie (2004). A new marketing approach to mass customization. International Journal of Computer Integrated Manufacturing. 17(7): 583–593. Piller, Frank Thomas (2003). Mass Customization: Ein wettbewerbsstrategisches Konzept im Informationszeitalter. Wiesbaden: Deutscher Universitäts-Verlag (Gabler). Randall, Taylor, Terwiesch, Christian and Ulrich, Karl T. (2007). User Design of Customized Products. Marketing Science. 26(2): 268–280. Schoder, Detlef, Sick, Stefan, Putzke, Johannes and Kaplan, Andreas M. (2006). Mass Customization in the Newspaper Industry: Consumers' Attitudes Toward Individualized Media Innovations. International Journal on Media Management. 8(1): 9–18. Schreier, Martin (2006). The value increment of mass-customized products: an empirical assessment. Journal of Consumer Behavior. 5(4): 317–527. Syam, Niladri B. and Kumar, Nanda (2006). On Customized Goods, Standard Goods, and Competition. Marketing Science. 25(5): 525–537. Zan, Huang, Zeng, Daniel D. and Hsinchun, Chen (2007). Analyzing Consumer-Product Graphs: Empirical Findings and Applications in Recommender Systems. Management Science. 53(7): 1146–1164.

Author Biographies Prof. Dr. Detlef Schoder chairs the Department of Information Systems and Information Management at the University of Cologne, Germany. Detlef Schoder was appointed reviewer to the German Parliament’s Lower House and is consultant to the European Commission. He was Visiting Scholar at Stanford University, University of California at Berkeley, and MIT and is on the editorial boards of several international journals covering e-Business. His research interests include Electronic Commerce, Ambient Intelligence, Social Network Analysis, and IT-based media innovations. He is co-founder of medieninnovation.com, a multiple award winning start-up for mass-customized information provision print and online. He was granted patents on "Individualized printed newspapers". Contact: www.wim.uni-koeln.de & medieninnovation.com | [email protected] Johannes Putzke is a PhD candidate and research associate at the University of Cologne. His research focuses on mass customization, social network analysis, complex system science and agent based modeling. In the context of mass customization, his analysis of the newspaper industry was published in The International Journal on Media Management. Contact: www.wim.uni-koeln.de | [email protected]

78


Dr. Kai Fischbach is an Assistant Professor with the Department of Information Systems and Information Management at the University of Cologne. He is also Associate Director of the Center for Applied Social Network Analysis (CASNA) in Cologne. Before entering his recent positions Cologne, he worked at the WHU Otto Beisheim School of Management (Germany) and has been a visiting scholar with the University of Illinois at UrbanaChampaign (USA). His research focuses on the formation of social and complex networks, open innovation, and the design of efficient information exchanges. Contact: www.wim.uni-koeln.de | [email protected]

1.4

Operationalizing Mass Customization – A Conceptual Model Based on Recent Studies in Furniture Manufacturing Emmanuel T Kodzi Jr. Operations Management, Strathmore Business School, Nairobi, Kenya Rado Gazo Wood Processing Group, Industrial Engineering, Purdue University, USA

The notion that mass customization (MC) can improve the competitiveness of the US furniture industry has implications for fundamental changes in the business models of furniture manufacturers. The industry has traditionally pursued a concept of competitiveness based on price rather than on value-delivery, which is more compatible with a MC strategy. Thus, a shift toward a MC strategy without a clear application roadmap is likely to have undesirable business consequences. In this study, we synthesize insights from previous research to conceptualize a value-delivery framework for making MC operational in the US furniture industry. We propose the "3P Operational Model" as a baseline for researchers and manufacturers exploring a link between MC and competitiveness. Pending empirical validation, we expect our model to have application for comparable manufacturing systems.

Introduction Mass customization (MC) is currently perceived as one of the means by which the US furniture industry can improve its competitiveness (Bullard and West 2002; Schuler and Buehlmann 2003). A seminar was held by the Forest Products Society in 2004 at Grand Rapids, MI, to examine U.S. competitiveness. According to Furniture Today Reports (2004), discussions at the seminar revolved around "manufacturing competitiveness and innovation, customer service, and historical perspectives on the U.S. and Chinese furniture industries". There was a call for the US furniture industry to focus on "innovation in design, marketing, distribution and mass customization", besides improving productivity. According to Oh et al. (2004), the current threats faced by the US furniture industry from less expensive imports is a signal to revise their strategies regarding consumer preferences; they view MC and speed to market based on understanding of consumer preferences, as key to the sustainability of the furniture industry. Lihra et al. (2008) discussed 79

80


MC in terms of its potential to offset the intrinsic production cost disadvantage of US furniture manufacturers. They observed a correlation between the level of product customization offered by four domestic furniture industry subsectors and their relative market success; and view MC as a successful business strategy for domestic producers competing with low-cost offshore manufacturers. Shultz (2006) also views the "continuing overall decline of growth and profitability" particularly of furniture manufacturers in the United States, as a challenge that can be overcome using a MC model. The sustainability potential of the MC model is also recognized by the financial sector. Venture Capitalists have been reported to directly support MC initiatives in furniture manufacturing. Two such companies partnered in the venture capital funding of Tennessee-based Smart Furniture Inc., which delivers customer-centric solutions for workspace and display configurations through a web-based interface to consumers in mass markets (Culp 2004). Furthermore, a recent study of the furniture industry in Indiana indicates that the average proportion of customized production output is expected to grow by 40% in the next 5 years (Iyer et al. 2006). These perspectives suggest that not only academics and governmental organizations, but also industrialists consider MC to be critical as a competitive strategy for furniture manufacturers in the US. However, for the trend toward a higher proportion of customized product manufacturing to impact the industry positively, current operational models must be reviewed. The traditional models have often focused on arriving at the lowest possible price, often at the expense of detailed customer preferences. According to Lawser and Schuler (2007), the tradition of making products to suit a company’s production process rather than to meet "changing customer tastes" is inconsistent with competitiveness. Individual companies will be unsuccessful at MC without a strong focus on delivering customer-value because the key to sustainability includes "creating ways to add more value" to both products and associated services as a way to counter the price appeal of imports (ibid). Thus the implementation of MC, if shown to be strategically applicable to contemporary US furniture manufacturing, must be initiated with the circumspection that a transition from a traditional price-based competition mode deserves (Kodzi and Gazo 2006). In this study, we examine fundamental business model changes that are required to implement MC successfully in furniture manufacturing. Because the study is essentially a conceptual synthesis of our understanding of MC and its connection with competitiveness, we first provide a background of price-based competition, make a case for value-delivery as an alternative competitive strategy, and clarify our approach to MC using previous definitions. By reviewing insights from previous research by the authors and others in furniture manufacturing, and from lessons in other industries, we conceptualize a value-delivery framework for


81

integrating MC into the operations of an existing manufacturing business. We then introduce the "3P Operational Model" and discuss the role of functional combinations of preparation, perception, and processing in a MC strategy, and propose it as a baseline for empirical research into links between MC and competitiveness. We conclude the paper with a discussion on research and managerial implications of the model. Conceptual Framework The subject of price-based competition in the US furniture industry has historic roots in the relative strengths of regional furniture clusters over time. Competitiveness in manufacturing was linked to geographic locations with lower-cost factors of production. While this approach initially served regional businesses well, sustainability was elusive. For instance, the competitiveness of cabinetmakers on the East Coast diminished as lower unit product costs were achieved by Midwestern factories (Ingerman 1963). Ohio State emerged as the leading center of furniture production, but was likewise surpassed by Michigan (Cater 2005). The local furniture industry in Michigan recognized the ephemeral nature of cost leadership and was spurred to redefine itself as a value-based furniture source rather than a cost leader. At the turn of the 20th century the focus on value, manifested through products displayed at the Grand Rapids Furniture Market, transformed the city’s reputation from a producer of cheap goods to "the Paris of Furniture Design" (Ames 1975; Hanson 1997). This pursuit of value-delivery and the necessary technology to create that value, enabled the furniture industry in Michigan to survive even though new manufacturing centers subsequently opened in North Carolina and Virginia (Sligh 2005). These two states later experienced massive plant closures largely in response to offshore competition (Quesada and Gazo 2006). The continued focus on value delivery can enhance manufacturing synergies for companies in Michigan and better equip them to respond to pricebased competition (Moore and Moore 2007). Though the dimensions of the competition are different today from what existed at the turn of the last century, there appears to be an overlap between the competitive market challenges currently confronting furniture manufacturers, and the historical development of furniture clusters in the US. Given the challenge of competing on price alone in a dynamic market, what lessons may be gleaned from a retrospective view of the furniture industry to inform the competitive strategies of contemporary manufacturers? One current dimension of price-based competition is the onset of globalization; locations with low-cost production centers outside the US must now be considered

82


as part of the furniture supply chain. Larger U.S. furniture manufacturers may try to beat the competition by consolidating their manufacturing operations and integrating imported products into their current product lines, or by relocating their manufacturing bases to low-wage countries (Lawser and Schuler 2007). The economic advantage derived from favorable factors of production does contribute to increased outsourcing and relocation of manufacturing overseas. Based on the high labor content in furniture manufacturing, it appears logical for US companies to set up operations in countries with low labor costs (Johnson 2004). Some large manufacturers may also transform their local operations into a retail mode and import enough furniture from offshore suppliers, if they have sufficient capital, or just import for redistribution to their existing customers. However, this response to globalization also implies that other organizations have the opportunity to outsource furniture from low-cost suppliers and sell directly to retailers. In this way, the competitive pressures on local manufacturers is intensified; and the growing market power of retail outlets further entrenches price-based competition, resulting in negative impacts on local furniture manufacturers. Another dimension of price-based competition in a mature industry like furniture is the commoditization of the product and the subsequent loyalty of customers to the price rather than to the product per se. Thus, it is easier to switch sources of furniture supply to the extent that furniture is designated as a commodity. The critical question, for which we evaluate historical parallels, is whether the ongoing substitution of offshore manufacturing/supply centers and the associated commoditization of furniture are based on long-term competitive strategy, or rather a knee-jerk reaction to price-based competition. Labor cost differentials between geographic regions cannot be sustained indefinitely, but to the extent that offshore production advantages persist, what business models are likely to preserve furniture manufacturing, or some aspects of it in the US — and for that matter, other countries with high labor costs? Schuler and Buehlmann (2003) suggest that amid extensive plant closures and offshore relocation in the last decade, some segments of the furniture industry survived mainly by adopting new business models including value-delivery. Through focusing on value-delivery, for example, the commodity perception discussed above is tempered, and furniture is portrayed as a means to harmonize utility, aesthetic, service and status attributes at end-use locations. It is our view that MC, as a unique manifestation of valuedelivery, resonates with the competitive strategy of the historic Michigan industry. In the next section we clarify our approach to MC and its relevance to the discussion on manufacturing competitiveness.


83

Adopted mass customization definition Several definitions of MC have been presented by various authors (Pine 1993; Alford et al. 2000; Mann and Domb 2001; Anderson 2003; Tseng and Piller 2003). We adopt the definition of Kodzi et al. (2007) that mass customization is "the fulfillment of customer-specific orders for defined segments of mass markets, at costs and lead times that communicate value rather than an associated penalty for personalization or order size". This definition captures the fundamental concept of offering products or services that better fit the needs of customers on a large scale, while accommodating the perspective of segmentation theory (Jiang 2000) – the view that heterogeneous markets comprise a number of smaller homogeneous markets. This view tones down the assumption of perfect heterogeneity of customers (as in custom manufacturing), and enables the provider of customized products to pursue its objectives (profit maximization, brand image enhancement, etc.) by crafting a tactical response to the demands of the smallest viable market segments. It makes business sense to define the broad needs of different market segments, and then provide the opportunity to customize orders based on specific customer preferences within that context. For example, it is conceivable that most early career professionals will not be included in the market segments that can afford high-end luxury cars or cruises. For this reason the configurations offered for such classes of products or services are not likely to be designed with early career professionals in mind. Notwithstanding such design discretion, there are several opportunities for customization regardless of the customer profile in a given market. Thus, customization is connected to the market positioning of a business, and restrictions that may arise from 'ability to pay' and related considerations, do not diminish the attribution of the terminology "mass" to the customization offer. In the specific case of furniture therefore, MC avoids the situation where customized furniture is sourced only for orders that are not time-sensitive and by customers who can afford the characteristically high premiums. Implicit in the adopted definition are also the strategic deployment of technology, improved manufacturing processes, and cultural change mechanisms which make MC cost effective for a business that welcomes customer participation in order fulfillment. With reference to the above discussion, how can a business sustain manufacturing competitiveness using MC principles? Competitive advantage An organization may achieve and sustain competitive advantage by leveraging its unique resources to take advantage of current and future business opportunities (Jenster and Hussey 2001). In this regard, the proximity of local US manufacturers

84


to a vibrant consumer market, and the increased importance of choice for the purchasing decisions of consumers suggest a logistics advantage that may be exploited using responsive manufacturing systems. These localized advantages may also be available to companies located in other countries where a critical mass of consumers is immediately accessible. Current advances in information technology and its interfacing with the manufacturing process offer complementary opportunities for furniture manufacturers (Noyelle 1986; Davis 1987; Gates 1995). However, the opportunities presented by increased customer choice also constitute threats for manufacturers that are unwilling to change their processes, because the demand for product variants adds significantly to manufacturing costs by increasing stock keeping units (SKUs), increasing setups and reducing average production runs. Capacity utilization targets that have been the traditional norm also result in excess finished goods inventory and force organizations into a "push mode". There is little room for reactive organizations to improve their competitive positions in markets characterized by increased turbulence driven by globalization and advancing technology (Huber 1984; Skinner 1985; Jaikumar 1986). On the other hand, by adopting alternative business models, local manufacturers can improve their capacity to respond quickly to uncertainty in the marketplace (Zhang et al. 2002), and develop a combination of distinctive competencies that position them to use increased customer choice to an advantage. A MC strategy increases the capacity of an implementing company to anticipate and adapt quickly to changes in the marketplace. Thus, in the context of market volatility, MC provides a strategic framework to develop sustainable competitive advantage, with payoffs in total cost reduction, customer satisfaction and organizational agility. The referenced payoffs will be elusive if customers in the US are unwilling to accommodate basic inconveniences such as participating in the configuration of their product preferences and waiting for the customized products to be delivered. For example, the use of an electronic interface in configuring product specifications is critical to the success of MC — whether this configuration is done at a physical point of sale or over the internet; and customers must be comfortable with relating to electronic renderings of their product preferences to order a customized product. In this regard, observed increases in e-commerce shipments may be a useful indicator of a predisposition towards "electronically-rendered" mass customized products, even though such online transactions also involve many "standard" products. E-commerce is forecasted by Forrester Research to increase very significantly in the US between 2005 and 2010 (Ji and VanBoskirk 2005). In the specific case of "furniture and related product manufacturing", the Annual Survey of Manufacturers indicates that e-commerce as a percentage of


85

annual shipments rose from 13.2 percent to 21 percent between 2003 and 2006. This growing percentage may be an indication of customers' willingness to use electronic media in purchasing furniture, and could be auxiliary to a MC strategy. Kodzi and Gazo (2006) also provide evidence indicating that US customers find it reasonable to wait longer for customized furniture than standard ones available in stores, within realistic thresholds. The business case for MC appears to be applicable within the context of local US furniture markets. It is useful in this discussion, to understand how MC has worked in other industries to derive implementation lessons for the furniture industry. Implementation insights from other industries Kodzi et al. (2007) note that an operational combination of modularity, agility, supply-chain integration and competitive cost, appears to be a key success factor in MC, based on an overview of MC applications in different industries (Pine and Pietrocini 1993; Kubiak 1993; Kotha 1996; Feitzinger and Lee 1997; Duray 2002; Fogliatto et al. 2003; Whitelock and Bardakci 2005). Applied to furniture, modularity in design may be expressed in terms of a basic framing for a wooden table, to which different profiles and dimensions of tops, legs, and other accessories may be attached without the need for structural adjustments. Agility refers to the responsiveness of a business in a dynamic marketplace; such businesses are focused on developing innovative ways to satisfy customer preferences in a given competitive context. This combination of attributes is a basis for developing the responsiveness, flexibility and adaptability needed in a mass customization strategy. Key requirements for implementing MC principles include, but are not limited to:

A crisp statement by the implementing company of what constitutes its key strengths and markets, and to what extent it is ready to invest in major process transformations such as re-engineering of manufacturing process, order flow patterns and user-interface (Alford et al. 2000).

A concerted effort to increase technological capacity and pursue integrated automated production that streamlines process design by using modules to increase flexibility and responsiveness within the manufacturing system (Pine et al. 1995).

A clear strategy under-girded by information technology (IT) that is used as a tool to interface seamlessly with all the unique units in the process (Strobel 2004) and a readiness of the entire supply chain to adjust to the changes that are necessitated by this strategic framework.

86


A case in point is Toyota’s unsuccessful implementation of MC in the early 1990s, resulting from the strategic inconsistency of retaining all the structures and systems of continuous-improvement organizations instead of adopting critical process transformations (Pine et al. 1993). Achieving MC essentially transforms a business (Alford et al. 2000). Therefore, to be capable of supplying unique goods and services to customers as a planned strategy without sacrificing cost control, product quality or speed, companies must actively redesign business processes for MC (Gilmore and Pine 1997). Toyota subsequently prepared for customization by introducing a "Dealer Daily" dealer-management network – one that integrates with a design-management and production-management system and has virtual links to dealership-management applications used by sales personnel and managers at car dealerships (Konicki 2002). Toyota took advantage of its strong dealership base to launch a new mass customization drive, and reported its then best-ever sales month in 47 years of business in the U.S. as May 2004 when sales exceeded $202m (North American-built vehicles accounted for 67 percent of sales). The brand of the company features significantly in its ability to attract customers to new products. Each customer has a concept of the product or service offering that can be expected from a given company, so customers may evaluate the decision to purchase a customized product not only by the uniqueness of the product offered but also by the reputation of the vendor. It would appear, then, that companies that are succeeding at MC have developed the ability to retain the technical attributes and core distinctive features of customizable offerings while accommodating a reasonable latitude of configuration options. Such competencies help to provide the desired flexibility in customization concomitantly with preserving the integrity of the business label. Therefore MC in the furniture industry must be evaluated for suitability in terms of whether manufacturers have developed the capacity and flexibility to respond to rapidly changing markets in a way that will preserve the integrity of their acclaimed distinctive capabilities. To superimpose the flexibility of customization on a valued brand, a company will usually require a new competitive business model, which requires fundamental changes and focused leadership to implement. The extent of organizational and process transformation required for successful MC will also be a function of which implementation mode is selected for a specific product or service offer. For example, MC will be implemented slightly differently depending on whether a furniture manufacturer chooses to interact directly with the individual customer, or operate through a retail chain. The implementation framework proposed by Gilmore and Pine (1997) has general coverage in terms of distinguishing between the levels of change associated with a


87

product and its representation, and suggesting four approaches (collaborative, transparent, adaptive, and cosmetic customization) to implementing MC (Da Silveira et al. 2001). Poulin and Montreuil (2003) also propose a model with 8 levels of mass customization (popularizing, varietizing, accessorizing, parametering, tailoring, adjusting, monitoring and collaborating) from a manufacturing and a retailing perspective to help clarify the input of customers at various stages of order fulfillment. Lihra (2005) adapted Poulin and Montreuil’s model to the assembled furniture industry as a basis for describing the relationships between personalization levels, customer involvement and manufacturing processes. He indicated that increasing personalization options also increases the complexity of factors needed to be managed by organizations. Our focus is on how to apply these implementation concepts without sharply increasing internal costs to manufacturers. Without preparatory process transformations such as lean manufacturing, the complexity arising from the level of information technology and advanced manufacturing needed to successfully implement and orchestrate MC, could be counter-productive. We now focus on some specific studies conducted in customized furniture manufacturing to develop our model for making MC operational. Implementation insights from conducted furniture studies Moore and Moore (2007) suggest that the practical application of MC is not as easy to achieve as the premise suggests. However, with reference to the experience of their Michigan-based company, they indicate that once MC implementation roadblocks are overcome, an efficient manufacturing model emerges that creates products only after actual demand has been captured. Based on previous MC research in the furniture industry that the authors and others have conducted, we distill key insights for MC implementation to guide the framing of our operational model:

Kodzi et al. (2007) investigated how process transformations impact MC capability for furniture manufacturers, by combining two case studies with an in-depth survey of selected businesses. The emerging prerequisites for adopting MC were related to information technology, lean manufacturing, labor skills, and production technology. The main challenges that needed to be overcome to implement MC successfully included reducing set-ups, reducing batch sizes, and managing the supply chain. It also became evident that manufacturers need to exercise discretion in the customization offer, taking cognizance of design and process constraints within the manufacturing system and finding the right balance between the proportions of production that are customized versus standard. Customization-capability is impacted not only by

88


organizational responsiveness, but also by organizational leadership and organizational learning.

Hauslmayer and Gronalt (2008) conducted exploratory analyses to test how different product mixes can result in distinctive outcomes in a production environment reconfigured for MC. Critical process changes include component modularization to allow for individual preferences to be configured. They used a case study of parquet flooring manufacturing as a reference for comparing results from a simulation study. Their research indicates potential reductions in inventory levels, and improvements in service rates with obvious opportunities for business.

Iyer et al. (2006) reported an expected increase in the level of customized production offered by furniture manufacturers in Indiana. This increase was associated with higher logistics costs, and dispersed customer locations and raw material sources. In other words, manufacturers recognized that they needed a more cost-effective way to meet the needs of customers within a widening range of locations. Focusing on "average" needs reduces service levels for a dispersed customer base, if this dispersion translates into more heterogeneity. A comprehensive systems approach is needed to support a MC strategy and lower internal costs associated with fulfilling customer-specific needs.

Kodzi and Gazo (2007) reviewed the implications of customer involvement in order determination, on the resources and capabilities of an organization. By the offer of customization, organizations invite customers to interact more directly with the manufacturing system. This interaction could result in either satisfaction or dissatisfaction depending, to a large extent, on the capabilities of the organization. A proactive technical service capability is required, especially in cases where customers are predisposed to opt for "expert guidance" in arriving at the product specification that best meets their preferences. For example, the technical implications of a customer’s choices need to be understood in real time to reduce hesitation regarding performance in service of a desired product configuration. Such customer hesitation could prevent the sale from being completed, with obvious repercussions on overall customization profitability. Whether the customization offer is over the internet or at a physical sales location, the organization must be adequately resourced to manage the aesthetic and technical combinations represented by each order, while minimizing the trade-offs between cost of service at the point of sale, and the satisfaction of the customer.

89


The insights discussed from the studies referenced above have implications for managing many interconnected factors in the context of delivering customerspecific value in a mass market with dynamic preferences. In the next section we conceptualize and propose a model for making MC operational and sustainable in this context of value-delivery. Model Development Value-delivery implies that furniture offerings do provide variety, good fit, and a measure of individuality for potential end-users. Consequently, distribution channels require a wider variety of smaller lot sizes to maintain high service levels for a heterogeneous customer-base. Figure 1 illustrates the forces that interact within the market environment of furniture manufacturers, and indicates the need for a strategic framework to develop responsiveness to demands of current and future customers.

Corporate Expectations Supplier Networks Pr ic

New

e

Co mp et it

Purchasing Experiences Social Networks

Needs Preferences Specifications ion

Needs Preferences Specifications

RETAILER

CUSTOMER

Sh

or

ea rl te

m ono

ion

More Variety

New Opportunities

P

es tim

at aliz son Per

Capa bilitie s

n Ec uctio rod

d

ic s

Sma ll

lot s iz

es

MANUFACTURER Market Forces Company Boundary Market Environment

Figure 1: Market forces driving customization-related transformations.

90


Buyers of customized furniture have a concept of what thresholds are reasonable in terms of lead time, for example, in assessing the match between their needs, preferences and specifications, and the product offering. These customer expectations are influenced by their networks and by purchasing experiences in other industries, such that furniture manufacturers must articulate a credible customer focus, when the demand patterns and specifications are variable. Figure 2 highlights three pillars of the framework for engaging the critical process transformations antecedent to crafting an effective response to the indicated pressures – preparation, perception, processing. We discuss these three functions in turn.


New

e

Co mp

Purchasing Experiences Social Networks

PERCEPTION

Needs Preferences Specifications et it

ion

Branding (Value distinctives)

RETAILER


CUSTOMER

Invitation to customers (Service capability; unique experience, innovation)

Sh

r te or

Configuration Process Capa b

ilitie

tim

es

a aliz son Per

(Options, Complexity, Experts)

Customer Input

s

d l ea

(Willingness to: express needs, invest time & wait)

tion

Order Software

New Opportunities Process Transformation (Lean imperative)

P

uc rod

Ec tion

m on o

ics

Product Architecture (Modularity, Critical Features)

Backend Research (Materials, Performance)

Supply Chain Support (Standardized material , Corp. Structure HR & culture, IT) Info. Technology PREPARATION

Order Processing Engineering detailing Tracking Systems Set-up reduction Advanced Manufacturing Material requirements (sort, grade, match)

Batch processes Cell-based Quality Feedback Team approach Cost Control Outsourcing

More Variety

Sma ll

lot

siz e s

PROCESSING Market Forces Company Boundary Market Environment

Figure 2: Principles of responding to changing market pressures.

(i) Preparation: Product families are redesigned on a modular platform to offer more options to customers, to increase flexibility within the manufacturing system, and to take advantage of scope economies. MC gives significance, not only to transformations that prepare the company to adopt technology-related


91

capabilities, but also to reengineering the entire manufacturing system. Information technology is recognized as an interface tool for coordinating the unique customization process. Responsiveness in the supply chain is also pursued to minimize unprofitable inventory flows to achieve high service levels without significant increases in internal costs. (ii) Perception: The business (re)defines its market identity — the distinctive features of the company’s offering are clearly recognizable as having the potential to satisfy individualized needs. The possibility of configuring specific products forms the basis for inviting customers to participate in the co-design process. Being a consumer durable, customers are likely to think through how a particular furniture model or feature might impact their everyday work or leisure. Thus, the invitation to participate in defining the product outcome positively affects customer perception of the company’s offering, and increases the likelihood of longer-term satisfaction with the products they have configured. The ability of a manufacturing business to retain the technical attributes and core distinctive features of a customizable offering helps to preserve the integrity of the business label while allowing customers the flexibility to configure the components in such a way as to meet specific needs. A service capability that is commensurate with the anticipated levels of direct customer involvement and collaborative interaction will also encourage customer participation in the codesign process. (iii) Processing: The business aligns the processing of customized orders with principles of MC — modularity, agility, supply-chain integration and competitive cost — to enable products to be manufactured on demand, and at a profit. For example, order software is updated to capture customer preferences accurately and transfer them seamlessly to a system that verifies the engineering detailing and tracks the order through team-based, process-oriented cells. Figure 3 demonstrates the non-trivial interaction of "Preparation", "Perception", and "Processing" for making MC operational in a manufacturing organization facing multiple pressures such as in the specific case of furniture manufacturing in the US discussed previously. The organization is more receptive to customer involvement, and the inherent learning is leveraged to foster long-term competitive resilience, through developing responsiveness, flexibility and adaptability. Taken together, the elements of preparation, perception and processing can support the investment decisions of furniture manufacturers that seek to assert their relevance in their current proximate markets.

92

HANDBOOK OF RESEARCH IN MASS CUSTOMIZATION & PERSONALIZATION Purchasing Experiences Social Networks


New

e

Co mp

PERCEPTION

Needs Preferences Specifications et it

ion

Branding (Value distinctives)

RETAILER


CUSTOMER

Invitation to customers (Service capability; unique experience, innovation)

Sh

r te or

Configuration Process Customer Input (Willingness to: express needs, invest time & wait)

Flexibility New Opportunities Process Transformation

Order Software

Product Architecture ucti Prod

(Modularity, Critical Features)

Backend Research

Supply Chain

Adaptability More Variety

ORGANIZATIONAL LEARNING

(Lean imperative)

s omic con on E

Order Processing Engineering detailing Tracking Systems Set-up reduction Advanced Manufacturing Material requirements

Sma ll

(sort, grade, match)

(Materials, Performance)

Support (Standardized material ,

Corp. Structure HR & culture, IT) Info. Technology

es tim

n atio aliz son Per

(Options, Complexity, Experts)

Capa bilitie s

d lea

Batch processes Cell-based Quality Feedback Team approach Cost Control Outsourcing

lot s izes

Responsiveness

PREPARATION

PROCESSING Market Forces Company Boundary Market Environment

Figure 3: Kodzi’s 3P operational baseline for mass customization.

Discussion and Conclusions By setting out the interdependencies between preparation, perception and processing in an MC context, the 3P model contributes to the discussion on bridging the gap between the strategic and operational considerations of MC (Åhlström and Westbrook 1999). This model provides a checklist for planning a MC strategy, and a baseline to empirically explore the links between mass customization and the competitiveness of furniture manufacturing in the US. If the operational principles are considered together, MC can provide a sustainable competitive advantage for implementing companies. Being a conceptual study, the 3P model requires some empirical validation. At this point, the model may also not completely silence concerns related to the liability of the manufacturer when newly configured products are delivered to customers without the traditional furniture testing procedures. Even though the customer is directly involved in the order fulfillment, the manufacturer does exercise discretion as the "expert", such that non-plausible combinations or parts


93

or features are eliminated. It is our view that since the product architecture is transformed, it is possible to test the performance of modules of product families separately, and then to simulate the performance of combined modules electronically. For this reason we also advocate delineation of the product configuration space based on distinct possibilities of module combinations to assure product integrity in collaborative furniture customization. Future research will address the extent of configuration possibilities offered by a manufacturer, and how that might impact the satisfaction of customers of furniture. To the extent that outsourcing and offshore manufacturing may strategically enhance the responsiveness of the furniture supply chain, rather than completely substitute domestic production, MC can help to retain local manufacturing profitably. Other things being equal, the collective strength of businesses implementing MC will improve the domestic market position of the furniture industry relative to imports. A stronger domestic positioning is also a plausible outcome in markets exhibiting similar characteristics as the US. As in the historical analogies previously discussed, value-delivery is superior to price-based competition in a dynamic market. Through MC, a strategic framework is made available for sustainable value-delivery.

References Åhlström, P., and R. Westbrook (1999). Implications of mass customization for operations management. International Journal of Operations and Production Management. 19(3): 262–274. Alford, D., P. Sackett, and G. Nelder (2000). Mass Customization: An Automotive Perspective, International Journal of Production Economics, 65(1). Ames, K. (1975). Grand Rapids Furniture at the Time of the Centennial. Winterthur Portfolio. 10, 23-50. Anderson, D.M. (2003). Mass Customization, the Proactive Management of Variety, CIM Press www.build-to-order-consulting.com. Annual Survey of Manufactures (2004-2006), U.S. Census Bureau. Accessed June 30, 2008 from www.census.gov/eos/www/2005/2005tables.html. Bullard, S. H., and C. D. West (2002). Furniture manufacturing and marketing: Eight strategic issues for the 21st century. Forest and Wildlife Research Center, Bulletin FP 227, Mississippi. Cater, J. J. (2005). The Rise of the Furniture Manufacturing Industry in Western North Carolina and Virginia. Management Decision. 43(6): 906–924. Culp, S. (2004) "Smart Furniture Closes Venture Capital Funding" Accessed July 5, 2008 at www.chattanoogachamber.com/newsandvideo/smartfurniture_09_04.asp. Da Silveira, G., D. Borenstein, and F.S. Fogliatto (2001). Mass Customization: Literature Review and Research Directions. International Journal of Production Economics. 72(1): 1–13. Davis, S. (1987). Future Perfect. Addison-Wesley Publishing Co. Inc, MA.

94


Duray, R. (2002). Mass Customization Origins: Mass or Custom Manufacturing? International Journal of Operations & Production Management. 22(3): 314–328. Feitzinger, E., and H. Lee (1997). Mass Customization at Hewlett-Packard: The Power of Postponement. Harvard Business Review. 75(1): 116–121. Fogliatto, F. S., G. J. C. Da Silveira, and R. Royerz (2003). Flexibility-driven Index for Measuring Mass Customization Feasibility on Industrialized Products. International Journal of Production Research. 41(8): 1811–1829. Gates, B. (1995) The Road Ahead. New York: Viking. Gilmore, J.H., and B.J. Pine (1997). The Four Faces of Mass Customization. Harvard Business Review. 75(1): 91–101. Hanson, N. L. (1997). The Furniture City. The Journal of American History. 84(3): 1003–1009. Hauslmayer, H, and M. Gronalt, 2008. Mass customization in the woodworking industry – a simulation study for the parquet flooring industry. International Journal of Mass Customization. 2(3/4): 179–199. Huber, G. P. (1984). The Nature and Design of Post-Industrial Organizations. Management Science. 30: 928–951. Ingerman, E. A. (1963) Personal Experiences of an Old New York Cabinetmaker Antiques. 84(5): 580. Iyer, A., S. Sommer, A. Thompson, and J. Mikals. (2006) Indiana Furniture Supply Chain Project, Final Report to the Indiana Department of Transport. Krannert School of Management, Purdue University, IN. Jaikumar, R. (1986). Postindustrial manufacturing. Harvard Business Review. Nov–Dec, 69–76. Jenster, P., and D. Hussey (2001). Company Analysis: Determining Strategic Capability John Wiley & Sons Ltd., England Jiang, P. (2000). Segment-based mass customization: an exploration of a new conceptual marketing framework Internet Research. 10(3): 215–226. Johnson, P. B. (2004). Furniture job erosion offers no signs of waning in North Carolina Knight Ridder/Tribune Business News. Accessed 10/2006 at http://tinyurl.com/nc36e8. Kodzi Jr., E.T., T. Lihra, and R. Gazo, (2007). Process Transformation Mandates for Manufacturing Customized Furniture, Journal of Forest Products Business Research. 4(8). Kodzi Jr., E.T., and R. Gazo, (2007). A Study of Mode-Specific Interactions between Customers and Manufacturing Systems in Mass Customization Proceedings, 2007 World Congress on Mass Customization, Boston, MA. Kodzi Jr., E. T., and R. Gazo (2006). Mass Customization in Practice – A Strategy that Wood Products Manufacturers can no longer ignore. Wood Digest. 37(7): 44–46. Konicki, S. (2002). Toyota Paves the Road to Customization. Information Week 6/2002. Accessed 01/2007 at www.informationweek.com/story/showArticle.jhtml?articleID=6502226 Kotha, S. (1996). From Mass Production to Mass Customization European Management Journal. 14(5): 442–450. Kubiak, J. (1993). A Joint Venture in Mass Customization. Planning Review. 21(4): 25. Li, C., and VanBoskirk, S. 2005. US Online Marketing Forecast: 2005 To 2010. Forrester Research Document. Accessed June 30, 2008 from www.forrester.com/Research/Document/Excerpt/ 0,7211,36546,00.html. Lawser, S. and A. Schuler, 2007. Operating Strategies for U.S. Furniture Manufacturers. Wood Digest, July 2007. Accessed July 5, 2008 from www.allbusiness.com/retail/retailers/ 10584859-1.html.


95

Lihra, T. (2005). Mass Customization of Wood Furniture: Literature review and application potential. Forintek Canada Corp. Lihra, T., U. Buehlmann, and R. Beauregard, 2008. Mass customization of wood furniture as a competitive strategy. International Journal of Mass Customization. 2(3/4): 200–215. Mann, D. L., and E. Domb (2001). Using TRIZ to Overcome Business Contradictions: Profitable ECommerce. Proceedings of TRIZCON2001, The Altshuller Institute, March 2001. Moore, K., and B. Moore, 2007. Re-inventing Manufacturing in Michigan. Blog entry on for metromodemedia.com published on Design Democracy. Accessed July 5, 2008 at www.designdemocracy08.com/abstract. Noyelle, T. (1986). Economic Transformation, Annals of the American Academy of Political and Social Science, Revitalizing the Industrial City. 488: 9–17. Oh, H., S. Yeon, and J. Hawley, 2004. What virtual reality can offer to the furniture industry. Journal of Textile and Apparel, Technology and Management. 4(1). Pine B. J. (1993). Mass Customization: The New Frontier in Business Competition. HBS Press. Pine, B. J., and T. Pietrocini (1993). Standard Modules allow Mass Customization at Bally Engineered Structures. Planning Review. 22(4): 127–137. Pine, B. J., D. Peppers, and M. Rogers (1995). "Do You Want To Keep Your Customers Forever?" Harvard Business Review, March 1995. Poulin, M., and B. Montreuil (2003). Implications of Personalization Offers on Demand and Supply Network Design: A Case from the Golf Club Industry, IEPM2003, Porto, Portugal. Quesada, H. J., and R. Gazo, 2006. Mass layoffs and plant closures in the U.S. wood products and furniture manufacturing industries. Forest Products Journal. 56(10): 101–106. Schuler, A., and U. Buehlmann (2003). "Identifying future competitive business strategies for the U.S. furniture industry: benchmarking and paradigm shift". USDA Forest Service General Technical Report NE-304.Northeastern Research Station. Newton Square, PA. Shultz, D. 2006. Competing in a commodity world: The Business Perspective of Mass Customization. Wood Digest, January, 2006. Skinner, W. (1985). The taming of lions: how manufacturing leadership evolved, 1780-1984. In K. B. Clark, R. Hayes and C. Lorenz (eds), The Uneasy Alliance: Managing the Productivity-Technology Dilemma (Boston, MA: Harvard Business School Press): 63–114. Sligh, R. (2005). Free Trade and the Future of Furniture. Acton Institute for the Study of Religion and Freedom. Accessed 01/2007 at www.acton.org/ppolicy/comment/article.php?id=244 Strobel, R. (2004). Motorola Paging Products Group, Boynton Beach, FL, in Oct, 1994 Issue of CIO Magazine. Accessed 01/2006 at www.cio.com/archive/101594/mass.html. Tseng, M. M., and F. T. Piller (2003). The Customer Centric Enterprise: Advances in Mass Customization and Personalization. Springer, Verlag, Berlin. Whitelock, J., and Bardakci, A. (2005). A Comparison of Customers' Readiness for Mass-Customization: Turkish vs. British Customers. School of Management, University of Bradford, UK. Accessed 10/2006 at http://tinyurl.com/l5vmsn. Zhang, Q., M. A. Vonderembse, and J. Lim (2002). Value chain flexibility: a dichotomy of competence and capability. International Journal of Production Research. 40(3): 561–583.

96


Author Biographies Dr. Emmanuel Kodzi Jr. is a Senior Lecturer at Strathmore Business School. He has been actively involved in both the world of business in various managerial roles; and in academic research and teaching. Through his connection with both industry and academia he has maintained a focus on manufacturing competitiveness and, more recently, how it interfaces with supply chain management. His doctoral research at Purdue University examined Mass Customization as a driver of competitiveness for US furniture manufacturers. He served a postdoctoral year at the Krannert School of Management at Purdue University (USA), then as Assistant Professor of Operations Management at Ashesi University (Ghana) before joining Strathmore Business School (Kenya). He has published in peer-reviewed journals on process strategy, manufacturing competitiveness, and instructional effectiveness and presented in several conferences. His major areas of teaching are Operations Management, New Product Development and Supply Chain Management. Dr. Kodzi’s current research interests include logistics systems in emerging economies, and the interface between supply chain responsiveness and new product development. Dr. Rado Gazo is a Professor of Wood Processing and Industrial Engineering at Purdue University. He teaches several courses including Properties of Wood, Secondary Wood Products Manufacturing and Furniture Design for CNC manufacturing. When not in a classroom, he conducts research and technology transfer in value-added wood products manufacturing and industrial engineering areas. He often works as a consultant to furniture companies. His research interests include competitiveness of furniture manufacturers and application of industrial engineering techniques to forest products manufacturing. Dr. Gazo has worked with over 100 companies, published over 150 publications, and given more than 100 professional presentations on the subject of secondary wood products manufacturing. Dr. Gazo is an active member and served on boards of Society of Wood Science and Technology, Forest Products Society and Consortium for Research on Renewable Industrial Materials. Contact: www.agriculture.purdue.edu | [email protected]

1.5

Beyond Mass Customization: Exploring the Features of a New Paradigm Nicola Morelli School of Architecture and Design, Aalborg University, Denmark Louise Møller Nielsen School of Architecture and Design, Aalborg University, Denmark

Technological and organizational developments are stretching the capabilities of industrial systems, which are now able to address the needs of smaller and more diversified target groups. Mass customization is the expression of such effort to adapt large production systems to customized solutions. At the same time though, substantial transformations in the social and economic conditions of our societies are challenging the basic assumption of the existing production systems. This is creating extreme differentiations in demand patterns and changing the role of customers in the production process. Mass customization may not be sufficient to address such changes. The authors of this chapter propose a new perspective in which customers have an active role in the value creation process via highly individualized and localized solutions. Such a perspective would change the role of industrial companies and the nature of their offering. This paper outlines the characteristics of the new perspective framework and explores some methodological directions for addressing new strategies.

Introduction Technological and organizational developments in industrial and economic systems in the last decades have triggered two apparently opposite trends: the first, towards globalization, is expanding markets and generating broad movements of production systems and facilities towards globalized solutions, whereas the second trend is creating new possibilities for individual people in local contexts to define sets of highly contextualized and personalized solutions. Although the contradiction between those two trends is not as dramatic as it appears, it presents the problem of the adequacy of the logical and organizational framework of industrial systems in the new situation. Industrial systems are now capable of supporting a high differentiation of the demand and the fragmentation of markets to smaller and smaller segments: industries are now targeting individu-

97

98


als, rather than groups. This is particularly evident in developed countries, where they can be supported by advanced services and infrastructures. However substantial changes in the structure of society, in lifestyles and in the infrastructures supporting economic systems are multiplying the diversities even in local contexts, thus creating a situation in which individual and highly diversified demand patterns represent a large part of the market. The question this situation poses is whether the existing production systems can satisfy such a heterogeneous demand without high costs, or whether the structures of the rules regulating such systems need to be revised: to what extent is mass customization integrating consumers as an active part of the production process? What is the capability of increasingly globalised production systems to address the needs of highly localized and individualized demand patterns? Such questions suggest a perspective shift that would emphasize new forms of innovation in which industrial solutions have some critical features (Manzini, Collina et al. 2004):

They are rooted on the network economy,

They are highly context sensitive in regard to both production and consumption aspects,

They include end-users in the production process; and

They allow for highly individualized solutions. The difference between the two perspectives may be seen as a paradigmatic shift or simply as an advanced stage of industrialization. The authors of this paper are well aware that the changes outlined above do not present any real elements of discontinuity that would clearly define a paradigm change. This paper will go through the progressive passage from integrated and mass produced products to highly customized solutions and refers to a paradigm shift only to help focusing on the fundamental elements of this epochal passage and to explore the potential of mass customization in this context. This paper will refer to the existing situation as the paradigm of industrial production and the emerging paradigm as the paradigm of highly individualized solutions. Two Perspective Views In the paradigm of industrial production the value creation process was conceptualized in terms of value chain (Porter 1985). According to this concept, value creation is not only sequential, but also implies that value is added along the production process, up to the moment in which the product is sold. In this framework, Ramirez (1999) observes that "customers were seen as destroying the


99

value which producers had created for them. [...] For producers, industrial value was 'realized' in the transaction which joined and separated them from customers." The new paradigmatic condition matured with technological advancements, which made work practices less linear and sequential. Distributed processing and concurrent engineering made the process of value creation more synchronic and interactive. This was the favorable condition to review the role of the customers. In the new framework value is no longer "added" until the point of sale – to be destroyed by the "consumers"- but is rather co-created by a network of actors, including customers, beside the traditional value producers (manufacturers, service providers). As it often happens during major paradigm shifts (Kuhn 1962; Arbnor and Bjerke 1997) the two paradigms are co-existing at the same time. However the ultimate presumptions supporting the old paradigm of industrial production, which could be very effective to interpret the logic of globalization, would probably not be sufficient to explain the emergence of new solutions that address localized and highly individualized needs. New methodological approaches emerged to correct the lack of explanatory powers of the old paradigm. Mass customization is probably one of those approaches. Here the rigidity of mass production was mediated by the new technological possibilities to diversify the offering, thus targeting very small and differentiated groups. The limit of the old paradigm, however, is in the persistence of industrial products as the link between producers and customers: in their attempt to provide a parsimonious definition of mass customization, Kaplan and Haenlein (2006) apply the term mass customization only to products. The two authors consider the definition of mass customization as "the act of integrating the customer in the value creation process to develop an individualized offering". This integration, they observe, is already inherently included in the services definition, therefore the use of the term mass customization in the context of traditional services would create a tautology. They propose, instead, to use "modularization" to refer to the delivery of highly individualized services at a lower cost. The definition of Kaplan and Haelein, derived from a study of the most relevant contributions in this area, is deeply rooted in the paradigm of industrial production, which focuses on production factors (cost reduction, flexibility, modularisation). Although this perspective recognizes the need for a better integration of the customer in the production process, it does not consider the implications of an active involvement of customers in any forms of co-production. Several authors (Pine, Peppers et al. 1995; Ettlie and Ward 1997; Zipkin 2001; Kaplan and Haenlein 2006) consider customers' integration in the production process as a way

100


of getting better information about customers' needs and preferences. Zipkin (2001) clearly explains how this process of elicitation should be integrated with flexible production processes and an adequate logistic. This perspective is still considering customers as served, therefore they are outside the real production process, whereas several cases are emerging (Von Hippel 2005; Tapscott and Williams 2006), in which customers' co-production demonstrates their capability to produce highly individualized and unique solutions. The authors reporting those cases have observed that they are undermining some critical elements in the logic of industrial production. This logic cannot be bent to individual solutions, unless a radical perspective shift is introduced. Such a shift has been (explicitly or implicitly) described as a shift from material production to value co-production (Ramirez 1999; Berger and Piller 2003), from value chain to value constellation (Normann and Ramirez 1994), from traditional productionconsumption systems to new systems which promote final users to the role of active co-workers (Manzini 2005; Morelli 2006; Morelli 2007).The perspective shift is therefore challenging the concept of mass customization, suggesting some fundamental questions about the validity of such concept, its characteristics and its implications in the new paradigm. The following sections will outline the progressive passages that brought about this perspective shift. Mass Customization and Product Architectures One factor that substantially contributed to the transformation from mass production to mass customization is the shift from a vertical/integrated industry to a horizontal/modular one. Fine (2000) describes such a shift as connected to a change from integrated product architectures to modular structures, which allow faster developments and frequent and profitable product upgrades (Fine 2000). Such a change is described as a "double helix". According to Fine (Figure 1), products begin their lives in integrated product architectures. In this phase manufacturers are exclusively using internal production capabilities. After some time, manufacturers will experience the pressure to disintegrate (modularize) the product architecture — in order to facilitate innovation processes, thus keeping up the fight against niche competition. The modularization also makes it possible to reduce the product complexity and to compensate for the organizational rigidity. Personal computers are an example of products, which have followed the loop of the "double helix" from an integrated to a disintegrated architecture. IBM, the first manufacturer of personal computers in the 80s initiated the disintegration of the computer architecture, which made it possible to produce the hard disk, the processor, the operative system etc. separately, and then assemble them afterwards


101

into a PC. This strategy also facilitated the integration of innovative components in the product architecture, without requesting any substantial changes.

Figure 1: The "double helix" (Fine 2002).

The process of disintegration and modularization of the product architecture in the PC industry made the concept of mass customization feasible. When the architecture of modules and interfaces was defined it was possible to customize the final offering by combining different components into a set of predefined options. This was the basis of the success of companies such as DELL. The process of disintegration and modularization, though, is not a final development stage of products' life. In the computer industry, for instance, some of the actors in the supply chain, such as Intel (processors) and Microsoft (operating system) were able to shift the focus from the single component to more integrated solutions. Windows operating system, for instance, was shifting the focus to the software (and use-related) components of a PC, thus proposing the integration of different functional units, such as web browsers, email, server operating system and multimedia contents. However it is worth noticing that the new integration happens at a different logical level, the hardware on which the whole process started becomes less relevant, while the integrated combination of software and

102


the possibility to offer a solution, rather than a simple product, becomes relevant. Seen from the old perspective of hardware manufacturers, such as IBM, this paradigm shift was initially not fully understood and caused serious strategic problems. Beyond the configurator The modularization of production is a necessary but not sufficient condition to support mass customization, nor the consequent aggregation to higher levels of integration. Once reduced to a modular structure, the company’s offering has to be supported by a configurator, that consists of a series of tools (including databases, examples, communication tools, toolkits) that help customers integrating the modules into a solution that satisfies their own needs. The economic conditions for the customization of products are defined by the configurator, which is the expression of the existing capability and degree of freedom built into a manufacturer’s production system. The configurator therefore defines the solution space in which the customers can search for a product that addresses their own needs. (Von Hippel and Katz 1998). The dimension and the customer’s freedom of choice in using a configurator are defined by the way modules are organized in the production system: it is larger when the user is able to manipulate and organize the basic elements (i.e. furniture components). Any customized product falling beyond the limits of a configurator would cost much more to the manufacturer, because it would require structural changes in the production system. The existence of several documented cases (von Hippel 2005; Tapscott and Williams 2006) in which users have customized products on their own, reveals a discrepancy between the solution space proposed by companies and the problem space identified by users. The problem space represents the specification of explicitly or implicitly defined needs of individual users. A broader definition of such a space would also consider desirable solutions, as an expression of needs that are not yet clearly defined or conditions that prefigure a value for the customers. The problem space expresses a request for solutions which are not necessarily satisfied by the products generated through a configurator. The discrepancy between solution and problem spaces defines the border line for mass customization. On one side of this line there are the characteristics and limitation of a production system, on the other side there are the customers' explicitly or implicitly defined needs as expressed before, regardless of the technical limitation given by the configurator. The two sides can also relate to two different kinds of knowledge (technical on one hand, tacit and user-related on the


103

other). The cost of stretching the solution space to a larger part of the problem space (thus integrating knowledge about new needs and desires into the solution space) would be too high for the manufacturer. In such cases the definition of the problem and its solution may be left to the main actors included in the problem space, i.e. to customers, who may be the best experts in identifying their own problems and developing their own solution (Von Hippel 2005). However, this would require a change in the role of the manufacturer, who, instead of proposing products, should offer support for users to develop their own solutions. Von Hippel proposes a toolkit, i.e. "user friendly" design tools that enable users to develop new product innovation by themselves. They are specific to the customers' design challenge and problematic area. Through the use of toolkits, users can create preliminary design, simulate or prototype it, evaluate its functioning and interactively improve solutions until satisfied. (Von Hippel and Katz 1998) In some cases the toolkit may introduce technical elements of the solution space, i.e. technology and knowledge derived from the manufacturer, in order to hook up the problem space to the solution space by generating consistency between the language of the customer and the technical language used in the solution space. (Von Hippel and Katz 1998). With respect to the configurator, the toolkit proposed by Von Hippel is a new step towards a perspective that sees customers as co-producers of their own solutions. In the customers' perspective indeed, the large variety of choices offered by a configurator is not an answer to their needs. They may in fact be disoriented by the wide range of choices available. The toolkit, on the contrary, is a form of empowerment of the customer, who can now focus exclusively on his own needs or expectations. The toolkit is in fact pointing towards a real customer-centered perspective: customers do not need products but rather highly individualized solutions: they do not need cars, they want to move from one point to another in the city; customers do not need money, but rather the capability of making their dreams real. The new perspective redefines the needs to a higher level, thus proposing new levels of integration. Highly individualized solutions involve new actors, beside the manufacturer, and increase the interaction of customers with the production system, giving them the chance to integrate different products and service at the local and global level. By doing this, customers will use all their own knowledge and skills to express their preference towards a particular producer/service provider. This value creation process may happen at the local level, through an exchange of knowledge based on geographical proximity, or at a trans-local level, by means of an exchange of knowledge through social networks that create logical proximities among members of communities that share the same interests. In all cases customers will

104


have a higher responsibility or even become the prominent actors in the value creation system. Such a large shift in perspective is already happening in many areas, from IT applications to kite surfing, from car sharing to Universities. It may not sound very appealing for manufacturers and service providers, but the increasing number of cases should encourage companies to have an active role in this process, rather than opposing it. What is happening in the real world has been anticipated on the web: after a first wave of applications on the internet, a new wave is now prevailing, which is assuming the applications of the previous wave as a commodity to activate social networks of exchange of knowledge. This is producing radical changes in the way the web is used. Likewise, in the material world, the first wave of industrial products is becoming the material support, but not the focus, of new trends in which:

Customers, rather than manufacturers, are the source of value creation

There is an intense cooperation between producers and customers

Such processes are intrinsically based on social and collective intuition

Production processes are being adapted to highly localized conditions, but their reproducibility relates to "trans-local" coincidences of social, cultural and economic conditions.

Defining the New Paradigm through its Characteristics The above mentioned trends can be seen as the root for a new paradigm condition that is going to change the way industrial companies and customers interact, in order to satisfy highly localized and individualized needs. Here below the trends will be analyzed in relation to possible strategies to support or address them. Customers, rather than the manufacturers are the source of value creation When shifting the focus from products to solutions, customers are generating new value by linking all the resources available in their own geographical or cultural context. IKEA customers, for instance, start designing their home with the help of the IKEA catalogue, but also by contacting friends who already bought the furniture they are thinking of buying. In the following stage they drive to IKEA, possibly borrowing car trailers or rooftop luggage holders from friends (recently from IKEA, too). They ask friends and family members to help them carry, transport and assemble furniture. The role of IKEA in this whole project is of course essential, but quite limited in time, if we consider the long process of


105

designing/planning, observing/ buying, transporting and assembling the furniture. The company however provides support in each of those phases: in the design phase the catalogue provides a visual support for the imagination (linking problem and solution spaces) especially for those customers who are not used to visualize the furniture in their home environment. The choice and purchase process is supported by the layout of the IKEA shops, the transport phase is facilitated by the special attention given to the packaging and the assemblage is facilitated by the extreme simplification of the knowledge required by customers to assemble IKEA furniture. In other cases, such as bank services (Table 1), the process of value co-production is facilitated by the decodification of complex technical and specialized knowledge into simpler knowledge units, in which information is collected in regard to a specific customer problem. Table 1: Case example of Jyske Bank. Bank services are now organized according to a complex disintegrated structure including several "modular" elements, whose integration is quite hard for the final customers, unless a consultant from the bank helps customers in generating configurations that are most appropriate for their own needs. In 2003 Jyske Bank, the fourth biggest bank in Denmark, promoted innovative ways of managing the relationship with clients. In order to make the offering from the bank clearly visible to the customers the internal space of each branch was redesigned, creating a sort of market space at each branch entrance. In this space, the bank’s offering are packed in boxes, similar to software packages (as in software packages, what is sold is information, not material products. Although people can download the same information from the internet, they often prefer to receive a package, as a sort of material proof of what they buy). Each box corresponds to an integrated solution to a specific aspect of the customer’s life (what should I do if I want to donate some money to my grandchild? What should I do if I want to move home? What should I do if I want to invest my money?...or to buy a new car?) Each pack provides knowledge about the services offered by the bank to support those customers' activities. The package has a barcode that activates a video on an information point where the customer can get an overview of the information included in the package. Other features of the layout, such as a coffee machine (with a special selection of coffee) and some reading space to get inspiration about travels, investments or home improvements offer an inspirational space, in which the customer should feel familiar and free to choose the services offered by the bank.

The perspective shift of Jyske Bank consists in the commitment to fully and actively involve the customers in the selection of the offering the bank can propose. In the most traditional relationship between a bank and its clients the bank is usually very active in proposing its offering to the customers. In the new perspective instead, it is the customer that is supposed to ask for a specific service. The integration of the modular structure for the service takes place in each of the

106


knowledge packages. It would be reasonable to say that the solution platform created by the bank is a semi-finished solution that allows the customers to integrate the service according to their own needs.

Figure 2: Jyske Bank’s solution packages and the information stand to get information on the packages' content.

New contact points for the cooperation between producers and customers Highly individualized solutions are possible when the contact between producer and customer is intensified. The production and consumption processes are often intersecting in a specific place, in which the user co-produces his/her own solution. This is the place where the various components of the production process (the products and any material support offered by the producer and the tacit knowledge and the needs of the customers) are integrated into a specific configuration. Beside the above mentioned examples of IKEA and Jyske Bank, other production sectors are defining specific contact points with customers. When looking at sport shoes, for instance, several companies promoted a direct contact with customers beside the possibility to personalize products online. Adidas, for instance, opened special stores in big cities, in which personal information could be collected, that refer to the customers' physical characteristic. Customers can provide information by running on a computerized pad, which records specific information about weight distribution on the foot, in conditions that are similar to sport activities. Besides being more accurate in the physical personalization of the shoes, this initiative requires the customer to "work" for defining his/her own solution, thus encouraging an active participation in the definition of the ideal


107

solution. The customers who are willing to pay more for this kind of shopping experience form a low percentage of the total market of those footwear companies, however, besides their own solutions, those customers also contribute towards providing lead-user knowledge that can be integrated in other mass produced products. Local solutions and collective intuition Several processes of co-production start from communities of customers/citizens, who take collective initiatives to solve emerging problems or opportunities. This has been particularly frequent in some countries in Northern-Europe, where citizens' associations have a long tradition for generating initiatives in different areas, from the distribution of agricultural products (the cooperative movement) to public schools, from the construction of wind mills to co-housing. The collective intuition and initiative may become an important source of innovation (Several cases of bottom up innovative initiatives have been collected in the EMUDE project; they can be found online at sustainable-everyday.net.). Although they are often the result of spontaneous and private initiatives, an organizational support may amplify the innovative power of such initiatives. Companies can offer organizational support to local production or sharing initiatives, such as urban farming systems (Table 2) or car sharing (Table 3) for instance, in order to address emerging needs and lifestyles, like the demand for sustainable solutions and the renewed attention to local products. Table 2: Case example of urban farming systems. The Urban Farming Project is a community based initiative in Middlesbrough (UK) for citizens to produce their own vegetables and sell them within the local community. The project was based on some simple elements: Windows allotments. i.e. Small boxes where people could grow vegetables. The allotments consist of boxes in different dimensions: small, medium and large. The largest ones are located in public spaces in residential areas, the smallest ones can be placed on windows edges. Unlike other urban farming cases (Manzini and Jegou 2005) the use of allotments makes it possible for people to cultivate their own vegetables without traveling to the outskirts of the city. Meal assembly Centres (MACs) are inspired by what happens in America, where soccer moms go along and basically construct dishes on what’s essentially like a production line. Customers go round various work stations building their meals with all their ingredients that have been pre prepared. Kitchen Playgrounds that encourage people to take part in the preparation of their food before sitting down to eat it, with the aim of linking vegetable growing to experimenting with new flavours and cooking Those elements were supporting local production while creating a market place for those products, in order to make the distribution of local products available on local markets.

108

HANDBOOK OF RESEARCH IN MASS CUSTOMIZATION & PERSONALIZATION Table 3: Case example car sharing services: The Hertz Delebil service.

Car sharing groups are emerging in various European countries as a solution to share the cost of cars and reduce traffic congestion in big cities. Car sharing systems of different sizes are created as spontaneous initiatives. The basic initiative consists in sharing the use of a car within a small group of people. The availability of free online booking systems is a good support to such small organizations. In some countries, like Denmark, in which the taxation on cars is very high, car leasing companies, such as Hertz, have seized the opportunity to organize car sharing initiatives around a stronger structure. "Hertz Delebil" is a service that provides the opportunity to book a Hertz car for a short period of time (even just one hour) using an online or phone service. Dedicated parking lots are distributed around the main Danish cities. After booking a car, members can access it by using a magnetic card. According to this structure, the membership in a car sharing system in one of the Danish cities gives access to car sharing systems in other Danish cities, too. From Hertz' perspective the new service does not imply major variations with respect to the traditional car leasing service. Hertz is in fact "relieving" local groups of people of some of the responsibilities in a car sharing system, working as a "mediator" among customers, making sure that cars are returned at the right time for the next user.

The emergence of such bottom-up initiatives requires companies to participate with a new methodological approach. While the most traditional corporate approach to problems such as food production or mobility is based on the companies' capability to offer products that address people’s needs, the new approach should include industrial companies as platform providers. The outcome of their activity should no longer be a product but rather a solution to specific needs. The platform to be provided should support communication, interaction, networking among people at the local level and between different local contexts. The generation of the platform represents a new methodological approach that requires companies to redefine their strategies, using any potential local source of innovation and activating local providers, citizens and institutions. Creating trans-local solutions The new perspective is joining elements of the past in a new combination, indeed it reframes local and highly individualized solutions, typical of the pre-industrial age, within parameters and criteria inherited by the industrial paradigm. Before the industrial revolution local and tailor-made solutions were based on craftsmens' activities. As such the solutions were not reproducible in any way. In this system the concept of innovation did not make as much sense as the concept of fit. The craftsman was not concerned about producing anything new as much as he was concerned about producing a perfect fit for his customer and for his local context.


109

In order to create the condition for reproducibility, the industrial revolution had to exclude the extreme case of individual solutions. Furthermore the need to create economy of scale called for an extension of design solutions to broader contexts, thus excluding any local specificity. The massification of solutions was compensated by a rapid innovation process, also triggered by competitive pressure. The new paradigm conditions can create a new convergence between local and individual solutions and industrial production. The condition for this to happen is to shift from economies of scale (based on material production) to economy of scope (based on organizational knowledge). This is possible by transferring the concept of modularization from products and components to organizational knowledge. The knowledge modules concern elementary components of a service, therefore they refer to activities that can be performed by local actors, thus making "tailor made" solutions possible for individuals and local contexts. The knowledge modules are the bricks of the architecture for local solutions; the mortar is represented by motivations and shared interests. The New Zealand-based company Ponoko, for instance, gives customers the possibility to design their own product and have it produced locally. Users are required to post their design to Ponoko, which mobilises a network of producers, located in the customers' area that will deliver the components of the prototype and the final product, ready to be assembled. Initiatives like this are based on a modular platform that can be partly or integrally reproduced in a different context, once the needed knowledge modules (local manufacturers and service providers) are identified in the new local context. The offering can be complemented with communication opportunities that link people from different contexts. Ponoko, for example includes customers' designed products in its web page, thus providing examples for other customers and offering existing customers the opportunity to come into contact with buyers in other parts of the world. The creation of online communities around those systems generates flows of knowledge between different contexts, which facilitate the trans-local development and evolution of such systems. Unlike traditional industrial systems, in which the transmission of information is often a threat to the company’s capability to keep its competitive advantage, the circulation of information about specific solutions can reinforce the market position of the company that provides the platform for such solutions, in the new system. The technological innovation embedded in the solution is becoming less important than users' capability to find their own solutions. The ability of empowering users in this sense is a critical success factor for companies.

110


Towards a Methodological Approach The new paradigm should define a new methodological approach for the development of highly individualized and localized solutions. This approach refers to:

A concept development activity that is much more focused on individual people, their context, their behavior and their capabilities

The attitude to develop local platforms

The capability to take advantage of economies of scope based on organizational knowledge

Focus on individuals, contexts, behaviors and capabilities The need for generating highly individualized solutions calls for new analysis and communication tools, beyond the techniques used today, that are mostly inspired by marketing methodologies. Marketing techniques are being developed, that increase the quantity and quality of information about user needs. Some of these techniques, focus groups for instance, are able to get direct input from users in the early stages of concept development. Those techniques however are often unable to collect contextrelated information, which would become very relevant when working on local or highly individualized solutions. They are unable of capturing tacit knowledge concerning customers' problem solving attitudes and capabilities, especially when it comes to very specific contexts. The analysis of users in context (Lindsay and Rocchi 2003) shadowing or filming them while performing routine activities, provides direct insights into the way customers interact with their local surroundings. Further information can be directly produced by the users, by inviting them to write down personal diaries, and to provide textual or visual notations about their daily life and their preferences (Gaver, Dunne et al. 1999). Such new tools introduce a new perspective for the development of innovative solutions, but also open new problems. Indeed, they provide qualitatively rich information about customers' life, preference, tacit knowledge and attitudes, but they open the question of how to translate such information patterns into operative indications for the design of new solutions. This problem could be addressed by generating personas, a well known technique in interface design that synthesizes the main tracts of the behaviors and attitudes of the main actors (Cooper 1999). A local activity can also be described by time sequences, transforming data from user observations into synthetic stories about routine activities and possibly


111

comparing routine activities of many actors interacting in the same context (Figure 3) (Morelli and Tollestrup 2007).

Figure 3: Time line comparison between the opening time of a local shop and its customers' daily routine (Jensen et al. 2006).

Time and time sequences are in fact critical factors in the development of new solutions. The new methodological approach would require companies' engineers to include time-related and behavioral components in the development of the new solutions. The traditional approach to digital modeling, focused on the physical and geometrical properties of objects and components and therefore static and synchronic, should be replaced with an approach that takes into account the whole life cycle of a product or service, thus modeling the behavior of the system in real

112


time simulations. A new approach of this kind is not new to manufacturing companies, Boeing, for instance, introduced the concept of life-cycle modeling in the design of the 7E7 projects. This approach allows thousands of suppliers to cooperate in parallel with Boeing in the development of new models, by simulating their life-cycle behavior, including operations and maintenance (Williams 2004). The new modeling is not necessarily based on digital technology, but is rather a participatory process, involving all the possible local actors. Use cases could be a good documentation tool for those modeling activities. Use cases, commonly used in software development to elicit requirements (Kulak and Guiney 2000; Leffingwell and Widrig 2000), have also been used in service design to produce real time simulations of different kinds of interactions within a service (Figure 4). The upper part of the figure describes the phases of user’s behavior, whereas the lower part describes the correspondent behavior of the system for each phase of user’s behavior (Morelli 2002, 2003, 2006). Developing local platforms The analysis of a local context defines a multifaceted picture of individual preferences, capabilities and tacit knowledge. As a product of an evolutionary process, local contexts and the activities developed in them are the result of the interaction between different skills, capabilities, tacit knowledge and personal exchanges. The context specificity of such processes represents the main challenge for a company that intends to produce local or individual solutions. The challenge consists in generating explicit organizational structures that catalyze and address those processes, whereas in the past the local innovation was only emerging through "natural" evolutionary processes. In order to make business sense, the organization of local activities and the utilization of local potential should be synthesized in modular architectures, in which each module refers to an actor (a service provider, a product manufacturer or a user with his/her own tacit knowledge) which holds the knowledge needed to solve a specific part of the solution. The main organizational task is to generate a solution platform that allows multiple solutions, by specifying sequences of events, interaction among modules, system boundaries information, physical and financial flows (Figure 5). Those platforms allow for a distribution of engineering power among the modules of the platform. Each module will be appropriately designed and organized at the local level, leaving the system organizer with the responsibility to negotiate the connection of those modules through an appropriate modeling activity (e.g. use cases, scenarios) that simulates the behavior of the system in time and space (see previous paragraph).

Figure 4: Use case technique used in the design of food delivery for ageing people.


113

114


Figure 5: Solution platform for a local food delivery system for elderly people.

Generating economy of scope The shift to the new approach is not simple, as it often requires a thorough rethinking of corporate strategies and roles, a perspective shift from a leading position of a vertical integrated organization to a coordinating position of a horizontal network of peers. The present approach to the development of new products is often based on a clear definition of roles between a server (the manufacturer, the service provider) and a client (the user). This approach is based on the assumption that the server should make something or perform a task on behalf of the client. Therefore this approach is relieving clients of any practical activity and often of any responsibility (Manzini 2005; Morelli 2007). This logic should be reversed in the new paradigm. Rather than thinking of relieving products or technologies, companies should think in terms of enabling solutions, which emphasize customers' capabilities and lead them towards the satisfaction of their very personal needs. The capability to activate and motivate local actors is the critical resource that supports the integration of heterogeneous systems of knowledge holders in local contexts into the higher level of the Fine’s double helix model. Furthermore, this resource allows companies to take advantage of economies of scope. Economies of scope refer to an increase of efficiency made possible by the multiplication of product variations (unlike economies of scale, which refer to the largest possible production of the same product). The main resource supporting the multiplication of product variations in the new approach consists of the organizational knowl-


115

edge acquired in each specific and local context. When appropriately codified and structured, this knowledge can be transferred to different contexts, in order to identify relevant actors and potentials in new contexts, thus generating and reproducing the genetic code for infinite solutions. Like with product families, such solutions share logical links and languages, but in this case the modular components are represented by knowledge holders. In this sense the opportunity offered by the new paradigm represents a new form of industrialization, referred to local or individual solutions. Concluding Remarks The aim of this paper is to provide arguments for a substantial perspective change, or a paradigm shift from industrial production to highly individualized solutions. The need for a new paradigm view is supported by the argument that mass customization, which represents the highest level of development of the industrial production paradigm, cannot be used to explain some emerging phenomena. In this paper some relevant contributions to the definition of the concept of mass customization have been discussed, emphasizing that, although the term includes several emerging trends in industrial production (such as the increase of flexibility, modularization, the capability to adapt industrial production to smaller and smaller target groups), it overlooks the role of the customers in the production process. Several contributions on mass customization describe the integration of customers in the production process as a mere process of elicitation of requirements. Other authors exclude that mass customization concerns services, because of the high degree of involvement of customers in the value-production process. This view is still based on a passive involvement of the customers in the production of value. Therefore the customer is in fact kept out or placed at the end of the process. The increasing relevance of services, the emergence of highly localized solutions and the increasing number of customers' autonomous initiatives to create individual solutions challenge this view, by requiring an intense involvement of new stakeholders, such as customers or local actors, in a co-production process. The outcomes of such a process are service-based solutions, rather than products, whereas the main focus is on the customer, rather than on the production system. In this perspective mass customization is an essential cause for reflection, because its characteristics can immediately be related to the emerging paradigm; however this concept is unable to explain some emerging changes. Furthermore the methodological approach required for the new solutions implies a logical shift of the focus (from industrial processes to customers' behaviors and routines) and the

116


extension of production systems (from global to local, from enterprises to individuals). Such a shift goes far beyond the ultimate presumptions of the paradigm of industrial production and therefore calls for new conceptual frameworks beyond mass customization.

References Arbnor, I. and B. Bjerke (1997). Methodology for creating business knowledge. Thousand Oaks, Calif.; London, Sage. Berger, C. and F. Piller (2003). Customers as Co-Designers. IEE Manufacturing Engineer (August/ September 2003): 42–45. Cooper, A. (1999). The inmates are running the asylum. Indianapolis, IN, Sams. Ettlie, J. and P. T. Ward (1997). US Manufacturing in the Early 1990s: the Chase and Challenge. Business Strategy Review 8(4): 53-59. Fine, C. H. (2000). Clockspeed-Based Strategies for Supply Chain Design. Production and Operations Management. 9(3): 213–221. Gaver, B., T. Dunne, et al. (1999). Design: Cultural Probes. Interaction. 6(1): 21–29. Jensen, R. D., C. Ø. Nielsen, et al. (2006). My Store. Department of Industrial Design. Aalborg, Aalborg University. Master in Industrial Design, 7.semester. Kaplan, A. M. and M. Haenlein (2006). Toward a Parsimonious Definition of Traditional and Electronic Mass Customization" Journal of Product Innovation Management. 23(2): 168–182. Kuhn, T. S. (1962). The structure of scientific revolutions. Chicago; London, University of Chicago Press. Kulak, D. and E. Guiney (2000). Use cases: requirements in context. New York Boston, Mass.; London, ACM Press; Addison-Wesley. Leffingwell, D. and D. Widrig (2000). Managing software requirements: a unified approach. Reading, MA, Addison-Wesley. Lindsay, C. and S. Rocchi (2003). Highly Customerised Solutions: The Context of Use Co-Research Methodology. Innovating for Sustainability. 11th International Conference of Greening of Industry Network, San Francisco. Manzini, E. (2005). Enabling Solutions for Creative Communities. Designmatters. (10): 64–68. Manzini, E., L. Collina, et al. (2004). Solution Oriented Partnership. How to Design Industrialised Sustainable Solutions. Cranfield, Cranfield University. European Commission GROWTH Program. Manzini, E. and F. Jegou (2005). Sustainable Everyday. Milano, Edizioni Ambiente. Morelli, N. (2002). Designing product/service systems. A methodological exploration. Design Issues 18(3): 3-17. Morelli, N. (2003). Product-Service Systems: a Perspective Shift for Designers. A Case Study: the Design of a Telecentre. Design Studies. 24(1): 73–99. Morelli, N. (2006). Developing new PSS, Methodologies and Operational Tools. Journal of Cleaner Production. 14(17): 1495–1501. Morelli, N. (2006). Globalised markets and localised needs: Relocating design competence in a new industrial context. Engineering & Product Design Education Conference Salzburg.


117

Morelli, N. (2007). Social Innovation and New Industrial Contexts: Can Designers Industrialize" Socially Responsible Solutions? Design Issues. 23(4): 3–21. Morelli, N. and C. Tollestrup (2007). New Representation Techniques For Designing In A Systemic Perspective. Nordes 07. Stockholm. Normann, R. and R. Ramirez (1994). Designing Interactive Strategy. From Value Chain to Value Constellation. New York, John Wiley and Sons. Pine, J. B., D. Peppers, et al. (1995). Do You Want to Keep Your Customers Forever? Harvard Business Review. 73(2): 103–114. Porter, M. E. (1985). Competitive Advantage: Creating and Sustaining Superior Performance. New York, Free Press. Ramirez, R. (1999). Value Co-Production: Intellectual Origins and Implications for Practice and Research. Strategic Management Journal. 20: 49–65. Tapscott, D. and A. D. Williams (2006). Wikinomics. How Mass Collaboration Changes Everything. London, Atlantic Books. Von Hippel, E. (2005). Democratizing Innovation. Cambridge, Massachusetts London, England, The MIT Press. Von Hippel, E. and R. Katz (1998). Shifting Innovation to Users via Toolkits. Management Science. 48(7): 821–833. Williams, A. (2004). Boeing 7E7 Project Pushes PLM Boundaries. Cadalyst. Zipkin, P. (2001). The Limits of Mass Customization. Sloan Management Review. 42(3): 81–87.

Author Biographies Dr. Nicola Morelli is an Associate Professor at the School of Architecture and Design at Aalborg University, Denmark. Before starting this position he worked at the Centre for Design at RMIT University (Australia) and at Politecnico di Milano (Italy). His research focuses on the role of industrial designers in the development of innovative Product Service Systems in the private and in the public sector. Together with the Research Unit on Service Design at Aalborg University, Nicola Morelli is involved in research projects aimed at providing methodological insights to support the process of designing productservice systems and in theoretical and strategic reflections about the development of highly customized and localized solutions. Contact: www.aod.aau.dk/staff/nmor | [email protected] Louise Møller Nielsen holds a M.Sc. Eng. in Industrial Design, and is currently PhD candidate at the Department of Architecture and Design, Aalborg University in Denmark. Her present research is concerned with the front end of innovation and multidisciplinary team interaction. Besides this her research interest is in the creation of experiential concepts and designing for hope. Contact: www.aod.aau.dk | [email protected]

1.6

Is the Best Product a Unique Product? Exploring Alternatives to Mass Customization with the Online Community of Threadless Adam Fletcher (formerly) Spreadshirt, Germany

Mass Customization has been portrayed as the ultimate form of marketing and the "business opportunity of the next millennium". This paper presents the results of a case study undertaken with the online t-shirt manufacturer Threadless and its Virtual Community. The literary assumption that consumers want unique products, following recent renewed interest in Mass Customization has prompted this research. The aim of this study was to look at an industry where it is technically possible to deliver a "pure" Mass Customization experience and then to look at businesses adopting different approaches to see what they offer the consumer. Threadless' business model aggregates opinions of user submitted designs and manufacturers the most popular into limited t-shirts. This studies looks at why this model is an attractive proposition for customers, community members and for Threadless. The results challenge a number of assumptions which can be found in the wider MC literature. Respondents at Threadless are willing to accept a product which they did not create and is not unique. This is providing that the product is at least limited, and that they have had involvement in its creation. The other key finding is that whilst Internet may offer enabling technologies which reduce the cost of individualization, these same technologies may also reduce the cost of aggregation. This allows businesses to group and manufacture for ever smaller markets of customers sharing the same needs. The author rejects the notion that a unique product created by its purchaser is the definitive product, calling for businesses to look for ways to combine both involvement and exclusivity in product creation utilizing these aggregation opportunities.

Introduction This paper will investigate consumers' attitudes to Mass Customization (MC) and collaboration for new product design conducted within a Virtual Community (VC). Research in this paper is centered on the virtual community of the online tshirt business Threadless.com. Threadless is an ongoing design competition in which users are invited to submit a design to the Virtual Community. Any registered members can rate the design out of five and can also provide qualitative feedback to the designer, usually in the form of modification suggestions or

118


119

positive/negative written feedback. A number of the highest rated design are then produced as t-shirts in short production runs and sold on the site. To follow is a critique of available literature in this area, followed by a discussion of the findings from a questionnaire completed by members of the Threadless VC and ethnographic research of messageboard communication. Conclusions are followed by recommendations for further research in this area. Terms Definition & Context Mass customization (MC) The definition of MC adopted by this paper is Hart (1995, p.1) stating that MC was "The use of flexible process and organizational structures to produce varied and often individually customized products and services at the low cost of a standardized, mass produced system." In short, increasing variety and involvement for every customer without losing manufacturing efficiency. It has been suggested that MC has not had the business impact that was anticipated (Lee et al. 1999), however MC can be seen in a wide variety of products/services and industries. Spreadshirt the online "make your own" apparel business allows anyone to design one off clothing produced and dispatched within 48hrs. Lego now actively engages their customer base in new product developments and allows users to design their own Lego sets online (for more about the Lego Factory see Berger et al. 2005). In the footwear industry most of the major trainer manufacturers allow for custom designing (such as Reebok with "Rbk CUSTOM" or Nike through "NikeID") and some offer completely custom products exactly molded to fit the customers foot (such as Adidas' "miAdidas" service, see Berger et al. 2005). Other MC products include Wine (Elite Vintners), Cameras (Leica) or the extremely successful build to order model used by Dell. One common feature in all the examples above is that they use the Internet as the key enabler, with which to interact with their customers. Advances in Information Technology and the Internet in recent years, particular in e-commerce and social network (e.g. Facebook), has created what is now a global marketplace. This new technologies are recognized as the key enabler for wider adoption of MC (Piller 2002; Von Hippel 1998; Schubert and Koch 2002; Pine et al. 1993; Fuller and Hienerth 2004) Virtual community (VC) This paper will look at how a VC can support MC activity. To date there is not one definition that has been adopted by the academic community for a VC. This

120


paper from Schubert and Ginsburg (2000) and Sawhney and Prandelli (2000), it is suggested that Virtual Communities are: "The union between individuals or organizations using electronic media to communicate within a shared semantic space on a regular basis" (Schubert and Ginsburg 2000, p.2). "The communities provide sociability, support, information, a sense of belonging, and social identity. This community is usually but not exclusively created around shared values or interests" (Sawhney and Prandelli 2000, p.5). Context MC as a concept has been discussed in academic literature with regularity since it was first articulated some 25 years ago. The idea of a customer designing a unique product or service was not a new concept even then, instead writings on MC were commenting on an increasing shift away from Mass Production towards individualized products, increasingly designed with input from customers. The Internet has the potential to change the way we buy and sell, empowering small businesses to sell on a global platform as if they were a multinational (Collin 1999), reducing barriers of entry (Porter 2001). The Internet facilitates direct and rich interaction with customers (Piller and Walcher 2005), allowing customers to give feedback and collaborate on new products. Internet sites such as Facebook or Ebay have shown that the power of the Internet lies in uniting individuals, developing a VC of loyal customers to support your brand. The proposed benefit of MC is that consumers are incorporated into the design process, designing the exact product or service they require. This will allow a business to develop a 1:1 relationship with their customers, understanding their exact requirements, helping to build a relationship with them which is impervious to competitors (Pine et al. 1993). Literature Review & Conceptual Framework What follows is a critical re-examination of the underlying assumptions behind MC. This aims to show that the current academic thinking regarding MC may be incorrect. This is followed by the presentation of an alternative approach to MC adopted by the online retailer Threadless and discussion of how this model might better support MC & innovative activities such as NPD. This paper takes the view of Hart (1995) who suggests that there are two different ways of defining and conceptualising Mass Customization. The first visionary definition is "the ability to provide your customers with anything they want profitably, any time they want it, anywhere they want it, any way they want." (Hart 1995, p.1)


121

While this is only meant as a "transcendent, absolute idea that exists solely in the ideal" (Hart 1995, p.1), it best describes the basic principal and inherent complexity of MC. The key difference between MC and the traditional one size fits all, "any color as long as its black" approach of Mass Production made famous by Fordism is the unique, tailored experience that the customer receives. You aim to reach large numbers of customers but simultaneously treat them as individuals (Davis 1996), whilst aiming to maintain the efficiency of mass production (Pine et al. 1993; Piller 2003). The other perhaps more realistic and practical definition offered by Hart is "The use of flexible process and organizational structures to produce varied and often individually customized products and services at the low cost of a standardized, mass produced system." (Hart 1995, p.1) The key difference in this second definition is the idea of not promising to produce anything a customer may desire but introducing flexibility, variety and where possible individualization into the experience. Note that this definition does suggest the possibility to "maintain the efficiency of mass production" (Pine et al. 1993; Piller 2003). Research suggests that in most industries this is at present still unrealistic, even with advances in manufacturing (such as CAD) and telecommunications (e-commerce etc) technologies. Reducing the trade-off between variants and production cost (Piller 2003) at present adding variety still results in additional operational costs (Brabazon and McCarthy 2004) and lost economies of scale. Definitions of MC in the academic literature have tended to overplay the need for MC to create a unique product or service. This paper rejects this notion and suggests that customer involvement is the fundamental principal of MC (Piller et al. 2004). The experience itself, more than what is created, provides the unique value for each individual (Prahalad and Ramaswamy 2004). The literature tends to draw a picture of two opposing and distinct strategies and mindsets (Lampel and Mintzberg 1996). In reality as suggested by the framework from this paper there is a continuum of strategies and it is the task of industry to decide which approach best compliments the needs of their customers and the capabilities of their business. Why mass customization? If there really is nothing simple about MC (Hart 1995) then why is this topic worthy of such academic and business interest? Teresko (1994, p.46) believes "If you can sell everything you make, mass customization is irrelevant". This like Harts visionary definition may be correct in an absolute, transcendent and idealistic way, however in reality, it is flawed. While today a business may sell

122


everything that it produces, only in a monopolised market are sales liable to be static and not subject to competition from rival businesses. While today you may sell everything you make, a competitor moving to MC and offering the customer increased choice may irreversibly change this. Just because sales are currently high, you may not be satisfying customers or selling profitably and fast enough to reduce potential revenue left sitting in stock. It also does not address the fact that by utilizing MC you are co-designing with the consumer which will allow you to "always sell everything that you make", but as suggested by empirical research (see Piller et al. 2004 for an overview), to sell it at a premium price. MC & the manufacturer For the manufacturer MC offers cost saving potential through better forecasting and reduced wastage. Estimates suggest the apparel industry alone loses over $300bn every year due to erroneous forecasting, heavy inventory and lost profits as a result of necessary discounting to reduce stock levels (Sanders 2001). The key difference with MC is that consumers are incorporated into the design process, and design the exact product or service they require. As manufacturers no longer have to predict demand for a product they may theoretically be able to keep smaller inventories of finished goods. Heavy discounts and promotions to move less popular products out of warehouses may become a thing of the past, achieving both manufacturer and consumer satisfaction (Lee and Chen 1999). Whilst potentially lowering stocks of finished goods, MC may have a negative effect on other areas of the manufacturing process. Large quantities and varieties of raw materials will need to be held to help support uncertain, fluctuating demand which could send inventory costs out of control (Lee et al. 1999). Some research has suggested that customers want and are willing to pay more for customized or non-standard products (Piller and Berger 2003; Piller and Muller 2004; Franke and Von Hippel 2003, see Piller and Muller 2004 for an overview of research conducted). The problem with this research is that it is often undertaken using a questionnaire. Measuring WTP using a questionnaire may leads to unrealistic results as consumers have an imagination about customization, but no experience of it (Piller and Muller 2004). It is worth noting that a large number of respondents in these studies had no previous experience customizing products. Pine (1994, p.14) suggests that "Customers don't want choice. They want exactly, what they want". Leaving aside the idea that any of us actually know what we want, it is unsurprising that many consumers would say they would be prepared to pay a premium for this experience. However, without experience in designing products (on or offline) customers may not be aware of the challenges in


123

articulating what they want, a number of sources suggest on the whole consumers find it almost impossible to do this (Berger et al. 2005; Zipkin 2001; Von Hippel 2005). They may also be unaware of the amount of time and involvement required in this process (Piller and Muller 2003). There is no guarantee that even with additional help provided by the manufacturer to provide the technical information in a format understandable to the customer (sticky information, see Von Hippel 1998 for an overview), the consumer is able to produce something that meets their desires. As Expectation Disconfirmation theory suggests this is only made harder as increased customers involvement will incrementally increase customer expectations of the final product. Mass customization & the virtual community It is suggested that recent renewed interest in MC is because of the introduction of new technologies in particular the Internet. These technologies often cited as key enablers of MC (Piller 2002; Von Hippel 1998; Schubert and Koch 2002; Pine et al. 1993; Fuller and Hienerth 2004). The Internet provides an efficient platform to reduce the often difficult and costly process of transferring a customers wants or needs to a manufacturer (Piller and Walcher 2005). As well as facilitating the efficient production of customized goods, Internet technology facilitates the personalization of customer relationships (Piller 2002). Amazons sophisticated recommendation system is an excellent example of this. Offering a tailored user experience to every customer based on their interest areas and previous purchasing behavior. One area that has received less focus in the literature is how encouraging collaboration amongst mass customizing customers can improve their MC experience. One Internet technology which can facilitate this collaboration amongst consumers is a Virtual Community. Despite that lack of consensus in what exactly constitutes a VC, they have existed online in various forms for approximately 30 years (Ridings et al. 2002). A recent report found that 79% of Internet users identified at least one community with which they maintained regular online contact (Rainie and Packel 2001). They are expected to have a significant impact on commercial companies fundamentally changing how they develop, price and promote their products (Hagel and Armstrong 1997). It is suggested in the literature that VCs are attractive to businesses as they provide a mechanism to:

Facilitate a stronger relationship between the firm and its customers (Brown et al. 2002; Hagel and Armstrong 1997).

124


Generate rapid-response and instantaneous feedback concerning different innovation projects throughout the entire innovation process (Ernst and Gulati 2003).

Access a community's publicly available knowledge base containing information detailing their likes, dislikes or demographics (Ridings et al. 2002). Online communities have been found to be highly innovative and can be found for almost every product or service (Fuller and Hienerth 2004), with research suggesting that many innovations originate in the user rather than the manufacturer domain (Von Hippel 1986; Piller and Walcher 2005).

Collaborative design is of interest as it may support MC using a community to encourage creativity and assist customers in making better choices than if they are left to design in isolation, selecting from a large variety of choices (Von Hippel and Tyre 1995; Franke and Shah 2003; Piller et al. 2005). Sawhney and Prandelli (2000) concluded that a business model that combined communities into product development empowers peripheral players, giving them the right to contribute their own experience and individual knowledge to the final output. Each consumer can add to the collective knowledge of the community adding knowledge from their individual experiences. Jeppesen and Molin (2003) believe that this user creation and development results in a longer product life and greater sales of the basic product. Some studies have proven that without conscious effort from the community sponsor, collaborative activity is already taking place. Franke and Shahs (2003) research concluded that "Without exception, the innovating community members we surveyed do not innovate in isolation or secrecy; they receive important advice and assistance from other community members" (p.158). Threadless – A more effective model of user lead manufacturing? Instead of thinking of opposites, this paper suggests that it is towards the centre of the continuum between MP & MC or Standardisation & Individualization that may offer the best fit for consumers and manufacturers. It’s suggested that conceptual polarization has lead management thinkers to ignore strategies which combine these logics (Lampel and Mintzberg 1996). One company which on the surface seems to combine these two strategies well and will be the focus of this papers research is the online t-shirt retailer Threadless.com. Threadless' business model cannot be categorised under Lampel & Mintzbergs framework as it has one fundamental difference from all the approaches, the customization & customer


125

input occurs at the earliest possible point in the value chain, at the conception stage. Threadless' business model is outlined below in Figure 1.

Figure 1: The Threadless business model.

In this approach the customer is in almost complete control over designing the product and determining what is to be manufactured. Every customer has an equal say and input but collectively they decide which product moves down the activity chain where the manufacturer handles the fabrication, assembly and distribution. In Threadless' case members of the community submit t-shirt designs which the rest can comment and score from 1-5. The top scoring designs are usually then manufactured (Threadless have the final say) and sold in limited quantities on the site. The winning designer receives $2000 in exchange for the rights to the design (and an additional $500 payment each time their design is turned into a print run of t-shirts). This research aims to look more closely at why the Threadless model is successful and what motivates consumers to participate in this VC. Conceptual framework It is believed that there has been an over emphasis in promoting MC as the future of manufacturing, jumping from one extreme to the other and ignoring strategies combining both these logics (Lampel and Mintzberg 1996). The fundamental assumption of MC is that with no obstacles, challenges or inconveniences, customers would rather create a unique product or service. Predictably today’s

126


business does not offer this hypothetical ideal for any but a handful of businesses. As a result it is suggested that there is a disconnect between the customers desire for involvement and individualization, and the manufacturers desire for economies of scale and predictable, manageable demand. The framework below outlines this perceived disconnect and recaps some points from the literature which encapsulate this (Figure 2).

Figure 2: Conceptual framework showing a MC disconnect between customers and manufacturers.


127

Research Questions, Methodology and Approach This section will outline the primary research methods, philosophy and strategy that underpin their use. The aim of this research is to try and answer the following research questions in relation to Threadless and its VC.

What aspects of the Threadless business model do its community members consider the most important?

Do all community members submit designs? If not, what factors stop members submitting designs?

What is it that keeps community members visiting and interacting with Threadless and its VC?

Does this business model support aggregation of user requirements? If so, how?

Sawhney and Prandelli (2000) believe that "a business model that combines communities into product development empowers peripheral players", how much evidence of this can be seen at Threadless?

To research Threadless, a case study approach was adopted. While only focusing on one community a case study approach offers an opportunity to gain a deeper insight into a relatively unexplored phenomenon (Jeppesen and Frederiksen 2004). It is suggested that Threadless is a unique community of co-design, and it is proposed that adopting an idiographic approach may help to highlight as many of the unique features of this community as possible (Bryman and Bell 2003). The research approach was essentially constructivist. While critics may argue that participant observation lacks reliability and may lead to bias, this is accepted, the Threadless VC is not an absolute reality where scientific measurement may occur. It is not disputed that research in this manner may lead to results which are not generalisable. Instead an interpretivist approach may suggest the existence of some key motivators or enablers which may exist in a number of different Virtual Communities. The goal is purely to seek a degree of theoretical generalisability from the results. The research comprised of a questionnaire completed online amongst community members and a participant observation of community communication. A questionnaire was used as it allows for further exploration of the emergent themes but allows for the drawing of a broader consensus from the community in general. It can help to understand the meanings they attribute to their acts and to the acts of

128


others (Bryman and Bell 2003). It can also be tailored to gather data on a specific topic (to help answer the Research Questions), whilst having questions which are more open in nature, complimenting the exploratory nature of this study. A link to the questionnaire was posted on the Threadless "Blog Forum" (threadless.com/ blogs) which is a public messageboard (the two names will be used interchangeably). Prior agreement had been received from Threadless to post the message and link. It was felt that this would encourage respondents to complete the questionnaire and raise its profile amongst the community. The second piece of research was a participant observation, monitoring and classifying all communication on Threadless' messageboard called the "Blog Forum". This observation lasted for seven days from the 1st to 8th of March 2006. Further analysis and classification of the comments posted below the first ten design submissions mentioned on the "blog forum" during this time was also completed. Participant Observation in this unobtrusive manner does raise a question of ethics (Paccagnella 1997). In this case all the data analyzed is considered as public discourse and viewable by all. To respect the privacy of the community members no names of community members or links back to the text were recorded. Only communication deemed relevant to this research was viewed & categorized. Results, Findings and Discussion In total, the questionnaire was completed by 204 visitors to the Threadless "Blog Forum". As all site visitors regardless of whether they are guests or community members have the ability to view the Threadless "Blog Forum", the total population is impossible to quantify. The "Blog Forum" is separate from the more common activities of rating, submitting or purchasing so is unlikely to be viewed by a majority of site visitors. In total 191 of these responses were deemed usable for this survey, the other 13 contained incomplete information. In total 219 users started the questionnaire giving an 88% (approx) started/completed rate. Over the seven day participant observation, a total of 422 messageboard threads (threads begin with a single member posting a title and message which other members may reply to) were viewed and had their topic category logged. Table 1 shows the different categories and the total number of individual threads in each. Thread types 3a-3e were deemed relevant to this study and likely to return data which would help answer the research questions. Although all threads were viewed and categorised, only type’s 3a-e (67 in total) were considered of interest to this research and also had their replies recorded and categorized to get a picture of the collaborative communication used on the Blog

129


Forum. In total 6751 replies occurred in all threads over the seven days. Of these 794 (11.6%) were in the threads categorized as 3a–3e on Table 1 and therefore the type of each reply was also recorded and categorised using a similar classification scale. It was a surprise to the researcher just how social the Blog Forum was, with the vast majority of topics started (over 80%) having nothing to do with Threadless or design in general, yet the frequency with which people returned to the forum was very high with more than 45% returning several times a day. Table 1: Number of threads by topic observed on the "Blog Forum". Thread classification 1.

Thread type Social - not T-shirt related

N. of threads 197

2.

T-shirt related - not Threadless

6

3a

Design help

8

3b

Request by designer for feedback on an Initial design (before the submission stage)

17

3c

Discussion of t-shirt undergoing/completed scoring

10

3d

Discussion of t-shirt undergoing/completed scoring (by the designer)

24

3e

Vote request for T-shirt undergoing scoring (by the designer)

8

4.

Discussing of a winning (printed) T-shirt

59

5.

Discussing Threadless/Talking to Threadless

28

6.

Requesting a reprint of a sold out design

9

7.

Suggestions of what Threadless should do next

5

8.

Request for help (general not Threadless related)

20

9.

Street Team (a Threadless loyalty scheme)

16

10.

Sponsor Post (Post from Threadless)

2

11.

Discussion of recent Purchase/What to Purchase

15

Total

424

What aspects of the Threadless business model do its community members consider the most important? When asked to prioritize from 1–7 the most important reasons why they purchased from Threadless, the top four variables (cited 1–3 in importance) are shown below in Table 2. Surprisingly and contradictory to the wider MC literature it was not

130


involvement which scored highest, instead it was "Innovative designs". In fact the results showed that involvement only ranked 3rd in importance. Customers liked the exclusivity offered by Threadless' limited batch production. Perhaps in a market dominated by large multinational brands Threadless' offer of a limited (although not exclusive) item is compelling enough. Verbatim feedback also supported this, respondent n.146 said "Uniqueness, that’s all I want and search (sic) from these shirts, it makes me feel as a single individual that doesn't follow the clothing trends that are made popular today." Respondent n.164 agreed and said "I feel the greatest reason people buy from Threadless.com is the short runs of unique, clever designs." Table 2: Responses most voted 1-3 for the question "What is the most important reason why you purchase from Threadless?"

Voted 1

Voted 2

Voted 3

Total 1-3 votes for category

Ranking

Innovative designs

52%

15%

6%

73%

1

Exclusivity of designs (short run production cycles)

9%

29%

15%

53%

2

Involvement in the design process

7%

17%

21%

45%

3

Price

10%

10%

21%

42%

4

The Threadless ethos and brand

6%

12%

14%

31%

5

Sizing and color ranges

4%

12%

15%

31%

6

Delivery times

12%

5%

7%

24%

7

In this case the quality of the design might be far higher than they feel they can have produced which is perhaps why 60% of the community have never submitted a design. So it is suggested that the community is being asked to make a trade off between the losses in uniqueness (compared to pure MC) for the increase in the design quality of what is produced.


131

Does this business model support aggregation of user requirements? If so, how? The strength of this type of NPD is in how it uses technology to simplify the cost of aggregating users' requirements. The ratings system acts as a form of market research gauging demand for an as yet un-manufactured product. During the research other examples of this aggregation were also observed. The participant observation of the Blog Forum highlighted several examples of small groups being formed around certain designs which were not popular enough to be put into manufacturing but were liked by a subset of the community. Thread ID:422 and a comment on Submission ID:10 are an example of this "The design was very popular here on the blogs, but threadless has told us they don't want to print it. Do we want to print it ourselves? if enough people are interested, i'll have the shirt printed up on my buck and sell it for cost here on the site..." (Reply on Thread ID:422) "Thanks for all the positive feedback everyone! (sic) and if this doesn't get picked, I will be printing it for my own company." (Comment on Submission ID:10) While not authorised by Threadless this shows that even sub-groups of community members can use the functionality provided by Threadless to find other users with similar design tastes. These groups may not be large enough to make them financially viable for manufacture by Threadless, but this showcases user lead aggregation which may lead to t-shirts being manufactured in smaller quantities for groups of users who meet on Threadless.. Empowerment of peripheral players Sawhney and Prandelli (2000) state that "a business model that combines communities into product development empowers peripheral players". How much evidence of this can be found within the community of Threadless? The answer to this question is dependent on who is judged to be the "peripheral players" within the Threadless community. If we consider that essentially Threadless is running a non stop competition then the key actors are Threadless and the people that submit designs. Without these there is no competition, nothing to rate or manufacture. With over 60% of respondents having never submitted a design this leaves the majority of our sample as so called "peripheral players". If we compare their behavior to those that have actually submitted a design to the competition, the results are surprising. Non-designers reported higher overall satisfaction, rated more designs, were more likely to post or reply to messages on the Blog Forum and had also brought more t-shirts in the past year. They have the same voting rights as everyone else and although they may be lacking the technical skills to

132


submit a design they are needed by the designers to improve the ratings of their designs. The result of this is that in the observation period 17 different threads asked for their (and the community in generals) feedback on designs before submission and 24 during the ratings process. This supports Sawhney and Prandelli (2000) showing that a strength in the Threadless business model seems to be in the way that it offers different ways for peripheral players to become involved, as one questionnaire respondent said "we get to become a part of the tshirt design we are helping somebody with… Involvement is a big factor. (Respondent n. 58) If we compare the Threadless business model against the original aim definition of MC from the Literature review which was "Aim to reach large numbers of customers but simultaneously treat them as individuals (Davis 1996)", the Threadless model seems to achieve this. The experience is different for every person visiting the site as they decide what to rate, how to rate it and what feedback to leave, this experience offers the unique value which Prahalad and Ramaswamy (2004) suggested MC provides. The actual t-shirt purchased is not unique but because designs are limited, distribution worldwide the likelihood of seeing somebody else in the design is small. Small batch production over one off items should help the manufacturer towards another MC goal — maintaining the efficiency of mass production (Pine et al. 1993; Piller 2003). Hart (1995) suggested that the homogeneous market was a thing of the past. The Threadless model contradicts this, the Internet represents a global marketplace which increases the likelihood whilst reducing the cost of finding multiple consumers with the same, homogeneous needs. It is believed that this highlights a potential gap in the MC literature which has over emphasised the value consumer place on uniqueness. Instead of thinking that MC "is the capability to offer individually tailored products or services" (Zipkin 2001), instead the ideal as suggested from the Threadless research maybe closer to a limited (but not unique) product with the usual high customer involvement. While the homogeneous market may be in decline the Internet offers an opportunity to find customers with homogeneous requirements. Piller and Walcher (2005, p.7) said that "The Internet provides an efficient platform to reduce the often difficult and costly process of transferring need information from customers to a manufacturer." Threadless shows that it also provides an efficient platform for customers to transfer need information to other customers supporting an aggregation of user requirements. Previously, getting this kind of need information through traditional market research would have been time consuming and laborious. By utilizing the Internet and reviewing community data Threadless can look at the number of maximum


133

ratings or average rating and from previous experience make an estimate of the number of sales that interest level will equate to. The Threadless approach delivers customer involvement but minimises some of the challenges that reduce manufacturer satisfaction with MC. Lee et al. (1999) said that for the manufacturer keeping large quantities and varieties of raw materials with uncertain, fluctuating demand could send inventory costs out of control. The community provides data to help gauge demand. Yet the consumer does not have to pick from a range of uniform, "average" products (Lee et al. 1999). Production complexity is reduced somewhat as you're not producing one off items. The research also seems to suggest that this business model helps overcome some of the other MC challenges cited in the literature review, this is represented by Figure 3 below which adapts the original conceptual framework in light of these new findings. The most relevant findings from this section can be summarized as follows:

The majority of users did not submit designs but still showed high involvement and higher satisfaction scores than designers.

Supporting the idea that consumer have difficulty articulating what they want, the most popular reason for not submitting a design was "Lack of Artistic Ability".

The primary reason for purchasing from Threadless was "Innovative Designs". Exclusivity was considered more important in the purchasing decision than involvement.

A large amount of support was available to designers and they showed a willingness to collaborate. Users provided feedback and suggestions to designers at every design stage. Community members who felt they lack the artistic ability to design could perform other functions in the design process (commenting, reviewing and rating). Conclusions and Limitations Results showed that Threadless offers an interesting balance of MC and niche production, giving customers a unique experience but asking them to agree collectively on which product best suits their needs. This aggregation allows the manufacturer to produce in small batches instead of producing one off unique items. This approach is further supported by the results of the questionnaire in which respondents placed "innovative designs" over involvement in the design process as the key reason they buy from Threadless. The first key finding was that consumers in this case seemed willing to make the trade off between creating a unique product as long as the received product was of

134


a higher standard that they would have been able to produce. This has potential impact for those businesses considering MC and offering a fully customized product.

Figure 3: Adapted conceptual framework.

The second key finding observed in the participant observation was the willingness of the VC to support each other and offer help and advice throughout the design process. This support can help overcome the problems that customers have during the MC process and reduce the support the manufacturer has to provide.


135

The final conclusion drawn from this research is that while technology is cited as removing communication barriers between manufacturers and customers, making it easier to receive want information, it is also suggested that this same technology facilitated the sharing of want information amongst customers. Threadless has developed a mechanism to aggregate this information and let its customers decide collaboratively the products which best meet their aggregated needs. It is proposed that this reduces the current disconnect between a manufacturers desire for production efficiency and reliable demand and a customers desire for involvement and unique (or at least limited) products. This may have implications for manufacturers trying with difficulty to offer the uniqueness that the MC literature proclaims customers want, when an equally effective albeit less radical approach may be available. The research hinted at a potential link between involvement and purchase intention. A majority of respondents felt that simply rating a design regardless of the rating given increased the chance that they would purchase it. Respondent 11 said that they felt they got "Way better scores when people feel they helped decide the final product..." This could have important implications as it may mean that regardless of the quality of the output simply being involved in the process might strengthen the likelihood of purchase. Contrastingly Expectation Disconfirmation theory "predicts that unrealistically high expectations will result in lower levels of perceived benefit than those associated with realistic expectations" (Staples et al. 2001;1) may suggest that increased involvement in the product creation process would only raises consumer’s expectations of that final product, which is an interesting potential paradox which would benefit from further research.

Appendix – Survey Instrument Q1

How old are you?

Q2

How many t-shirts have you purchased from Threadless in the past year?

Q3

Since your first purchase from Threadless, what percentage of ALL your t-shirt purchases have come from Threadless?

Q4

How many times a week do you visit the Threadless website?

Q5

Do you rate potential designs at Threadless?

Q7

Do you think it is more likely that you will buy a t-shirt that you have rated highly (given a rating of 3 or more), than one you have rated 2 or less?

Q9

How often do you view it?

Q10 How often do you post or reply to messages?

136


Q11 Have you ever posted requesting design assistance on the blog forum? Q12 Did you receive a response from other members offering to help? Q13 Have you ever responded to anyone posting, asking for design assistance? Q14 Approximately how many times have you responded offering assistance? Q15 How many designs have you submitted to Threadless? Q16 If you haven't submitted a design, why not? (Tick as many of the below as you agree with) Q17 How many of these have been selected for production? Q18 What do you consider to be the most important reason why you buy t-shirts from Threadless? Q19 Do you feel that you are part of a wider Threadless Virtual Community? Q20 If Yes, do you feel a sense of attachment to this community?

References Berger, C. Moslein, K. Piller, F. & Reichwald, R. (2005) Designing modes of co-operation at the customer interface: learning from exploratory research. European Management Review. 2(1): 70–87. Brown, S.L., Tilton, A. & Woodside, D (2002) The case for on-line communities. The Mckinsey Quarterly, 1. Collin, S (1999) Doing Business on the Internet. 3rd Edition, Kogan Page: London Davis, S. (1996) Future Perfect Addison-Wesley; Reading. Ernst, H. & Gulati, R. (2003) Virtual Customer Integration – Bringing the Customer back into the Organization. Franke, N. & Shah, S. (2002) How communities support innovative activities: an exploration of assistance and sharing among end-users. Research Policy 32(2): 157-178. Franke, N. & Von Hippel, E. (2003) Satisfying Heterogeneous User Needs via Innovation Toolkits: The case of the Apache Security Software Research Policy. 32(7): 1199–1215. Füller, J. & Hienerth, C. (2004) Engaging the creative consumer. European Business Forum (EBF); Issue 19 (Autumn 2004). Füller, J. Bartl, M. Holger, E. & Mühlbacher, H. (2004) Community Based Innovation: A Method to Utilize the Innovative Potential of Online Communities Proceedings from the 37th Hawaii International Conference on Systems Sciences 2004. Available at http://tinyurl.com/lcewxj. Hagel & Armstrong (1997) Net Gain: Expanding markets through virtual communities. Boston, MA: Harvard Business School Press. Hart, C. (1995) Mass customization: conceptual underpinnings, opportunities and limits International Journal of Service Industry Management. 6(2): 36–45. Jeppesen, L. & Frederiksen, L. (2004) Why firm-established user communities work for innovation: The personal attributes of innovative users in the case of computer-controlled music instruments Working paper available at ideas.repec.org/p/ivs/iivswp/04-02.html [Accessed 10th March 2006] Lampel, J. & Mintzberg, H. (1996) Customizing Customization Sloan Management Review, Fall 1996: 21–30.


137

Lee, S.E. & Chen, J. (1999) Mass customization Methodology for an Apparel Industry with a Future Journal of Industrial Technology. 16(1). Lee, S. Barua, A. Whinston, B. (1999) The complementary of Mass Customization and Electronic Commerce. Economics of Innovation and New Technology. 9(2): 81–109. Paccagnella, Luciano. (1997). Getting the Seat of Your Pants Dirty: Strategies for Ethnographic Research on Virtual Communities. Journal of Computer-Mediated Communication. 3(1). Piller, F. (2002) Customer interaction and digizability: a structured approach to mass customization. Working Paper, Available at www.mass-customization.de/download/pil2002-2.pdf [Accessed 05th October 2005.] Piller, F. & Berger, C. (2003) Customers as Co-Designers. IEE Manufacturing Engineer. August/Sept 2003. Piller, F. Moeslein, K. & Stotko, C. (2004) Does mass customization pay? An economic approach to evaluate customer integration. Production Planning & Control. (15)4: 435–444. Piller, F. & Muller, M. (2004) A New Marketing Approach to Mass Customization. Int J. of Computer Integrated manufacturing. 17(7): 583–593. Piller, F. Schubert, P. Koch, M. Moslein, K. (2005) Overcoming Mass Confusion: Collaborative customer co-design in online communities. Journal of Computer-Mediated Communication. 10(4). Piller, F. Walcher, D. (2006): Toolkits for Idea Competitions: A Novel Method to Integrate Users in New Product Development, Journal of R&D Management. 36(3): 307–318. Pine, J. (1993) Mass Customization: The new frontier in business competition. Harvard Business School Press; Boston. Pine, J. Victor, B and Boynton, C. (1993) Making Mass Customization Work Harvard Business Review. (71)5: 108–111. Pine, J. (1994) Customers don't want choice, Managers Journal, Wall Street Journal, 18th April, p.A14. Pine, J. (1998) Welcome to the Experience Economy. Harvard Business Review. 76(4). Porter, M. (2001) Strategy and the Internet. Harvard Business Review. 79(3): 62–79. Prahalad, C. Ramaswamy, V (2004) The Future of Competition: Co-Creating Unique Value with Customers. Boston: Harvard Business School Press. Rainie, L., & Packel, D. (2001) More online, doing more. Pew Internet & American Life Project. [online] Available from http://preview.tinyurl.com/nakmxn. Ridings, M. Gefen, D & Arinze, B. (2002) Some antecedents and effects of trust in virtual communities. Journal of Strategic Information Systems. 11:271–295. Sanders, H. (2001) Financial rewards of mass customization. MCPC 2001, Proceedings of the 2001 World Congress on Mass Customization and Personalization, Hong Kong University of Science and Technology. [online] available at http://preview.tinyurl.com/m4b2yq. Sawhney, M. & Prandelli, E. (2000) Managing Distributed Innovation in Turbulent Markets. California Management Review. 42(4): 24–54. Schubert, P. & Ginsburg, M. (2000) Virtual Communities of Transaction: The Role of Personalization in Electronic Commerce. Electronic Markets. 10(1): 45–55. Schubert, P. & Koch, M. (2002) The Power of Personalization: Customer Collaboration and Virtual Communities. Eighth Americas Conference on Information Systems. Teresko, J. (1994) Mass Customization or Mass Confusion. Industry Week, June 20, 1994.

138


Staples, S. Wong, I. & Seddon, P. (2002) Having expectations of information systems benefits that match received benefits: does it really matter? Information and Management. 40(2): 115–131. Von Hippel, E. (1986) Lead users: a source of novel product concepts. Management Sciences. 32(7): 791– 805. Von Hippel E. & Tyre, M. (1995) How Learning by Doing is Done: Problem Identification in Novel Process Equipment. Research Policy. 24(1): 1–12. Von Hippel, E. (1998) Economics of Product Development by Users: Impact of Sticky Local Information. Management Science. 44(5): 629–644. Von Hippel, E. (2005) Democratizing Innovation. Cambirdge, MA: MIT Press. Zipkin, P. (2001) The Limits of Mass Customization. MIT Sloan Management Review. 42(3): 81–87.

Author Biography Adam Fletcher first became interested in the concept of Crowdsourcing whilst at University, where he wrote his thesis on the topic. He is passionate about Virtual Communities, Open Innovation and chocolate. He has worked for several years as a Brand Evangelist managing Crowdsourcing contests and Virtual Communities. Adam started his career at Microsoft in the UK, then moved to Germany as the Industry Ambassador for the apparel platform Spreadshirt. At Spreadshirt he ran his own successful Crowdsourcing contest "The Open Logo Project 1.6". This led to an opportunity creating a smartphone management platform called Virtual Mobile which saw him live for a year in sleepy Auckland, New Zealand. Adam now lives once again in East Germany where he is selfemployed as an Internet Marketer. Adam is available for consulting, speaking opportunities, further research, a beer – in fact just about anything that gets him out of the house. Contact: www.thezig.co.uk | [email protected].

1.7

Before Pine and Dell: Mass Customization in Urban Design, Architecture, Linguistics, and Food William Mitchell MIT Design Lab, Massachusetts Institute of Technology, United States Ryan Chin MIT Design Lab, Massachusetts Institute of Technology, United States

Long before B. Joseph Pine II established a viable economic strategy around the concept of Mass Customization, and Dell Computer’s execution of a custom build-to-order strategy, combinatorial theory (configuring of modular components) and generative systems have been employed in biological systems, grammatical sentence structure in linguistics, and also in architectural and urban design. This paper will trace the conceptual roots of Mass Customization through the examination of historical precedents: (1) Design of cities via biological analogy (Aristotle); (2) Architectural form via grammatical analogy (Mitchell); (3) Precis des Lecons d'architecture (Durand); (4) Combinatorial Optimization (Newell, Simon); and (5) Culinary Arts. We will then discuss the limitations of such combinatorial methods and then lay out a conceptual framework for achieving high levels of customization using combinatorial methods. The work on the MIT Concept Car by the Smart Cities group of the MIT Media Lab will illustrate these principles.

Introduction B. Joseph Pine II (1993) describes the fundamental principles of mass customization by writing: "The best method for achieving mass customization — minimizing costs while maximizing individualized customization — is by creating modular components that can be configured into a wide variety of end products and services. Economies of scale are gained through components rather than the products; economies of scope are gained by using modular components over and over in different products; and customization is gained by the myriad of products that can be configured." This paper will examine closely the rules that govern the use of modular components and their relationship to product architecture in differing industries.

139

140


Design of Cities via Biological Analogy One of the earliest recorded dialogues on the use of combinatorial thinking (Mitchell 1978) is credited to Aristotle in Politics (section 1290) called "On Parts of Animals," where he discusses at length methods for the design of a city. Aristotle uses the following biological analogy to formulate a generative model consisting of interchangeable parts: "If we were going to speak of different species of animals, we should first of all determine the organs of sense and instruments of receiving and digesting food, such as the mouth and stomach, besides organs of locomotion. Assuming now that there are only so many kinds of organs, but that there may be differences in them — I mean different kinds of mouths, and stomachs, and perceptive and locomotive organs — the possible combinations of these differences will necessarily furnish many varieties of animals. (For animals cannot be the same which have different kinds of mouths or ears.) And when all the combinations are exhausted there will be as many sorts of animals as there are combinations of the necessary organs." William J. Mitchell, in his book Computer Aided Architectural Design (1978), further explains Aristotle’s use of this analogy. He writes: "In other words, he described a generative system for potential animals. He then continued on to argue that, in a similar way, designs for potential cities can be broken down into their essential constituent parts, listing the alternatives for each part, then taking various different combinations of alternatives." Grammatical Combination The foundation of languages is based in part on combinatorial strategies applied to the rules of grammar. Given the long and storied field of linguistics, this paper focuses on just a simple example of how combinatorial strategies help in the formation of sentences (English). Mitchell (1989) writes in The Logic of Architecture: Design, Computation, and Cognition: One powerful way to do this is to introduce the idea of grammatical combination of parts. We can, if we so choose, specify in the type definition of an architectural vocabulary an element that is only instantiated in certain kinds of combinations of other elements. That is, we specify certain external relations in the type definition. The analogy here with the parts of speech is close; it is essential to being an English noun that is only


141

instantiated in English sentences in certain kinds of combinations with other words, as given by the rules of English grammar. Thus not every string of English words is an English sentence: only strings that comply with the rules of English grammar count as sentences. He continues to introduce the idea of replacement rules which is a powerful substitution tool to generate new sentences. More specifically, recursive replacement allows designers to repeatedly apply a rule in order to create potentially infinite design sets. Mitchell (1989) considers a square that is divided into 4 smaller squares by splitting the square at its mid-point along any edge and connecting the midpoint to form new squares. By applying this rule again for each subsequent square a new solutions space is quickly populated. He notes that the Taj Mahal’s division of paths and canals exemplifies this pattern. Recursive replacement more commonly occurs in language; the following recreation of Mitchell’s original sentence tree (Figure 1) depicts a structure that can generate 32 different sentences for 5 variables with only two values for each variable (25 = 32).

Figure 1: Sentence tree (recreated from Mitchell 1989).

The power of grammatical combination is evident by the near-endless variation of properly formed sentences. Dell computer adapted similar techniques to create a

142


"build-to-order" strategy by allowing different combinations of selected (configured online) components within a given modular sensitive structure. If we examine Dell’s syntactic structure we realize that they employ bundling which is directly linked to their supply chain management system. For example, when inventory is high for a particular component they will rapidly change its configuration options to offer better prices for overstocked items which it commonly bundles with other components (Anderson 2004). Precis des Lecons d'Architecture (Durand) Even before the L'ecole des Beaux Arts and other classical schools of architecture were established, Leonardo da Vinci utilized combinatorial strategies to help him generate designs for churches (central planned). He realized that if he began with the simplest spatial forms (square, octagon, circle, or dodecagon), he would arrive at every conceivable central-plan church, without taxing his imagination, by the mechanical addition of circular, semi-circular, or octagonal ancillary spaces to the principle and cross-axes of his basic figures (Mitchell 1977). 16th century Italian architect, Andrea Palladio utilized parametrically defined sets of rules (grammars) for laying out building geometries. His masterpiece the Four Books of Architecture, describe implicitly grammatical combinations for the creation of Italian villas. The work by George Stiny, William J. Mitchell, and Larry Sass further explores the Palladian Grammar through generative demonstrations of Palladio’s unbuilt work. William J. Mitchell (1977) notes that the challenges of top-down or bottom up design processes is best exemplified by classical approaches: ...was based upon systematic exploration of alternative ways in which various elements from a fixed vocabulary could be assembled in different combinations to generate architectural forms (Banham 1960; Summerson 1963; Hernandez 1969). J. N. L. Durand’s Precis des Lecons d'Architecture (1803) began with a profusely illustrated discussion of different ways in which building elements (columns, walls, etc.) could be assembled to generate sets of potential "combinaisions horisontales" (plans) and "combinaisions verticales" (elevations), then continued on to discuss urban design in analogous terms. ("De meme que les murs, les colonnes, etc., sont les element dont se composent les villes. / Just as walls, columns, etc. are the elements from which buildings are composed, so buildings are the elements from which towns are composed)


143

J.N.L Durand’s Precis des Lecons d'Architecture provides a graphic illustration of architectural elements, their proportioning rules, and rules of assembly for each constituent component (Figure 2) into a complete formal building composition.

Figure 2: Precis des Lecon’s d’architecture illustration plate.

William J. Mitchell’s (1989) book, The Logic of Architecture, discusses the complementarity of these approaches and the need for higher level abstraction to resolve conflicts of directional approaches. Today, practically all architects have moved away from the classical traditions and styles of the Beaux-Arts, but still employ the key principles of element combination such as structural steel beams and columns, window modules, plug-in mechanical systems, standardized fixtures, and so on. Combinatorial Optimization (Simon) Herbert A. Simon (1968) reintroduces the concept of 'satisficing' in his book, The Sciences of the Artificial, whereby he makes a distinction between finding the "optimal" vs. "satisfactory" answer to a problem. Optimal answers often require exhaustive and thorough evaluation and analysis of all possible solutions

144


(combinations, in the case of mass customization). Whereas satisfactory answers are solutions that fit a more limited set of criteria, but are no less acceptable than optimal solutions when balancing the tradeoffs of time, effort, and cost of finding the optimal. Yet combinatorial optimization as a theory of filtering a solution space is both a historic and present day method of generating solutions and selection key solutions. (Other forms of combinatorial generation and optimization such as genetic algorithms (GA) have evolved from the fields of artificial intelligence and computer science. This form of heuristic search is particularly effective in the design of complex structures in the building industry.) Simon (1968) writes, "The well-documented methods of finding optimizing algorithms such as linear programming, control theory, and dynamic programming have been developed in research universities by some of the most distinguished logicians and mathematicians." Simon recognizes that design logic applied to world design problems does not always allow an optimum answer. He considers the combinatorial problem of the traveling salesman (Simon 1968) to illustrate this point: ...given the geographical locations of a set of cities, find the routing that will take a salesmen to all the cities with the shortage mileage. For this problem there is a straightforward optimizing algorithm (analogous to the max algorithm for chess): try all possible routings, and pick the shortest. But for any considerable number of cities, the algorithm is computationally infeasible (the number of routes through N cities will be N!). Given the computational power of even any modern desktop computer Simon’s example probably has less power to convince us that computational brute force is a viable solution. Mainframes aside, the question of resources still is more pertinent than ever when dealing with real world problems. Simon offers several strategies in applying design logic such as finding alternatives, Mean-Ends Analysis (MEA), and the logic of search (heuristic), each of which are detailed extensively in his writings. Culinary Arts The preparation, cooking, and presentation of a meal is an combinatorial task. Produce, meats, diaries, grains, etc. are combined within a set of rules that govern their taste, color, texture, appearance, preparedness (cooked level), and even how it should be consumed. Fast food lends itself to the use of components because customers desire a balance of choice and speed of preparation. Often this is accomplished by the pre-processing of basic modules, for example a hamburger joint will slice tomatoes, onions, shred lettuce, and have condiments in packages ready for quick assembly of a custom burger. This section focuses on the

145


relationship between product architecture and combinatorial methods of endproduct assembly. Given the world’s vast culinary history, we will focus solely on the fast food culture of the United States. The tree diagram described in the previous section on grammatical combination proves useful in illustrating the architecture of many culinary delights. Using the following food archetypes, we will diagram the product architecture of the following food categories:

The Sandwich

The Pizza

Sushi (Maki roll)

Chinese Combination Platters

Sandwich architecture A sandwich is defined as "two or more slices of bread or the like with a layer of meat, fish, cheese, etc., between each pair" (dictionary.com 2007). Given this definition we can construct a tree diagram consisting of two branches (Figure 3). Two or more slices of bread form one branch; the elements of the filling form the other. We can further subdivide the tree by moving down any branch of the tree. For example, the filling branch can be parsed into different categories like vegetables, fruit, meat, fish, cheese, sauce, etc. as described in the Filling tree (Figure 4).

Figure 3: Sandwich tree.

Figure 4: Filling tree.

One common instantiation of the sandwich tree is the Ham and Swiss cheese sandwich. The following tree diagram describes the architecture of such a sandwich (Figure 5).

146


Figure 5: Ham and Swiss cheese sandwich tree.

Such simple syntactic structures provide an elementary system of substitutions to create vast varieties of possible solutions for any given architectural genre. For example, a hamburger (now classified as a type of sandwich) can be created by substituting the bread with a hamburger bun; changing the fillings with the new elements of a hamburger patty and mustard; and subtracting the Swiss cheese (Figure 6).

Figure 6: Hamburger tree.

The architecture of a sandwich is based primarily on a layering scheme which places emphasis on the interface between materials. With the notable exception of the "Open Face" sandwich, most sandwiches depend on the binding of materials through surfaces, for example, mayonnaise is spread on one side of bread to bind

147


it to the next layer (perhaps a slice of meat). It is through the connectivity of these surfaces to each other that give the product some structural cohesion. The character of the layering is determined by the precision of the interfaces. A sloppily-made sandwich, where not much attention is paid to proper layering, will fall apart. A typical English afternoon cucumber sandwich with its crust trimmed illustrates a congruency of its material character. Often sandwiches with an overabundance of filler require additional structural care (a toothpick) beyond the cohesion created by sauce or some other binder. When the rules (or grammar) of the sandwich are radically violated the product begins to fade from the archetype. Pizza architecture A pizza is defined as, "a flat, open-faced baked pie of Italian origin, consisting of a thin layer of bread dough topped with spiced tomato sauce and cheese, often garnished with anchovies, sausage slices, mushrooms, etc." (dictionary.com 2007). As opposed to the sandwich, the pizza is primarily divided into three branches (Figure 7). The rules of substitution still apply to pizzas as they do with most culinary archetypes. To create a pepperoni pizza, we simply elaborate the tree by adding pepperoni and cheese to the toppings category (Figure 8).

Figure 7: Basic pizza tree.

Figure 8: Pepperoni pizza tree.

Common to both the pizza and sandwich is the established hierarchy within the product tree. At the highest level, the dough (pizza) and bread (sandwich) represent a complete module. The dough may be divisible (usually when making large batches, dough makers subdivide them into loaves for easy handling), but is treated as a complete module. The yeast, flour, water, and other ingredients used in making dough are bound together and cannot be separated after preparation. The pizza dough cannot be reversed after combining the wet and dry ingredients.

148


However, when treated as distinct modules (e.g., toppings before baking) we can exercise more flexibility. The tomato sauce is usually comprised of crushed tomatoes, herbs, and olive oil. Chefs normally bundle these ingredients together. If a customer requests tomato sauce with no oil olive in it, then the chef will either have to create a new sauce or perhaps refuse the order. Tradeoffs in bundling occur because the economies of scale favor making large batches of pre-made modules. This coupled with the reversibility characteristics of the product at any point in the product lifecycle and structure often determines which parts of the product architecture become nodal points (joints) in the tree. Sushi (Maki Roll) architecture In the Sushi tradition, a Maki Roll is defined to be: "cold boiled rice moistened with rice vinegar, usually shaped into bite-size pieces and topped with raw seafood (Nigiri-zushi) or formed into a long seaweed-wrapped roll, often around strips of vegetable or raw fish, and sliced into bite-size pieces (maki-zushi)" (dictionary.com 2007). Below is a simple diagram of a Maki roll that is divided into an outer skin, Sushi rice, inner filling, and sauce (Figure 9). Using the same method of substitution like that illustrated in the sandwich, a Tekka (Tuna) Maki (Figure 10) can be created by placing a piece of tuna as the inner filling. A Sake (Salmon) Maki Roll can be created by swapping out the Tuna for Salmon.

Figure 9: Maki roll tree.

Figure 10: Tuna maki roll tree.

As opposed to the sandwich and pizza architecture, the Maki Roll has a much higher degree of interchangeability of its parts. For example, a California roll inverts the outer and inner skin. An "inside-out" California Maki tree diagram illustrates this principle (Figure 11). The formation of a California Maki includes the use of substitution and interchangeability of components. The following incomplete list of elements (ingredients) illustrates this point (Table 1).

149


Figure 11: California maki roll tree. Table 1: List of sushi components. Outer skin Dried seaweed Sushi Rice Egg Eel Fish Roe Cucumber Tuna Yellowtail Salmon

Sushi rice

Inner filling

Sauce

Sushi Rice

Eel Tuna Yellowtail Salmon Cucumber Avocado Fish Roe Dried Seaweed Egg

Wasabi Spicy Mayo

The inherent flexibility at the highest levels of the Maki tree diagram yields an almost inexhaustible reel of possible solutions. As we add more components to the ingredients list, more possible solutions become possible. There are notable exceptions (soft shell crab) particularly in Americanized sushi such as the use of unroll-able ingredients like soft shell crab, which is almost never used as an outer skin, primarily because the culinary value of soft shell crab is in deep frying the entire crab. Fine chopping of this component is possible, but would ruin its inherit integrity as a monolithic and singular module. Sushi is similar to pizza and sandwiches. The final product is held together by the cohesion created by its interfaces. The act of rolling sticky rice (inherent cohesion), which encapsulates the other components, is one step in the product development process. Again, the preciseness in executing (e.g., cutting) and

150


binding the interfaces can be judged by a number of metrics like taste, presentation, proportions, smell, and so on. Chinese combination architecture With all due apologies to the endless variety of traditional Chinese food, we will focus on Americanized Chinese food which is found in the neighborhoods of Chinatown throughout the U.S. and abroad. In particular the concept of a Chinese combination platter (normally served for lunch) is a food archetype warranting critical analysis. A Chinese combination platter is defined as a complete meal with at least two distinct elements which can be served separately (e.g., egg rolls, Kung-Pao chicken, shrimp Lo Mien, etc.). Figure 12 illustrates the minimum number of branches in a combination platter. The architecture of the combination platter is highly additive. Different combination platters can be built ad infinitum by simply adding more items. Like the previous examples, substitution is one of the keys to producing variety in the final product (Figure 13).

Figure 12: Minimum branch tree.

Figure 13: Maximum branch tree.

A typical platter served at a Chinese restaurant can be described as follows (Figure 14). Like Maki Sushi, Chinese combination platters also have high levels of interchangeability. A combination platter can have two appetizers instead of one appetizer and one main dish. Such flexibility allows the restaurateur to cater to the wishes of the customer. Depending on whether you are eating in or taking out, combination platters have different strategies for packaging. Restaurants normally serve Chinese combination platters on a simple plate (some have small divisions to keep sauces separate). Take-out orders come either in distinct boxes or in a styrofoam or aluminum container. The container normally subdivides for ease of serving the proper portions and becomes, in essence, the equivalent of the binder of the product. The Bento box in the Japanese tradition is a more evolved and refined example of such a platform (normally served in the restaurant).


151

Figure 14: Chinese food combination platter tree.

Why food? Dissecting these common food type places focus on some key themes that reoccur throughout product and service development. Combinatorial strategies as demonstrated by abovementioned precedents and the selected culinary examples have significant cultural and economic resonance. History shows these strategies inherit the characteristics of vast variety, flexibility, and adaptability to differing contexts. The endless variety created by the product schemas has generally enriched each food category. Modern society is now faced with the key issues of efficiency in the design, manufacture, and delivery of both commodity and non-standard products and services. To respond to these changes and the changing wants and desires of the user/customer, the restaurateur can employ a strategy of high modularity to achieve the adaptability needed to suit unique customer requirements. Many establishments allow customers to create their own dishes by either modifying the menu or even to create dishes from scratch. This is often enabled by the use of pre-made modular components. Oishii, an upscale Japanese restaurant in suburban Boston, popularizes customer created sushi by posting a top ten sushi creations list in the restaurant. The best creations are eventually integrated into their main menu. The power of emergence in the creative process has yielded both unpredictable and pleasant surprises. Poor combinations are also allowed, but are eventually filtered out by the chef (an expert) and popular opinion (top ten worse sushi list). The culinary arts have developed over time as an evolutionary tradition with experimentation as a key driver for improvement and creativity. The food chain, California Pizza Kitchen (CPK) popularized a new genre of pizzas by introducing new combinations and utilizing non-traditional toppings such as Peking Duck, Tandori Chicken, Japanese eggplant, and so on. Similar to the Maki example, CPK introduced new ingredients (modules) to the parts list and generated new

152


designs. CPK has been able to accomplish 1) good flavor combinations, 2) aesthetics in presentation, and 3) a powerful brand by reinterpreting traditional pizza making. These metrics add up to an economically viable business based on combinatorial processes. Mass customizers will need to look across all industries for scaleable and modular processes based on the smart structuring of product architectures to be successful in the new global world market. Limitations of Combinatorial Methods Herbert Simon’s traveling salesmen parable constructs a computationally intensive problem of combinatorial optimization. Another set of underlying limitations to combinatorial methods is minimizing cost. Herbert Simon’s "Design as Resource Allocation" (1968) states two generally implicit ways design is concerned with cost: First, conservation of scarce resources may be one criteria for a satisfactory design. Second, the design process itself involves management of the resources of the designer, so that his efforts will not be dissipated unnecessarily in following lines of inquiry that prove fruitless. More and more cost calculations have been brought explicitly into the design procedure, and a strong case can be made today for training design engineers in that body of technique and theory that economist know as "cost-benefit analysis." In design, as opposed to mathematics, cost constraints creep into the design process as argued by Simon. Another limiting factor in combinatorial methods is the power of the designer herself. Historically an innumerable set of design processes ranging from randomization to rationale principles of emergent grammars have been utilized by designers. Part of this process described by Mitchell (2001) is the ability to intuit using design sensibility. He describes here the limitations of combinatorial search through the example of Mathias Roriczer’s writings on the design of cathedral pinnacles (1486): Creative "re-reading" of shapes, and the subsequent production of variants based upon such re-readings, is an important part of manual design processes. Any computational scheme that prematurely imposes a definite way to parse a composition into parts and subparts will inappropriately constrain a designer’s capacity for creative generation of alternatives. It will become a Procrustean bed. Some alternatives simply won't be considered because of stylistic preferences or the lack of proper filtering or fitness functions to reduce the solution space to meet cost/time constraints. Frank Gehry’s design process is well documented as a nonlinear exploration of combinations. Often his process begins with an exploration


153

of programmatic requirements of a building in the form of program blocks. These blocks are used early in the design process to explore a functional layout of adjacencies and spatial relations under certain site constraints. As the process evolves, assumptions and hierarchies are broken if a powerful design emerges, thus challenging and changing the original framework. George Stiny (2001) discusses the same issue when using shape grammars: Design is more than sorting through combinations of parts that come from prior analysis (how is this done?), or evaluating schemas in which division are already in place. I don't have to know what shapes are, or to describe them in terms of definite units -atoms, components, constituents, primitives, simples, and so on – for them to work for me. In fact, units mostly get in the way. How I calculate tells me what parts there are. They're evanescent. They change as rules are tried. Classical combinations in the culinary world have evolved over time in complex ways. Some classic combinations like "meat and potatoes" almost undoubtedly result in harmonious and delicious meals if executed properly. With trade came the cross-fertilization of goods with herbs, spices, meat, fruits, and vegetables, and new combinations of flavor became possible. Before America was explored, Gazpacho in Spain did not contain tomatoes. Today, fusion cooking dominates experimental cuisine and is rapidly becoming a part of our food vocabulary. Taste scientists research combinations of flavors in order to invent new mixes, however, the limitations of cost apply to food as they do to most endeavors. Popular TV shows like "the Iron Chef" also show the power of the creative individual to generate designs out of seemingly disparate combinations of elements (food in this case) that would have never been tried even with exhaustive optimization. Current Work in Mass Customization at MIT The Smart Cities Group of the MIT Media Lab has explored mass customization in the product design space since it was formed in 2003. Led by Professor William J. Mitchell, the group has worked with sponsors from industry and students throughout the MIT community to examine the role of design and product architecture in mass customization. The group launched the MIT Concept Car project in 2003 to co-develop with General Motors a vehicle that would showcase MIT design, engineering, and technology. The design of a mass customizable vehicle became one of the key goals of the project. This was enabled by establishing a modular product architecture consisting of in-wheel electric motors called "Wheel Robots" that plug in to the chassis of the vehicle. The Wheel Robots recombine all the essential drive

154


mechanisms of a traditional automobile into the wheel hub space of four tires. Independently-controlled direct electric drive motors provide propulsion and eliminate the need for gear boxes and a traditional transmission. The group also embedded steering, suspension, and braking in the space of the wheel hub. These design maneuvers provided both flexibility at the product architecture level and performance benefits (i.e., better handling) because the Wheel Robots are upgradeable and replaceable as stand-alone modules similar to a disc drive in a computer. The Wheel Robots only require 1) power, 2) signal, and a 3) mechanical connection to the chassis to function. The radical re-modularization of the traditional vehicle architecture disentangles an integrated complex system into a simpler product platform with localized complexity (in the wheels), thus freeing up the cabin interior and exterior for more customization. A traditional vehicle would require internal space for the powerplant and mechanical couplings to transit power to the wheels. This space now can be used for additional space for passengers or for storage. This vehicle architecture fundamentally changes vehicle manufacture from an integrated assembly system to a distributed modular system. For example, Wheel Robots, after going through rigorous engineering development and manufacturability analysis, can be produced en mass at centrally located manufacturing plants, whereas the body/cabin can be regionally or locally designed and manufactured. Wheel Robots would simply be shipped to each local assembly plant to be plugged into the rest of the vehicle. This separation allows each manufacturer the flexibility to customize for specific contexts created by cultural and regulatory differences. The Wheel Robot modular product architecture allows designers the freedom to explore areas formerly constrained by traditional platform layouts. The following matrix (Figure 15) illustrates the variety of designs based on this strategy. High customizability in the "end-product" becomes possible once a flexible modular architecture is in place. Ensuring high variety in the resulting endproducts requires standardization of the interfaces between modules. For example, a Wheel Robot architecture is only successful if the connection between the Wheel Robot unit and the chassis is standardized for power and signal protocols. The mechanical connection would also need to be designed for interchangeability across platforms. Once this is in place, then open market competition based on these standards will provide not only variety in the Wheel Robots themselves (i.e., high performance vs. compact size wheel robots), but also in the overall vehicle itself (For more on the design and engineering development of the MIT Concept Car project please visit cities.media.mit.edu).


155

Figure 15: Automotives designed based on the Wheel Robot vehicle architecture (Images by: Patrik Künzler, Mitchell Joachim, Marcel Botha, Raul-David "Retro" Poblano, and William Lark, Jr.).

Conclusions History shows us that combinatorial processes have been embedded in our thinking and in our methods of production. Our need to individualize thus far is tempered by the balance between efficiency of production methods and the ability to match (or configure) a desirable and manufacturable solution to each customer. This has been accomplished by the use of modular product architectures, some with the generative power to produce endless variations in the end-product. This paper has introduced a number of combinatorial strategies employing various levels of sophistication in their syntactic structure ranging from the complexities of language to simple substitution of product modules. Mass customization as an economic model itself will also need to establish itself as viable lifecycle process and not just simple configurations tied to complex supply chain networks. More research in the areas of open innovation, customer co-creation, generative design, solution space optimization, and rapid manufacturing will begin to shape a vision of mass customization as a highly sophisticated means of making individualized products and services.

156


References Anderson, D.M. (2004) Build-to-Order & Mass Customization. Cambria, California: CIM Press. Dictionary.com. (2007). Dictionary.com definition of the word sandwich. [online]. Available from: dictionary.reference.com/browse/sandwich [cited 2 April 2007]. Dictionary.com. (2007). Dictionary.com definition of the word sushi. [online]. Available from: dictionary.reference.com/browse/sushi [cited 2 April 2007]. Dictionary.com. (2007). Dictionary.com definition of the word pizza. [online]. Available from: dictionary.reference.com/browse/pizza [cited 2 April 2007]. Pine, II, B. Joseph (1993). Mass Customization: The New Frontier in Business Competition. Boston, MA: Harvard Business School Press. Mitchell, William J. (1977). Computer Aided Architectural Design. New York: Petrocelli. Mitchell, William J. (1989). The Logic of Architecture. Cambridge, Mass.: MIT Press. Mitchell, William J. (2001). Vitruvius Redux: Formalized Design Synthesis in Architecture. In: Erik K. Antonsson, Jonathan Cagan (ed.). Formal Engineering Design Synthesis. New York: Cambridge University Press. 11. Simon, Herbert (1968). The Sciences of the Artificial. Cambridge, Mass.: MIT Press. Stiny, George (2001). How to Calculate Shapes. In: Erik K. Antonsson, Jonathan Cagan (ed.). Formal Engineering Design Synthesis. New York: Cambridge University Press.

Author Biographies William J. Mitchell, Professor of Architecture and Media Arts and Sciences at MIT, holds the Alexander W. Dreyfoos, Jr. (1954) Professorship and directs the Media Lab’s Smart Cities research group. He was formerly Dean of the School of Architecture and Planning and Head of the Program in Media Arts and Sciences, both at MIT. He teaches courses and conducts research in design theory, computer applications in architecture and urban design, and imaging and image synthesis. A Fellow of the Royal Australian Institute of Architects, Mitchell taught previously at Harvard’s Graduate School of Design and at UCLA. His most recent book, Placing Words: Symbols, Space, and the City was published by MIT Press. His earlier books include: ME++: The Cyborg Self and the Networked City; E-Topia: Urban Life, Jim—But Not As We Know It; the edited volume High Technology and Low-Income Communities (with Donald A. Schon and Bish Sanyal); City of Bits: Space, Place, and the Infobahn; Digital Design Media (with Malcolm McCullough, two editions); The Reconfigured Eye: Visual Truth in the Post-Photographic Era; and The Logic of Architecture: Design, Computation, and Cognition. Contact: cities.media.mit.edu Ryan Chin is a fourth-year PhD student at the MIT Media laboratory in the Smart Cities research group. He is building the car of the future – the CityCar – a foldable, stackable, sharable, electric, two-passenger vehicle that rethinks urban mobility. The project, a collaboration with General Motors, tackles the problems of parking, congestion, energy efficiency, and carbon emissions in cities. In 2007 Chin lead a team of Media Lab students


157

in the creation of the RoboScooter – a lightweight electric folding scooter designed as a clean, green mobility solution also for cities – in collaboration with Sanyang motors (SYM) and ) and Industrial Technology Research Institute of Taiwan (ITRI).Both projects are exploration platforms for urban design, mass customization, and technological innovation. Chin at MIT earned a master of science in media arts and sciences and a master of architecture; and bachelor’s degrees in civil engineering and architecture from the Catholic University of America. Contact: cities.media.mit.edu | [email protected]


2

Consumer Studies & Marketing Aspects

159


2.1

Typology of Potential Benefits of Mass Customization Offerings for Customers: An Exploratory Study of the Customer Perspective Hans H. Bauer Department of Business Administration and Marketing II, University of Mannheim, Germany Anja Düll BASF SE, Ludwigshafen am Rhein, Germany Dennis W. Jeffery Lexware GmbH & Co. KG, Freiburg, Germany

In order to utilize the full potential of mass customization (MC), the first objective of this study is to develop a differentiated typology of potential benefits which different MC offerings can generate for consumers. Therefore, a theoretical-conceptual approach is complemented by 20 in-depth interviews. The results suggest that style customization especially is capable of generating symbolic, emotional, hedonic and epistemic benefits, and seems to be attractive with regard to presents. Fit and functionality customization mainly offer possibilities for the generation of functional benefits, such as quality and comfort, but also have positive effects on the physical health. Furthermore, personal and economical benefits are of importance. The second objective is to analyze the central factors influencing the evaluation of the attractiveness of MC offerings and the demand of different age groups. For this purpose focus group surveys with pupils, students, middleaged employees and over 50s were conducted. Type, extent and place of customization, service support and type of purchasing decision could be identified as central factors of MC offerings. Budget and time shortage, use of attractive alternatives to mc, certain individual characteristics and dissatisfaction with standardized products represent important consumer-related factors. Finally, differences between the regarded age groups are laid open.

Introduction Due to fierce competitive conditions in consumer markets many companies show an increasing interest in mass customization (mc). By offering customizable mass products companies expect to realize significant advantages in competition through the generation of enduring customer value (Pine 1993; Fiore et al. 2001;

161

162


Reichwald and Piller 2006). In this regard, there is still a lack of comprehensive marketing research. The majority of existing articles on MC analyze the concept from the supplier perspective. The customer perspective has only been partly researched so far (Kaplan and Haenlein 2006; Hunt 2006). Until now, there are only few findings concerning the benefits customers expect from mass customized offerings beforehand (Schreier 2005), which factors are central for the evaluation of the attractiveness of such offerings and what customers want from certain mass customized offerings (Goldsmith and Freiden 2004). Since this is a basic requirement for the value-oriented and target group appropriate market development, the article at hand aims at contributing to the existing MC literature by providing basic exploratory findings. Literature Review The concept of mass customization Mass customization offerings in consumer markets enable customers either to actively align a standardized product according to one’s own specific requirements or at least to deliberately initiate its customization before the actual purchase (Schneider 1998; Steck 2003). Consequently, the alignment postulates a minimum of customer activity and customer integration in the phase of prepurchase (Kaplan and Haenlein 2006; Ihl et al. 2006) and results in an individualized product (Schneider 1998; Bardakci and Whitelock 2003). The specification of the desired customization can either be carried out online by using product configuration systems or can be performed in actual shops at retail (Ihl et al. 2006). Stylistic and functional customization options can be distinguished as specific MC approaches: stylistic customization options allow customers to individualize optical or other sensual product components. Functional customization options permit customers to choose, specify or omit various functional aspects of a product. In certain cases, products can be customized in order to meet specific requirements of the human body (body-fit customization) (Piller and Müller 2004; Reichwald and Piller 2006). The extent of customization can vary to a high degree. Normally, modular constructed MC systems simply enable customers to select certain attributes from a set of pre-defined product features (e.g. selection of a specific color or function) (Kaplan and Haenlein 2006). However, MC offerings can provide a much higher extent of customization, e.g. style customization, which provides the opportunity for customers to create their own product design.


163

By increasing the extent of customization the uniqueness of a product rises, but concurrently so do costs and time of production. From a customer’s perspective this affects purchase price and waiting time which have to be sacrificed for the product. In various studies, the extent of customization is only determined implicitly and imprecisely by the fundamental goal of achieving MC products at low costs comparable to the costs of standardized mass products or at least at a low cost (e.g. Shen and Ball 2006; Kaplan and Haenlein 2006). In order to include the whole spectrum of MC offerings, there is no limitation in reference to a specific extent of customization in this article. Thus, the existing trade-off for the customer between a high degree of customization and low prices can be analyzed. Literature review In some basic conceptual articles the interest of consumers for MC is derived from the generally observed social change, which determines the increasing importance of values such as individuality and hedonism (Snyder and Fromkin 1977; Schneider 1998; Blaho 2001; Tepper Tian, Bearden and Hunter 2001). In addition Hart (1995) points out that customers' susceptibility towards customizable offers is determined by the uniqueness of their needs and the gap between desired and realized aspects in a certain product category (q.v. Guilabert and Donthu 2006). Furthermore, in other conceptual articles the costs associated with MC are discussed. These costs have a negative impact on customer acceptance. They include the perceived risk when purchasing a customized product, the payment of a price premium, the acceptance of waiting time and the time and effort involved (Bardakci and Whitelock 2003). So far, it has not been analyzed how the readiness of customers to accept these costs varies with different offers. Shen and Ball (2006) clarify that the value of an individualized offer from the customer perspective results out of the perceived product and process benefits less the perceived process costs and monetary expenses. However, the customer response towards MC offerings still has to be researched comprehensively (Simonson 2005). The limited number of empirical studies about MC from a customer perspective can be divided into two categories. The first category focuses on questions during or after the process of the actual customization. The second category sets in before customization. The majority of articles relates to the first category in which the handling of product configuration systems and customer satisfaction during and after product customization is the primary concern (e.g. Huffman and Kahn 1998; Franke and Piller 2001; Kurniawan, Tseng and So 2006; Dellaert and Stremersch 2005; Bharati and Chaudhury 2006; Schreier 2006; Franke and Schreier 2006). Concerning the first category, especially the work of Schreier is relevant for the

164


present study. He appraises the value increment of self-designed products via toolkits and theoretically interprets this increment subsequently (Schreier 2005; Schreier 2006). Schreier also identifies the theoretically derived benefits (functional benefit, pride, uniqueness and process benefits) in 48 interviews with consumers which were conducted after customization (Schreier, Mair am Tinkhof and Franke 2006). Nevertheless, it has to be stressed that the subjects in the context of the survey were encouraged to customize a product. Hence, they provide no insight concerning the expected benefits which initiate the purchase of a MC offering by a customer (Dellaert and Stremersch 2005). Although an optimization of the actual product configuration only makes sense if it becomes clear which customers wish MC and what benefits they expect from it, a lack of research in this second category has to be stated. As yet, solely isolated studies focussing on a specific product range have been conducted (Euroshoe Consortium 2002; Fiore, Lee and Kunz 2004; Kreuzer, Kühn and Michel 2007). Within the EUROShoE project for instance, an attempt to evaluate the market potential of customized shoes in the European shoe industry has been undertaken (Euroshoe Consortium 2002). The customer readiness to use a MC offering in the clothing industry has been explained in causal models by the need to experiment with one’s appearance, the need to highlight one’s individuality, the optimum stimulation level (Fiore, Lee and Kunz 2004) and the need for uniqueness (Kreuzer, Kühn and Michel 2007). Only Hunt (2006) compares several products (mobile phone, book bag, desk chair and alarm clock) with each other in his survey. Yet, he cannot provide empirical evidence for the moderating effect of consumption visibility on the relationship between the value of a MC product and its determinants. In his model these determinants are: involvement in functional benefits, involvement in symbolic benefits, centrality of visual product aesthetic, the need for uniqueness and the need for optimization. He calls for further research on the importance of product categories for the evaluation of MC offerings. So do Shen and Ball (2006) and Schreier (2005). The individual difference variables which have been used for the explanation of the positive attitude towards MC offers do not bear any concrete reference to the offer. In addition, the compilation of these variables in Figure 1 shows that they particularly can be put in relation to the customized product (especially to the style customization option) and not in relation to the process. Apparently, the potential of MC has not been explored and used completely so far. Hence, apart from the Fiore et al. (2001) study there is no discrimination between different types of customization (stylistic and functional). The extent of customization is barely brought up as a vital issue. The influence of the place of customi-

165


zation (shop versus internet) on the evaluation of the attractiveness of the offerings has not been analyzed either (Ihl et al. 2006). It is assumed though, that these offering characteristics determine the attractiveness of the offering to a great extent. Analyzing individualized newspapers Kaplan (2006) determines to what extent the determinants "perceived usefulness" and "perceived ease of use" affect the acceptance by customers. By analysing the perceived usefulness, a reference to a concrete offering can be made. However, once again there is no systematization and precise definition of the various potential benefits which consumers expect from MC. Finally, a common shortcoming of numerous studies is the fact that exclusively students act as test subjects. This means that the results can only be generalized to a limited extent (Fiore et al. 2001; Kurniawan, Tseng and So 2006; Fiore, Lee and Kunz 2004; Bharati and Chaudhury 2006; Hunt 2006). Individual difference variables with positive influence on attitude towards MC

Needs

Additional individual difference variables

Put in concrete terms and in relation to… Product

Process

Need for individuality

Need to experiment with appearance

Need for uniqueness, Need for unique products Need for optimization

-

Centrality of visual product aesthetic

Optimum stimulation level

Involvement in symbolic benefits Involvement in functional benefits

-

-

Figure 1: Individual difference variables identified in the relevant literature.

Aim of study and procedure overview This study addresses some of the research deficits identified in our literature review. First of all our intention is the development of a differentiated and comprehensive typology of the potential benefits of MC offerings which are relevant for customers in the pre-purchase phase. Furthermore, the central factors influencing the evaluation of the attractiveness of MC offerings are to be analyzed and the demand for MC for different age groups concerning different product groups is to be exposed. The procedure is organized into three phases. Phase 1 and 2 serve the purpose of developing a differentiated typology of potential customer benefits. In Phase 1 which is merely conceptual, selected theories of customer value are transferred to

166


the concept of MC in order to develop the structure of potential customer benefits. Through this the theoretical foundation is ensured. In Phase 2 the precise definition and completion of customer benefits takes place based on empirical data. Qualitative in-depth interviews were chosen as the appropriate method of data collection. This method allows for a profound discussion of the topic with the test subjects and to expose psychological relations (Greenbaum 2000; Salcher 1995). Phase 3 serves for the exploration of the central factors influencing the evaluation of the attractiveness of MC offerings and of the demand for MC for different age groups with regard to different product groups. An adequate focus group survey is applied as a method of exploration (Greenbaum 2000). Typology of Potential Benefits Development of a benefit typology structure From the customer perspective products are viewed as a bundle of benefits, not attributes (Lai 1995). These benefits represent the variable degree of need satisfaction a consumer subjectively experiences through a product offering. It becomes clear that benefit is neither clearly supply nor merely demand-oriented. In fact, benefit establishes the interconnection between customers and offerings (Gerth 1965). In the course of the buying process, the actual benefits of goods and services cannot be judged by customers. Hence, the purchasing decision is normally based on the anticipated degree of satisfaction, i.e. the subjectively expected benefit (Perrey 1998). In the following, selected theories of customer value are transferred to the concept of MC in order to develop a conceptual structure of customer benefits. This is done with the objective to develop a typology of potential customer benefits of MC offerings. Thereby, it becomes easier to decide afterwards which aspects have to be focused upon in order to ensure a customer-oriented configuration and communication of the offerings. Initially, a distinction concerning the source of customer benefit is carried out. MC offerings are characterized particularly by process components. This stands in contrast to standardized offerings. Therefore a distinction is made in this regard. We proceed in accordance with Holbrook (1996), who distinguishes between active and reactive value, depending on whether value arises from an activity or an object. Thus we differentiate between the (active) customization “process” and the customized “product” as the final (reactive) result of customization. In order to obtain an improved understanding of the consumers’ expected benefits and to make them more tangible, various studies perform a differentiation of the


167

total value into several types of benefits (e.g. Vershofen 1959; Sheth, Newman and Gross 1995; Lai 1995; Holbrook 1996). This distinction is not organized uniformly. In the following, the essential types of benefit which can be transferred to MC offerings are presented in order to refine the structure of customer benefits. First of all, according to Sheth, Newman and Gross (1991) functional value can be asserted as a product benefit. Functional value arises from the tangible or technical features and the constitutional capacity of a product. It corresponds to Vershofen’s base value. Vershofen (1959) made a significant contribution to the conclusion that base value is by no means the sole pivotal criteria in the context of a buying decision. Instead, consumer choices are influenced by the emotional and intellectual components of additional value in many cases (Berekoven 1979). In the context of mc, the additional value of a product can be further subdivided. Lai (1995) describes the benefit acquired from the compatibility and consistency in a product constellation as a holistic benefit. This is interesting for MC offerings, because products can be aligned coherently. A symbolic benefit arises from a product’s capacity to serve as a symbol through which a person can gain social status. Vershofen (1959) denotes this type of benefit as prestige benefit. Holbrook (1996) refers to this as other-oriented benefit and Lai (1995) terms it as a social benefit. The importance of this type of benefit can be increased by the subjectively perceived uniqueness of MC products. Moreover, an emotional benefit can result from an individualized product. This includes the product’s capacity to arouse positive feelings or affective state of mind (Lai 1995). Holbrook (1996) emphasizes the self-orientation of this type of benefit. The process benefit of customization can be divided into hedonic and epistemic benefit. Hedonic benefit refers to the benefit acquired from a product’s capacity of meeting a consumer’s need for enjoyment (Babin, Darden and Griffin 1994; Lai 1995). Due to the fact that customers become active during the course of mc, there is a great potential to experience enjoyment and fun within the scope of this activity. Epistemic benefit refers to the benefit acquired from a supplier’s capacity to provide novelty or to meet a desire for knowledge (Lai 1995). Generally, this type of benefit can occur by integrating customers into the process of product specification (customers perform a new role) or particularly by implementing sophisticated procedures and technologies (such as product configuration systems and body scanners). Precise definition of benefits In order to precisely define and complete the customer benefits, 20 half-hour indepth interviews were conducted from the 29th of September until the 30th of

168


October 2006. Thereby, an additional differentiation concerning the type of customization was realized (stylistic versus functional customization). The recruiting of the test subjects was carried out via posters and flyers in Mannheim. Besides, attention was called in some newsgroups. To ensure a comprehensive representation of potential benefits, both men and women (50% respectively), as well as a broad age spectrum (20 to 60 years) without and with MC experience in different product classes was considered. In order to guarantee comparability, the interviews were conducted by means of a semi-structured interview guideline. As a start the concept of MC was explained. The following questions were verbalized openly. By using the laddering method, the deep-rooted views of the individuals could be uncovered (Greenbaum 2000). Furthermore, 16 visualized product images served as stimuli. The interviews were recorded digitally, transcribed and evaluated via qualitative content analysis (Mayring 2005). In the process, an inductive generation of categories was carried out (Mayring 2003). Results Concerning the product benefits, the identification and precise definition of functional benefits could be accomplished. The central points of functional customization (Figure 2) are a superior perceived product quality (by realization of the desired product configuration), an increased convenience in using the product (e.g. by omitting unwanted product attributes). Customers expect increased convenience, fitting accuracy and sometimes even positive effects on their physical health from the body-fit customization option (e.g. by adjusting the dimensions of a product to the body when having back problems or by determining the product ingredients when being allergic to certain substances). Regarding style customization an aesthetic benefit acquired from the product’s capacity to present a sense of beauty could be identified. Besides, the holistic benefit of harmonizing several products in terms of color became apparent. This is particularly beneficial for customers who consider product aesthetics to be of high importance. But customers also expect a holistic benefit by the possibility of customizing a product’s functionality because of the improved match with other products. Symbolic benefits were only identified while making use of style customization. This offers the possibility to demonstrate one’s personality through the optically individualized product (“style that suits me”) and like that to differentiate oneself from others (“unique style which no one else has”).

169


Besides, regarding style customization a kind of pride in the self-customized product could be revealed as an emotional benefit. As another emotional benefit of both stylistic and functional customization the pleasurable feeling of indulging in something special could be identified. Furthermore it became obvious that the test subjects expect a lower degree of annoyance, which also leads to an emotional benefit (Figure 2). The epistemic benefit has explicitly been established as a process benefit. Many test subjects evaluate active product customization as an interesting and novel experience. They regard style but also functional customization to have the potential of satisfying their need for variety and novelty. Expected fun and entertainment were revealed as concrete definitions of hedonic benefits, particularly in reference to style customization. Source of Benefit

Type of Benefit

Functional

Holistic Aesthetic Product

Symbolic

Emotional

Epistemic Process

Hedonic Personal

Concrete MC offering

Economical Temporal

Type of Customization Functional Higher quality/ functionality, higher convenience, better fitting accuracy, sometimes positive effects on health Visual match with other Compatibility with other products products Higher degree of aesthetics Self-expression, differentiation from the crowd Pride Pleasurable feeling of Pleasurable feeling of indulging in indulging in something special something special Enjoyment of product/less Enjoyment of product/less annoyance annoyance New insight, New insight, experimentation experimentation Fun, entertainment Control and influence Control and influence (ealize own ideas) (ealize own ideas) Improved price/performance Improved price/performance ratio, lower total price by ratio omitting unwanted features Shorter Shorter searching time searching time Stylistic

Figure 2: Differentiated typology of potential customer benefits.

170


Moreover, a new type of benefit could be identified in the category of "process benefits of customization". We term this type of benefit a personal benefit. It becomes clear within customization of both functional and stylistic aspects by the possibility of increased control and influence on the product’s configuration. According to customers, they feel better if they are put into a position where they can realize their own ideas. Besides especially style customization offers the possibility of implementing own ideas to a high degree. Apart from the customized product and the process of customization a third source of benefits has been discovered. This source of customer benefit is denoted as concrete MC offering, because the resulting benefits are based on the characteristics of a concrete offer (payment of a price premium, acceptance of waiting time). Primary, these characteristics of the MC concept create costs for customers. However, customer benefits in the form of economical and temporal benefits could also be identified here (Figure 2). An absolute economical benefit is particularly expected when choosing functional customization. So, customers expect a price reduction by omitting unwanted product features. An improved performance/price-ratio was also mentioned regarding style customization by improving the efficiency of a product. Due to a shortened searching time, customers expect a temporal benefit. This refers most notably to body-fit customization, but also to style customization. Central Factors in the Evaluation of the Attractiveness and the Demand of MC Offerings with different Age Groups Objective and procedure In order to reveal the central factors within the course of customer evaluations of MC offerings and to determine the demand for MC in reference to different product groups, four focus groups (group 1 = pupils; group 2 = students; group 3 = middle-aged employees; group 4 = over 50’s) were conducted from the 29th of November 2006 until the 17th of January 2007. Test subjects were recruited via posters and flyers at schools, at the university and at central shopping areas some weeks before. Finally, a total of 30 test subjects participated in the four focus groups with a duration of 60 to 90 minutes. The interviews were semi-structured and provided sufficient scope for broad discussions. The main part of the discussions was directed to the MC concept, its characteristics, chances, limits and evaluation. In addition, the demand for MC offerings in reference to different product groups (clothing, accessories, means of transport, technological products, books/cds, care products, games, furniture, domestic appliances, sports equipment,


171

sportswear, gardening products and foods) was visualized by means of a point rating system. In the course of this procedure each test subject was handed five star shaped stickers, which were assigned to the 14 visualized product categories, to a "?-poster" or an "unnecessary-poster" depending on the individually perceived attractiveness of the offering. The "?-poster" stands for desired product categories which were not given, the "unnecessary-poster" for the non-existing demand for MC offerings. The independency of the valuations was ensured. The interviews were recorded, transcribed and supplemented with the point allocation. Afterwards, they were interpreted by means of qualitative content analysis. Results Certain features of an offering as well as consumer-related factors were identified as the central factors determining the evaluation of the attractiveness of MC offerings. Concerning the offering, it became obvious that the type of product has a significant impact on the evaluation of the attractiveness of MC offerings for all test subjects. Regarding high-involvement purchases, i.e. major purchases, purchases which are realized less often and purchases which are of high importance to customers, MC offerings are valuated higher in attractiveness. In contrast, the costs associated with MC are perceived to be very high regarding lowinvolvement purchases. Likewise, the type and extent of customization determines the perceived attractiveness to a high degree and have an influence on customers` readiness to accept MC specific costs. In total, a higher beneficial contribution was attached to functional and especially to body-fit customization in many cases compared to style customization. The latter is perceived as "rather enjoyable" by the test subjects in reference to some products (especially in regard to accessories and if high emotional involvement is existent) but seems to be primarily attractive for one-time trying or to give away. The interest in extreme customization is high. At the same time, some of the interviewed consumers point out that they are prepared to accept higher costs when they contemplate an extreme MC offering. Then, on the other hand they are not willing to accept any compromises. For all test subjects, the place of individualization exerts an extensive influence on the attractiveness of MC offerings. On the whole, retail shops are rated higher. Although the internet holds the advantage that consumers do not have to frequent a shop twice, it is not considered to be attractive. The poor suitability of the medium is particularly criticized with regard to clothing and body-fit customization and leads to increased effort and risk. From the customers' perspective the offered service represents a decisive factor. Customers demand consultancy, price

172


transparency, exchange guarantees and the adherence to delivery time because they perceive the purchase of MC products to be of high risk otherwise. The evaluation of the attractiveness of a MC offering is also affected by central factors on the part of customers. Factors which encourage adoption are the dissatisfaction with standardized products, but also certain individual difference variables which can be addressed by MC offerings (such as need for variety, need for enjoyment, health and self-indulgence orientation). Factors which hinder adoption are budget and time shortages, the use of attractive alternatives to MC and individual difference variables which stand in contrast to MC (such as impatience, enjoyment of shopping, low confidence in own skills, lack of product knowledge). Specific results per age group Differences within the age groups could be particularly discovered concerning the rating of MC as unnecessary, the assessment of the attractiveness of the place of customization and minor differences concerning the demand for MC regarding specific product categories. In addition, it became clear that the factors encouraging and limiting adoption are the same but differ in their relevance within the age groups. In contrast to the assumption that MC offerings are especially attractive for young people (Goldsmith and Freiden 2004), the group of pupils only showed the third largest interest in customization options. They allocated 16.6% of the available points on the "unnecessary"-poster and stated that the main reason was the price premium charged for individualized products. Due to their limited budget, the price represents an important decision criterion. Besides, it showed that pupils like being creative and enjoy designing. They do this at home themselves, since this represents an attractive alternative. Moreover, a great interest in trying out new things could be observed in the discussion groups. For example, one of the study subjects reported that she frequently individualizes products via a configurator on the internet; however, she does this just for fun and has not bought any individualized products so far. The dissatisfaction with available standardized products is not particularly apparent. Bad mode of operation of technological products, as well as the existence of too many but not all desired product attributes is mentioned. Concerning shoes and trousers the bad fit of many products is criticized. Almost all of the pupils use the internet on a daily basis and feel comfortable about handling it. In comparison to the other groups regarded, they would most likely make use of customization offerings on the internet, since this is experienced to be convenient and they feel informed more objectively on the internet


173

than in a shop. However, it became apparent that pupils enjoy going shopping in the city now and again. They do not want to abandon this completely. A need for MC was particularly revealed for technological products and clothing, followed by furniture, accessories, sportswear and means of transport (Figure 3). The group of students showed the highest interest in customization options. They allocated none (0%) of their points on the "unnecessary"-poster. Their dissatisfaction with standardized products is more distinctive than the pupils'. Regarding technological products and means of transport they primarily complain about the impractical and inconvenient equipment as well as the existence of too many functions and the non-existence of desired product attributes. With regard to clothing, they criticize the fit of trousers and shoes as well as the appearance of specific pieces of clothing. They also emphasize lacking fitting accuracy and functionality with respect to sportswear and sports equipment. The students are also familiar with the internet; nevertheless, they would prefer the shop with regard to clothing and body-fit customization. The students show a strong need for fun. They show a demand for MC concerning technological products, clothing, accessories, furniture and sportswear. Additionally, they have a demand for customized sports equipment, care products and audio-books. Furthermore, one of the test subjects expresses the desire (product) for an individualized television program. The middle-aged employees show the second largest interest in customization options after the students, although, time shortage is a hindering factor. Their dissatisfaction with standardized products is comparable to the students'. What’s new is that they also address health issues (e.g. the suitability of specific customization offerings such as mattresses with adjusted slatted frame or individualized desk chairs when having back problems). They use the internet to a high extent too. With reference to most products and customization options they prefer the shop however. Desired shopping experience and consultancy are stated as reasons. As a whole, a high consciousness for health matters as well as a distinct leaning towards self-indulgence was expressed. The interest in MC concerns similar products as stated by the students. Additionally, domestic appliances and textiles are mentioned. The 50+ group exhibits the lowest interest in customization offerings as measured by the "unnecessary" evaluation. They allocated 20% of the points on the "unnecessary poster" (Figure 3). Simultaneously, they show the highest interest in the product category clothing (27.5% of the points). In this regard the fit, but also wearing comfort and own visual ideas are central aspects. Particularly female test subjects complain about the feeling of being pressured to follow fashion. The dissatisfaction of the 50+ group with standardized products and services is the highest. In certain areas, many persons within this group accept

174


the offerings of make-to-order producers such as tailors and cabinetmakers. Apart from furniture, there is a high need for customization and dissatisfaction with regard to functional aspects of domestic appliances within the 50+ group. Only few of them feel confident using the internet, some do not use it at all. All of them would make use of customization options only in shops. Health, as well as selfindulgence plays a major role. One of the test subjects expresses the desire for individualized (body-fit) hearing devices (product). Ranking

1

2

Pupils 16.60%

Students 17.50%

Employees 17.50% Technic. Products

Over 50s 27.50%

Clothing

Clothing

Technic. Products 13.30% Furniture

Technic. Products 12.50% Accessories

12.50% Clothing

10% Accessories

10% Furniture Sports Equipment

10% Furniture Means of Transport

6.60% Sportswear Means of Transport

7.50% Sportswear Care Products

5% Accessories Sportswear

12.50% Furniture Household Appl. 5% Accessories Means of Transport Care Products Textiles -

3

4

-

5% Means of Transport Books/CDs

16.60%

0%

5 Unnecessary

Clothing

Sports Equipment Books/CDs Household Appl. Textiles -

-

10%

20%

Figure 3: Results of the rating system.


175

Finally, it was also discussed to what extent individualized products are suitable for presents. Especially concerning small products and style customization groups 1 to 3 perceive the idea to be very attractive. The personal touch and token gesture which can be expressed thereby are experienced as very positive. The 50+ group is the only one that judges giving away individualized products as inappropriate. They expect difficulties in meeting the recipient’s taste. Summary and Outlook Although an ongoing interest in MC can be observed both in practice and research, there is still a considerable lack of research with regard to the customer perspective. While the actual configuration behavior and the subsequent customer satisfaction have been analyzed in several studies, up to now, there is no differentiated typology of benefit requirements which customers expect from MC offerings beforehand. But not until the potential benefits are laid open as a whole can the full potential of MC be utilized in an extensive and targeted way. Therefore, the first objective of our study was to develop a differentiated typology of potential benefits. In the first instance, a theoretical-conceptual approach was chosen with the objectives of creating a structure and a scientifical foundation. Furthermore, in order to precisely define the customer benefits, in-depth interviews with 20 test subjects were conducted. The interviews clarified that instead of functional benefits especially aesthetic, symbolic and emotional benefits do arise from style customization. Thus, it is recommendable to communicate these additional benefits in an emotional way. Expected hedonic and epistemic benefits play the central role concerning the customization process. The style customization process also offers the opportunity of having control and realizing own ideas. This potential can be utilized by an event-oriented marketing for instance. On the other hand, customization of functional aspects mainly offers possibilities for the generation of utilitarian benefits such as quality and comfort, but with the body-fit option also positive effects on physical health. Hence, MC offerings must be designed in such a manner that these possibilities can be realized. The focus group study pointed out that customers often rate these customization options to be more beneficial and therefore are willing to accept the higher costs associated with mc. Consequently, these advantageous aspects have to be pointed out to customers. With regard to functional customization, a potential personal benefit can result from the customization process. This is because customers take up a new and active position in the course of MC which allows for an increase in decision power and is perceived as beneficial. Regarding customization of

176


functional aspects, consumers expect an economical benefit in terms of a lower price by omitting dispensable product features. Moreover, this customization option can reduce customer displeasure, especially relating to the "over engineering" of technological products. With regard to body-fit and style customization the customers regard their shortened searching time for an adequate product as beneficial. The second objective of the study was the analysis of the central factors concerning the evaluation of the attractiveness of MC offerings and the MC demand of customers from different age groups with respect to specific product categories. For this, a focus group study with four age groups (pupils, students, middle-aged employees and over 50s) was conducted. The characteristics of MC offerings displayed in Figure 4 as well as central factors on the part of customers proved to be the central factors. Type of product

MC attractive with high-involvement purchasing decisions, i.e. with larger and seldom made purchases and products that are regarded highly important (otherwise the costs are considered to be very high)

Type of customization

A higher benefit is seen in functional and especially body-fit options increased willingness to accept costs

Extent of customization

Extreme customization is valued increased willingness to accept costs, but then less willingness to compromise regarding customization Low customization suited to one-off try out for fun or use as a present, particularly with style

Place of customization

Shop in many cases more attractive, especially with body-fit option and clothing (internet rated here as unsuitable) and for larger purchases (high risk perceived)

Service support

Service (consultancy), price transparency, exchange guarantee and adherence to delivery time are requirements (reduce perceived risk) Figure 4: Central features of the offering in MC evaluation.

On the part of customers, the central factors encouraging adoption were identified as the dissatisfaction with standardized products and highly distinctive individual characteristics. Mc can target these factors and generate corresponding benefits. Time and budget shortages, the use of attractive alternatives to MC as well as specific individual characteristics, which cannot be addressed with MC were recognized as the central factors hindering adoption. The differences existing within the age groups are presented in Figure 5.

177


Dissatisfaction Strong needs

Pupils

Students

Employees

Over 50s

Low

Medium

Medium

High

Like to have fun

High degree of health consciousness, selfindulgence orientation

High degree of health consciousness, self-indulgence orientation

Strong desire to try out new things, like to be creative

Time and budget shortage

Low budget, oncost represents main criteria

Implementation of attractive alternatives to MC

Like designing products themselves

Time important factor

Use offerings from individual producers

Figure 5: Differences within the age groups concerning the customer-related factors encouraging and hindering MC adoption.

In contrast to former assumptions, it is not only young people who exhibit a high interest in mc. Pupils only show the third largest interest. Tight budgets and the independent product customization can be stated as reasons for that. Students show the highest interest in MC offerings, followed by employees. For these however, time plays a more critical role. The demand for MC with regard to specific product categories is similar in groups 1 to 3: Technological products, clothing and furniture are at the top of their requirements list. Accessories, means of transport, sportswear and sports equipment score well too. A great potential for MC offerings results from these products if addressed to persons who are young or middle-aged. Although over 50s are dissatisfied with many standardized products, they still judge many MC offerings to be unnecessary because they are happy to make use of the offerings from make-to-order producers. They have a high interest in MC offerings for clothing. Concerning furniture and domestic appliances, they also find the offerings attractive. A shop concept proved to be essential for the effective development of this target group. In the context of communication, the present dissatisfaction which can possibly be overcome by MC should be emphasized. The highly pronounced needs which have been identified within the different age groups also represent interesting indicators which should be considered in future concepts of MC offerings.

178


Our study contributes to the existing research of mc. However, the qualitative approach to research does not enable the generation of quantitative statements. Hence, it is important to build on the results in a quantitative way. In order to implement the identified benefits in the future, the development of adequate scales is of particular importance in the first instance. Furthermore, the quantification of the influence of specific benefits on the willingness to make use of concrete MC offerings would provide useful insights. At this point, a segment specific analysis or a customization type (style versus functional) specific analysis would also be of interest. Future research activities should address product categories which are most notably attractive for MC from a customers' perspective and have not been researched comprehensively so far (e.g. technological products, furniture, accessories).

References Babin, Barry, Darden, William and Griffin, Mitch (1994). Work and/or Fun: Measuring Hedonic and Utilitarian Shopping Value. Journal of Consumer Research. 20(4): 644–656. Bardakci, Ahmet and Whitelock, Jeryl (2003). Mass customization in Marketing: the Consumer Perspective. Journal of Consumer Marketing, 20(5): 463–479. Berekoven, Ludwig (1979). Die Bedeutung Wilhelm Vershofens für die Absatzwirtschaft. Jahrbuch der Absatz- und Verbraucherforschung. 25(1): 2–10. Bharati, Pratyush and Chaudhury, Abhijit (2006). Product Customization on the Web: An Empirical Study of Factors Impacting Choiceboard User Satisfaction. Information Resources Management Journal. 19(2): 69–81. Blaho, Robert (2001). Massenindividualisierung: Erstellung integrativer Leistungen auf Massenmärkten. Bratislava: Ševt-Verlag. Dellaert, Benedict and Stremersch, Stefan (2005). Marketing Mass-Customized Products: Striking a Balance between Utility and Complexity. Journal of Marketing Research. 42(2): 219–227. Euroshoe Consortium (2002). The Market for Customized Footwear in Europe: Market Demand and Consumers' Preferences. euroshoe.itia.cnr.it/ Euroshoe/links/links.htm. Fiore, Ann, Lee, Seung-Eun and Kunz, Grace (2004). Individual Differences, Motivations, and Willingness to Use a Mass Customization Option for Fashion Products. European Journal of Marketing. 38(7): 835–849. Fiore, Ann, Lee Seung-Eun, Kunz, Grace and Campbell J. (2001). Relationships between Optimum Stimulation Level and Willingness to Use Mass Customization Options. Journal of Fashion Marketing and Management. 5(2): 99–107. Franke, Nikolaus and Piller, Frank (2001). Value Creation by Toolkits for User Innovation and Design: The Case of the Watch Market. The Journal of Product Innovation Management. 21(6): 401–415. Franke, Nikolaus and Schreier, Martin (2006). I made this myself! Exploring Process Utility in Mass Customization. Proceedings of the Summer American Marketing Association Conference. Chicago 2006, 14–15.

179


Gerth, Ernst (1965). Die Bedeutung des Verbrauchsnutzens für den Absatz. Berlin: Duncker and Humblot. Goldsmith, Ronald and Freiden, John (2004). Have it your Way: Consumer Attitudes toward Personalized Marketing. Marketing Intelligence and Planning. 22(2): 228–239. Greenbaum, Thomas (2000). Moderating Focus Groups: A practical guide for group facilitation. Thousand Oaks [et al.]: Sage. Guilabert, Margarita and Donthu, Naveen (2006). Mass Customization and Consumer Behavior: The Development of a Scale to Measure Customer Customization Sensitivity. International Journal of Mass Customization. 1(2/3): 13–23. Hart, Christopher (1995). Mass Customization: Conceptual Underpinnings, Opportunities and Limits. International Journal of Service Industry Management. 36(6): 2–36. Holbrook, Morris (1996). Customer Value: a Framework for Analysis and Research. Advances in Consumer Research. 23(1): 138–142. Huffman, Cynthia and Kahn, Barbara (1998). Variety for Sale: Mass Customization or Mass Confusion. Journal of Retailing. 74(4): 491–514. Hunt, David (2006). A Consumer Perspective on Mass Customization. University of Missouri-Columbia. edt.missouri.edu/Winter2006/Dissertation/HuntD-052506 -D4001/ short.pdf. Ihl, Christoph, Müller, Melanie, Piller, Frank and Reichwald, Ralf (2006). Kundenzufriedenheit bei Mass Customization. Die Unternehmung 60 (3), 165-184. Kaplan, Andreas and Haenlein, Michael (2006). Toward a Parsimonious Definition of Traditional and Electronic Mass Customization. Journal of Production Innovation Management. 23(2): 168–182. Kaplan, Andreas (2006). Factors Influencing the Adoption of Mass Customization: Determinants, Moderating Variables and Cross-National Generalizability. Göttingen: Cuvillier Verlag. Kreuzer, Michael, Kühn, Richard and Michel, Stefan (2007). Die praktische Relevanz individualisierbarer Massengüter aus Sicht der Nachfrager. Die Betriebswirtschaft 67 (4), 399-417. Kurniawan, Sri, Tseng, Mitchell and So, Richard (2006). Consumer Decision-Making Process in Mass Customization. International Journal of Mass Customization. 1(2/3): 176–194. Lai, Albert (1995). Consumer Values, Product Benefits and Customer Value: A Consumption Behavior Approach. Advances in Consumer Research. 22(1): 381–388. Mayring, Philipp (2003). Qualitative Inhaltsanalyse: Grundlagen und Techniken,. Weinheim, Basel: Beltz. Mayring, Philipp (2005). Neuere Entwicklungen in der qualitativen Forschung und der Qualitativem Inhaltsanalyse. In: Mayring, Philipp and Gläser-Zikuda, Michaela (eds.). Die Praxis der qualitativen Inhaltsanalyse. Weinheim, Basel: Beltz, 7–19. Perrey, Jesko (1998). Nutzenorientierte Marktsegmentierung: ein integrativer Zielgruppenmarketing im Verkehrsdienstleistungsbereich. Wiesbaden: Gabler.

Ansatz

zum

Piller, Frank and Müller, Melanie (2004). A New Marketing Approach to Mass Customization. International Journal of Computer Integrated Manufacturing. 17(7): 583–593. Pine, Buddie (1993). Mass customization: the new Frontier in Business Competition. Boston, Massachusetts: Harvard Business School Press. Reichwald, Ralf and Piller, Frank (2006). Interaktive Wertschöpfung: Open Innovation, Individualisierung und neue Formen der Arbeitsteilung. Wiesbaden: Gabler. Salcher, Ernst (1995). Die Psychologische Marktforschung. Berlin [et al.]: de Gruyter. Schneider, Paul (1998). Produktindividualisierung als Marketing-Ansatz. Schesslitz: Rosch-Buch.

180


Schreier, Martin (2005). Wertzuwachs durch Selbstdesign: Die erhöhte Zahlungsbereitschaft von Kunden beim Einsatz von "Toolkits for User Innovation and Design". Wiesbaden: Deutscher Universitäts-Verlag. Schreier, Martin (2006). The Value Increment of Mass-Customized Products: An Empirical Assessment. Journal of Consumer Behavior. 5(4): 317–327. Schreier, Martin, Mair am Tinkhof, Astrid and Franke, Nikolaus (2006). Warum "Toolkits for User Innovation and Design" für ihre Nutzer Wert schaffen: eine qualitative Analyse. Die Unternehmung. 60(3): 185–201. Shen, Anyuan and Ball, A. Dwayne (2006). How do Customers Evaluate Mass Customized Products? Proceedings of the Summer American Marketing Association Conference, Chicago 2006, 388–389. Sheth, Jagdish, Newman, Bruce and Gross, Barbara (1991). Consumption Values and Market Choices. Cincinnati [et al.]: South-Western Publishing Company. Simonson, Itamar (2005). Determinants of Customers' Responses to Customized Offers: Conceptual Framework and Research Propositions. Journal of Marketing. 69(1): 32–45. Snyder, Charles R. and Fromkin, Howard L. (1977). Abnormality as a Positive Characteristic: The Development and Validation of a Scale Measuring Need for Uniqueness. Journal of Abnormal Psychology. 86(5): 518–527. Steck, Werner (2003). Die Individualisierung der Kundenbeziehung im Finanzdienstleistungs-bereich. Hamburg: Kovac. Tepper Tian, Kelly, Bearden, William O. and Hunter, Gary L. (2001). Consumers' Need for Uniqueness: Scale Development and Validation. Journal of Consumer Research. 2(1): 50–66. Vershofen, Wilhelm (1959). Die Marktentnahme als Kernstück der Wirtschaftsforschung. Berlin, Köln: Heymann.

Author Biographies Prof. Dr. Hans H. Bauer leads the Institute of market-oriented Management at the University of Mannheim. Contact: www.bauer.bwl.uni-mannheim.de | [email protected] Anja Düll was Research and Teaching Assistant at the Department of Business Administration and Marketing II at the University of Mannheim. Her research focuses on consumer behavior, customer trust, the customer centric development of innovative offers, and the acceptance of innovations in consumer markets. In her thesis project she analyzed the customer perspective in mass customization. Today she is inhouse strategy consultant at BASF SE. Contact: [email protected] Dennis Jeffery graduated from the University of Mannheim in 2006. His thesis titled "Active individualization of mass products in consumer good markets" focused on the customer perspective on mass customization. Thereafter, he continued his research at the Department of Business Administration and Marketing II at Mannheim University as a coauthor of the article at hand. Presently, he lives in Freiburg, Germany, and is employed as a Service Manager at Lexware. Contact: [email protected]

2.2


Kate Herd Product Design and Engineering, School of Engineering and Information Sciences, Middlesex University, United Kingdom Andy Bardill Product Design and Engineering, School of Engineering and Information Sciences, Middlesex University, United Kingdom Mehmet Karamanoglu Product Design and Engineering, School of Engineering and Information Sciences, Middlesex University, United Kingdom

As mass customization develops, there is increasing understanding of how this practice can be implemented in terms of manufacturing capability and expertise, data transfer and management, and product architectures and business processes; however these critical elements fulfill only part of the story. The notion of designing for co-design is still relatively under-researched; co-design can be seen to consist not only of the specific codesign activities during product creation, but of the entire purchasing experience for the customer co-designer. This chapter presents a conceptual model and a design tool to support design for co-design, encompassing issues of increased emotional connection and positive customer experience.

Introduction As a paradigm persuasively championed as a business response to the saturated, segmented markets of the late twentieth and early twenty first century, mass customization (MC) became a reality when new manufacturing technologies, data transfer, and management methods became enablers for this new method of product creation. MC combined the efficiency of mass production with the differentiation possibilities of customization, creating variety through flexibility and quick responsiveness (Pine 1993, Tseng and Piller 2003). Central to much of this work has been the notion of 'solution space', a conceptual container for the breadth of product possibilities available to a customer co-designer, established through the assessment of product architecture, range, overall company strategy and manufacturing capability (Berger and Piller 2003), "the pre-existing capability 181

182


and degrees of freedom built into a given manufacturer’s production system" (Von Hippel and Katz 2002). This focus on the functional application and requirements of MC fulfils only part of the story. Mass customization by its very nature consists not only of the tangible product or service offering, but of the co-design experience for the customer. This experience differs from purchasing a mass produced product as it requires engagement and participation in the creation process. According to Schreier (2006, p.319) "the success of outsourcing certain design tasks to customers, that is, the success of mass customization, depends heavily on efficient and effective manufacturer-customer interaction". The notion of designing for co-design, however, remains relatively under-researched. Gamble et al. (2006, p.245) describe a customer experience as "a blend of a company’s physical performance and the emotions that it evokes"; how then can a MC customer experience be best understood, managed and designed for? This chapter proposes a conceptual model and a design tool to support the development of codesign experiences. The product envelope seeks to develop the existing notion of the solution space into a more complete model which reflects the wider context within which the solution space resides, unpacking the experience which surrounds and provide access to the tangible MC product at the core of the model. Useful as a hypothesis generating tool, the product envelope can be used to inform our understanding of the customer corridor, the customer journey through the codesign experience. The experience matrix brings these together to offer a framework within which opportunities for design can be clearly identified. Designing for Co-Design: The Story So Far MC alters the traditional product development process – whilst design remains a "conscious and intuitive effort to impose meaningful order" (Papanek 1997), MC moves towards a two-stage model – the first, the realm of company/designer establishing the solution space, the second, that of customer as co-designer; this second stage fundamentally changes the role of the customer from consumer of a product, to a partner in a process of adding value (Reichwald et al. 2004). However most customers are still buying made-to-order products manufactured in a mass production system – they are far from being "a very creative consumer" as predicted by Toffler almost thirty years ago (Piller et al. 2004). The literature describes a spectrum of research in the area of designing for co-design and has identified a number of salient design considerations, these include:

Offering a solution space which encompasses the designs that co-designers wish to create (Von Hippel 2001)


183

Minimising the potential complexity experienced by the co-designer, keeping their expenditure in the buying process as low as possible, whilst providing clearly perceptible benefits (Berger and Piller 2003, Kumiawan et al. 2003).

Enabling customers to engage in co-design activities without extra training, and for them to learn through cycles of trial and error (Von Hippel 2001).

Reducing cognitive overhead, which lies not only in extent of choice, but also in areas such as lack of understanding about which solution meets their needs, uncertainty about the behavior of the supplier, and uncertainty regarding the purchasing process, ordering and paying in advance for something that’s only been seen virtually (Franke and Piller 2003).

The extent and nature of choice in terms of product customization, which needs to be sensitively controlled and presented to avoid mass confusion, the external complexity resulting from excess variety (Piller et al. 2003). These findings often fall into one of two areas for investigation: (i) issues surrounding the contents of the solution space and (ii) communication and application of the contents through an appropriate product configurator. Kaplan et al. (2007) state that the customization process cannot be separated from the customized product, yet there are a limited number of empirical studies undertaken that investigate the relationship between MC, product configurators and customer interactions (Franke and Piller 2003). The literature reports customer satisfaction and engagement to be dependent upon a range of factors; amongst others, Huffman and Kahn (1998) identify the nature and operation of a product configurator, the presentation of data, and type of customer input required during the purchasing process, whilst Fiore et al. (2002) describe the enhancement of individuality, and the engagement in an exciting experience. Little research exists which draws together these issues to help illuminate the wider considerations surrounding both the customer co-design experience from co-design to receipt of product, and the now "fuzzy" practice of designing for co-design as a product designer, whose final instantiation of a product is at best only predictable within a given solution space.

The Need for a Coherent Model: The Emotional Connection If the solution space describes and defines the stable processes and product architectures of a MC product (Tseng and Piller 2003), and the entire co-design experience is an intrinsic element of a MC product, then there is a need to develop a conceptual model for MC product offerings that encompasses this wider context

184


within which the solution space resides, the entire co-design and product purchasing experience. As markets have become segmented, product development has begun to move beyond traditional considerations of usability and functionality. Where customers had been viewed in the past by traditional marketers as rational decision makers, and customer decision making as a simple problem solving process (Schmitt 1999), "...it has been suggested that these factors no longer drive consumer choice. Consumers now look for more from the products that they buy; they are looking for pleasure and the fulfilment of their emotional needs" (Porter et al. 2005). Desmet and Hekkert (2007) describe product experience as "a multifaceted phenomenon" (p.59) comprising aesthetic pleasure, attribution of meaning and emotional response. It is easy to assume that increased product performance heightens levels of customer satisfaction, but the relationship is far more complex than this (Shen et al. 2000). Trends indicate that users are expecting increasing levels of "connection" with everyday products (Demirbilek and Sener 2003), customers want products that "dazzle their senses, touch their hearts and stimulate their minds" (Schmitt 1999); "…emotions are often not merely elicited by the product 'as such'. When a product elicits an emotion, it is always induced by a specific product-subject relationship and within a specific context" (Desmet 1999). These relationships, contexts and experiences are intrinsic to the co-design experience and must therefore form an essential part of the product envelope model. These new user-centred approaches have been referred to amongst others as: "emotional ergonomics" (Bennett 2003), "the new human factors" (Jordan 2000), and "supra-functional needs" (Weightman and McDonagh 2003). Pine and Gilmore (1999) describe this as the emergence of the "experience economy", a new economic era in which successful companies must create memorable events or experiences which engage their customers in a personal way. Fiore et al. (2002) discuss the role that co-design may play in the desire for experiences, and highlight the active engagement of the customer as central to the co-design experience. The MC literature provides evidence of established methods from the fields of product design and engineering being utilized in designing MC products. Take as an example product family architecture, design axiom principles, and portfolio architecture (Krishnapillai and Zeid 2003). New methods include Toolkits for User Innovation Design (von Hippel and Katz 2002). However, as Hernandez, Allen and Mistree (2003) describe, to date, no method has been devised to aid designers in the successful application of these approaches a coherent and systematised manner for mass customization. As of yet, few direct connections


185

appear to have been made between the new user-centred approaches and the literature in the field of MC. As sociologist John Clammer observes "Shopping is not merely the acquisition of things, it is the buying of identity" (Lewis and Bridger 2004, p.13). In fact, it is a broadly held view that our possessions are both a contributor to, and reflection of, our own identities; making things a part of our "self" by creating or altering them appears to be a universal human belief (Belk 2001). Mass customization keys into this desire for creation as it is an approach that is fundamentally driven by an individual customers' emotional connection with the product, exemplified by their participation and engagement in the co-design experience. "Experience is used to sell…but…experience is as much a product of what the user brings to the situation as it is about the artefacts that participate in the experience. What this position implies is that we cannot design an experience. But with a sensitive and skilled way of understanding our users we can design for experience" (Wright et al. 2003, p.52). To design for a co-design experience we therefore need to understand the customer, their behavior, expectations and the interface between the customer and the configurator. Understanding the Co-Design Experience To design a co-design experience for MC, there is a need to define the term. We posit that the MC "co-design experience" consists not only of the specific codesign activities through the product configurator during product creation, but of the entire purchasing experience from the beginning of co-design activity through to the receipt of the customized product and beyond. This definition appears imperative in a situation where, unlike purchasing an off the shelf product, there will frequently be a delay between the co-design activity and receipt of final product. A survey of the literature within the field of MC highlights a breadth of research utilizing a wide range of research methods; this variety stems from both the nature of the research problems and the discipline of the researchers. There appears to be little enquiry investigating the wider aspects of the customer co-design experience as defined in this chapter. MC research can generally be seen to focus within a specific stage of the co-design experience. Some researchers utilize a methodology which attempts to "recreate" an MC purchasing environment, for example work by Kumiawan et al. (2003), and Kamali and Locker (2002), or else take an empirical approach to understanding the motivations and choices of a consumer as they are asked to go through a pre-selected MC purchasing process, for example work by Huffman and Kahn (1998) and Bee and Khalid (2003). Much of the

186


research concludes with a completed design at the product configurator. Other researchers use quantitative methods to investigate customer perception and understanding of MC without any participation in a co-design experience, for example Fiore et al. (2004) and Dellaert and Dabholkar (2007). Each piece of research within the MC literature is contributing to the knowledge of this relatively new paradigm, and their research problems and chosen research methods each offer useful insights into specific elements of the co-design experience. However, if, as Fiore et al. (2004, p.845) describe, "active engagement of the customer is central to the experience of co-design", it is important to select appropriate research methods to capture the entirety of the co-design experience for the customer; the tracking of customer behavior on websites, for example, could be seen as simply "capturing data" rather than the rich information source of customer observation, such as can be found in empathic design techniques in product design (Leonard and Rayport 1997). As the "rationally acting user has been transformed into a complex, emotional experiencer" (Mattelmäki 2006, p.20) it is important to recognise that MC research must not remain formulaic, turning customer experience into mathematical equations. As Rosenthal and Capper (2006) describe, formal market research techniques will often fail to detect opportunities for product innovation. Alternative approaches are needed to elicit the subtle, tacit customer needs, requiring product designers to "go beyond the view of a product as a set of explicit performance features and functions and to consider the implications of the physical and emotional context of product use" (p.216). The selection of appropriate research methods is paramount in deriving insightful data relating to the co-design experience. "We are often testing the wrong things: products instead of behavior, acceptability instead of experience" (Gobé 2007, p.223). Adjoining disciplines such as marketing, in particular areas such as Customer Relationship Management (CRM) and Customer Experience Management (CEM), as well as in product design areas relating to user-centred design, can all offer theoretical and practical techniques to support existing MC research. It is important to note that "design research is paradoxical: it is both imaginative and empirical" (Johnson 2003). Imaginative since "the customers' ability to guide the development of new products and services is limited by their experience and their ability to imagine and describe possible innovations" (Leonard and Rayport 1997) yet grounded in evidence since no business wants to feel their research has been "made up" by the research department (Johnson 2003). Research techniques are needed which enable us to empathise with the co-designer; "we need not only a window into the user’s life, but also an explanation of how he sees things in that window" (Mattelmäki 2003, p.121). Research methods such as contextual


187

observations and user interviews have become standard practice within the discipline of product design. "Traditional market research asks people questions about what they do and the design is based on what they say. But there is a difference between what they say they do and what they really do" (Myerson 2001). New approaches to customer co-design research in MC may reveal elements of customer experience which have not yet been identified, thereby offering new opportunities for design. "Design probes" are one research method which may be useful in supporting further research in the field of MC. Defined by Mattelmäki (2003, p.120) as "self documentation packages for gathering data on people’s actions and the contexts in which they take place…[they] provide people with tools for reflecting and projecting their opinions and feelings", and offer a research method for exploring the co-design experience. Their advantage lies in the "rich, textured understanding of user need" (Gilmore 2002, p.31), recognising that "the truth is that there is no average person out there" (p.32). Carrying out research with small numbers of users using these methods falls inline with the nature of MC where everyone is an individual; "Empathic research methods…if skilfully used, can yield much inspiration from small numbers of subjects" (Moggridge 2007, p.434). Travelling Through the Customer Corridor: Mapping the Touch Points One method of understanding what occurs in a co-design experience is to break the experience down into the "touch points" which construct it, the "instances of direct contact either with the product or service itself, or with representations of it by the company or third party" (Meyer and Schwager 2007). As Slassi (2005) describes, it is tangible touch points which make the experience real, enabling brand to be savoured, remembered and communicated. Touch points must be considered in their totality in order to create a "clear and consistent unified customer experience" (Moggridge 2007, p.422), providing a consistent narrative which runs throughout the experience. This narrative is important as brands and products need to weave a story around them which not only has emotional appeal but also communicates an authentic message – Caterpillar sells shoes on the back of its rugged work image, while customers of the Body Shop buy its beliefs along with its products. (Lewis and Bridger 2004). The role of brand is important to the co-design experience, and the concept of "Brand DNA" suits this theory building exercise well; the term has been used to cover a variety of perspectives such as specific words with positive associations to a brand (Marketing Research 2006), and the aesthetic form elements that contribute to a brand identity (Smyth and Wallace 2000), but is used here to infer

188


that brand can be embedded in all aspects of a product offering. Every brand consists of an essence consisting not only of the product or service, but the mental constructs associated with it. "Effective branding is about the emotions design creates" (Gobé 2007, p.xxxi) This essence, or construct, evokes a range of meanings which are powerful due to their social and cultural relevance (Bedbury 2004). As Lewis and Bridger (2004) describe, it is a brand’s ability to trigger emotional responses that will often provide it with a wining edge over less familiar products and services; in fact brand can apparently transcend logic when making purchasing decisions "As AOL executive Bob Pittman remarked: I remind people all the time that Coca-Cola does not win the taste test. Microsoft is not the best operating system. Brands win" (p.33). It is important for a mass customizer to consider the touch points within their product offering as these exist not only within the product itself, but throughout the entire co-design experience. "When a person buys a service, he purchases a set of intangible activities carried out on his behalf. But when he buys an experience, he pays to spend time enjoying a series of memorable events that a company stages – as in a theatrical play – to engage him in a personal way" (Pine and Gilmore 1999). At each touch point, "the gap between customer expectations and experience spells the difference between customer delight and something less" (Meyer and Schwager 2007, p.120). Schmitt (1999) describes these as touch points as "experience providers", Gilmore and Pine (1999) refer to them as "cues", highlighting the importance of each cue portraying a consistent theme to the customer to construct the desired experience. Service design refers to the "customer journey"; "the framework of a customer journey helps you think about the experiences and touchpoints that exist before and after the most obvious points of a service" (Moggridge 2007, p.435). Meyer and Schwager (2007) use the marketing term "customer corridor", the series of touch points that construct the purchasing process. Reichheld (2001) sees the customer corridor as the entire lifecycle of a customer’s interactions with a company and its products; key activities become doorways within the corridor, "It’s a good model for study of customer behavior, because what determines value is the sum of relative benefits and drawbacks, advantages and disadvantages, that consumers experience at each doorway along the corridor" (p.201). Figure 1 describes a generic customer journey. When we compare the generic customer journey in Figure 1 to a MC co-design experience, it becomes apparent that it is missing the co-design experience at the product configurator. Although there will be some variation between MC product offerings, a generic "MC customer journey" is offered in Figure 2. This "MC customer journey" describes the "stages" (key activities) that exist within a MC co-design experience.


Figure 1: The customer journey (Watkins 2007).

Figure 2: The MC customer journey: stages of a MC co-design experience.

189

190


Since making the customer co-design experience visible is the first stage in understanding a co-design experience, the "MC customer journey" offers a useful framework for developing the notion of a "customer corridor" within this research. We posit that the "customer corridor" represents a conceptual space through which the customer co-designer travels when purchasing a MC product. The corridor is made up of a series of "stages" (framed by the MC customer journey) and the touch points which construct the experience. Visualising a "customer corridor" enables mass customizers to understand where design effort is required to create the desired co-design experience.

Figure 3: Generic MC customer corridor (online purchase).

It is important to recognise that a co-design experience is not always a linear route. A useful metaphor is that used by Service Design company LiveWork (Moggridge 2007, p.422) who describe on-ramps and off-ramps in a customer experience "…so you're not talking about the main road of content flowing through, but how people access it, how they leave, what they do with it when they're finished with it". This appears particularly relevant to many existing


191

product configurators where, for example, designs can be saved and returned to at a later date. We can consider these on-ramps and off-ramps as "doorways" within the "customer corridor" where the co-designer can enter or exit the co-design experience, for example saving a design or logging into your account. Based upon the purchasing experience of a range of MC products, Figure 3 describes a generic MC customer corridor, indicating the stages, doorways and touch points within the MC product offering. When looking at a generic customer corridor, we see a single co-designers interaction with their purchasing experience. When we consider the doorway into resources such an online communities within the website, it becomes evident that the creation of these resources (accounts/blogs/groups etc where co-designers can share, talk and discuss) offers the potential for bringing customer corridors together (Figure 4).

Figure 4: The creation of communities.

Whilst mapping of customer experience is not new in business, (for example see Figure 1, Tseng et al. 1999), the application of these techniques to MC, in

192


combination with user-centred approaches remains unexplored, and offers potential for insights into previously undiscovered areas of customer co-design experience. What the customer corridor does not offer however, is information on how to produce a better co-design experience. The Product Envelope Model: Designing for "Pleasure" The product envelope model seeks to develop the existing notion of the solution space into a more complete model which reflects the wider context within which the solution space resides. In light of the need to design a positive co-design experience, we have highlighted user-centred approaches as offering opportunities for understanding and developing this area. Jordan’s work on "designing pleasurable products" (2000) was one of the first to explicitly address the creation of positive feelings in product use, explaining users' emotional response to products through an framework of four types of pleasure. A number of other tools and techniques can be found in the literature, these include Norman’s (2004) work on "emotional design"; the Kano Model of Product Quality (Matzler et al. 1996); the framework of product experience (Desmet and Hekkert 2007) and Kansei Engineering, best defined as "the implementation of the customer’s feeling and demands into product function and design" (Nagamachi 2002, p.289). Jordan’s (2000) work is different, concerning itself not only with what a product should do and how it should do it, but with who the product is for and how will it provide benefits for them as individuals, members of social groups and members of society. In developing a conceptual model of an MC product offering which we call the product envelope (first described in Bardill et al. 2007), Jordan’s four pleasures framework (2000) is particularly useful as a hypothesis generating tool since it provides not a theory of pleasure, but rather a framework to help those involved in the design process take a structured approach to understanding the entire spectrum of benefits a product can bring. A four pleasures analysis of MC reveals the following points: Physio-pleasure Is concerned with the physical body, everything from anthropometrics and ergonomics, physiological need or benefit, through to positive feedback from the sensory organs; touch, taste, smell, hearing and sensual pleasure. In the context of product design Jordan (2000) discusses tactile and olfactory product properties; how does a product feel when holding and touching it, how does the interior of a new car smell? Physio-pleasure can be experienced both from the product and as a result of its function.


193

Physio-pleasure in MC relates to the physical aspects of a product, including product, packaging, configurator etc., principally based on material properties and their subsequent form. The nature of "evidencing" a MC design becomes a key consideration; evidencing relates to the type of feedback provided to the codesigner to communicate a product’s physical attributes. If a co-designer’s sensory perceptions are limited, for example through the use of an online product configurator, how can physio-pleasure best be catered for? This is particularly important in a situation where the product is yet to exist. The presence of "uncertainty" in the buying process infers that understanding the brand DNA, and its relevance to the physical product will be of prime importance; for example if customizing an iPod online with Apple, assumptions could be made on behalf of the consumer regarding material quality, type and finish according to the brand DNA. Physio-pleasure may already exist through any one-to-one experience with the product, and the customer corridor would only contain options which furthered this relationship. The customer would not want to "design out" the iconic essence of the iPod, but rather enhance it through its perceived appropriateness to the individual’s personal pleasure construct. Likewise it can be assumed that Apple would not want to spoil the brand image through color and material choices which do not match with their brand DNA. This can clearly be seen their recent offer to provide free laser engraving on new iPods purchased online (Apple Store 2008). The placement of text, choice of font and point size is carefully controlled to balance the customization against strict design requirements in line with their brand DNA. Brand DNA is an intrinsic part of physio-pleasure in terms of product semantics (both in retaining brand integrity, and providing coherence and recognition in customer perception). Socio-pleasure Refers to relationships with others – individuals, groups and society as a whole. Socio-pleasure is drawn from the aspects of products that confer social, material or cultural status, help to construct personal identity and/or stimulate desirable social interaction. These product qualities give positive feedback to the owners about their personal view of themselves in society. Socio-pleasure can also be derived from changes in technology; for example the creation of cyberspace made everything "local", there is socio-pleasure gain when individuals and groups became able to both express and retrieve information in a manner not previously possible. Socio-pleasure in MC relates to the relationship between a producer, their product, and one customer, who will draw socio-pleasure from elements of design relating to the brand and the facilitation of positive social interaction that the co-design

194


and ownership of the product confer and stimulate. An important consideration is identity construction through product consumption relating to particular brands; "the perfume does not just smell nice, it signifies or promises sexuality, femininity, prestige, getting a partner" (Slater 1997). After purchasing a customized bag from Timbuk2 (2008), the company responds with an email stating "You've now joined a community of individuals who value quality construction, self-expression and urban style" (personal email from Timbuk2). Socio-pleasure is also gained from perceived value derived from uniqueness of output and pride of authorship, reporting on their co-design experience and subsequent recognition of their creative input. Companies such as Nike iD (2008) and my K-Swiss (2008) incorporate these values within their product envelopes; enabling the consumer to "sign" each product with a self-selected word/phrase enhances the emotional connection between consumer and product. Pumas Mongolian Shoe BBQ (2008) has an online user gallery in which customized designs can be posted and ranked, purchased, edited and shared. Timbuk2 has a link to the photo management and sharing site Flickr (2008), showing public photos tagged with Timbuk2, thereby facilitating a visual community of users. Positive co-design experiences can stimulate a desire for not only the product but also for the wider co-design and purchasing experience. Socio-pleasure is also concerned with the construction of a narrative running through the co-design experience, offering a coherent experience for the co-designer. Psycho-pleasure Refers to a user’s cognitive interaction with a product and their subsequent emotional reaction. Outcomes are observed to give more emotionally satisfying results (psycho-pleasure) when products enable users to complete complex tasks with little cognitive demand. Psycho-pleasure can also be drawn from perceived product benefits; product sounds which convey quality, product forms which appear streamlined or reflect the notion of power. The vacuum cleaner that sounds powerful and whose bulging form appears to struggle to contain the large powerful motor inside offers psycho-pleasure to those looking to complete the domestic chore faster and more conveniently (Jordan 2000). Psycho-pleasure is drawn from the co-designers cognitive interaction and their subsequent emotional reaction with the tangible and intangible elements of both the product configurator, purchasing process and resulting product. As the codesign experience becomes an important part of socio pleasure in embedding desire for the MC product within a social group, the quality of the co-design experience is mediated, in part, by psycho-pleasure. Product configurators that enable complex tasks to be undertaken for small levels of cognitive overhead,


195

explicitly supporting customers in specifying their needs and making informed choices, will provide positive co-design experiences (Piller et al. 2003). The literature in fields of interaction design, experience design, Human Computer Interaction (HCI), ergonomics and many other associated areas is burgeoning with guidance on how to effectively configure the configurators; yet it is important to recognise that the provision of psycho-pleasure suffers similar tensions to physiopleasure, in that the customer is interacting with a virtual product. Hence, the codesign experience needs to address not only how the product and brand DNA is evidenced, but how feedback relating to design decisions are communicated to the co-designer. Psycho-pleasure is also important in the period between the codesign activity and receipt of the MC product. Jordan (2000) describes the phenomenon of cognitive dissonance (p.44), "the search for evidence that confirms what a person wishes to believe". After the purchase of a product, in particular an expensive product, people may wish to be reassured they have made the correct purchasing decision. This appears particularly relevant to MC, where there will frequently be a delay between purchase and receipt of the product. As an example of a positive psycho-pleasure experience after co-design and payment, co-designers purchasing customized products from Freitag (2008) are rewarded with email updates of their order; "It will take some time to wash, manufacture and package (about 2–3 weeks). You'll be notified by email after each major step of production. Just give us some time to provide you with our best quality"; "We have cut out your bag. It will now be cleaned and cleaned and cleaned. And once that is done, we will send you another email. Woohaa!" (personal emails from Freitag) Ideo-pleasure Relates to peoples' values. In the context of products this ranges from aesthetics to ethics. It includes taste, moral values and personal aspirations. It defines how people do, and would like to, see themselves; for example owning a HarleyDavidson reaffirms a perceived identity of a rider who is a rebel, reinforcing a "more exciting, less conformist self-image" (Jordan 2000, p.53). Ideo-pleasure can be drawn from the values that a product embodies; is it fair-trade, organic, sustainable, engineered, crafted or "designer"; does it reflect the values of a particular era or culture? Ideo-pleasure can also be drawn from a positive relationship between the product as "art form" and its aesthetic effect on the user’s environment. Ideo-pleasure in MC is a combination of judgements and values, relating to the product and brand; these are difficult to define since they are personal and constructed from a complex mix of objective and subjective components. There is

196


an evident link between socio and ideo-pleasure in the construction of the self; socio-pleasure is driven by the social interaction and positioning of an individual within social groupings whereas ideo-pleasure is driven by the potential ownership of, and experience gained through the MC product, enhancing the individuals' construction of themselves. Perceptions of brand can therefore be seen to play an important role in ideo-pleasure, both throughout the product and the co-design experience; what does the MC product say about the co-designer? The NikeID "London Studio" (2008) offers a team of "design consultants" to help you design your NikeID trainer. What does this communicate about the brand? You are no longer entering a shop, but rather a "studio". Therefore, in terms of ideo-pleasure we can begin to consider not only what the brand and your design says about you, but what your choice of "design consultant" portrays. What must be established are the commonalities that exist between an individuals ideo-pleasure constructs, and whether these are permanent, trend-based, transitional or related to a phase in the formation of personal psychology. It is apparent that the investigation of ideopleasure will benefit from user-centred approaches where research can focus on the individual. The first generations of the product envelope model (Bardill and Herd 2006, Bardill et al. 2007) informed by the four pleasures analysis of MC, build upon existing knowledge within the MC literature. The product envelope (Figure 5) consists of a core element, the solution space (as currently defined in the literature), surrounded by levels of service, experience and interaction. These levels are mapped against the four pleasures to construct an initial representation of the customer co-design experience. The model contains a number of key ideas which are summarized below.

The product envelope is generated by the producer of the MC product (where producer refers to the wider role of the organization selling the product and can encompass the designer, manufacturer, marketing team etc)

As a customer co-designer, you penetrate the envelope and engage with a number of experiential layers before reaching the solution space where the MC product resides; these layers are interconnected and the co-design experience will not necessarily provide a linear route through the envelope

When creating a product envelope, regular traversal of the design line is required to ensure all regions of the envelope are integrated; this is important in ensuring coherence in the customers perception of the entire MC product offering

Brand is important in differentiating between product envelopes, in circumscribing the envelope and in permeating through the core of the envelope.


197

This is an essential consideration, as brands generate choice, simplify purchase decisions, offer quality assurance, and reduce risks involved in purchase (Karjalainen 2003)

The four pleasures framework (Jordan 2000) is a useful tool as it can be used to stimulate discussion relating to both the product, and co-designer and codesign experience

Figure 5: The first generation of the product envelope model (Bardill et al. 2007).

Whilst remaining a useful model, we have found the representation of levels of experience restrictive since these suggest a set number of sequential or hierarchical stages within a co-design experience. The mapping of the four pleasures works at a basic level but indicates separation rather than connection; the co-design experience needs to be placed within the larger context of the product envelope. In light of the discussions of the four pleasures analysis of MC, it becomes evident

198


that the four pleasures framework can be used as a key method of defining and informing the customer co-design experience, and that there are overlaps between the experience of these pleasures in relation to both company practises and individual co-design experiences. A successful MC co-design experience will be designed to stimulate all four pleasures (Figure 6).

Figure 6: Defining customer co-design experience using the four pleasures framework.

We therefore propose a revised product envelope model (Figure 7) that builds upon previous work, to further unpack the customer co-design experience to consider that:

The product envelope is generated by the producer of the MC product; as a customer co-designer, you can penetrate the envelope from any direction (or number of directions) depending on your motivation for the co-design experience; for example are you engaging in MC because you love the physical product (physio-pleasure), or do buy into the brand values and want to be part of the social group wearing those customized trainers (socio- and ideopleasure) (Figure 8)


199

The design of the product envelope (the design line), and resulting co-design experience is characterized by all four pleasures being designed for within all elements of the MC customer co-design experience, positioned within an awareness of the wider pleasure constructs. When creating a product envelope, regular traversal of the design line is required to ensure all regions of the envelope are integrated; this is important in ensuring coherence in the customers perception of the entire MC product offering

Brand remains important in differentiating between product envelopes, in circumscribing the envelope and in permeating through the core of the envelope. Brand is constructed from a blend of all four pleasures.

Figure 7: The product envelope model.

Constructing a Coherent Experience: The Experience Matrix As described in Figure 3, an MC "customer corridor" represents a conceptual space through which the customer co-designer travels when purchasing a MC product. The customer corridor indicates the "stages" of their journey; the "doorways" through which the co-designer can enter and exit the co-design

200


experience; and the touch points which represent their interaction with the company. If touch points are seen as the connectors which hold elements of the experience together, then these connectors, if designed correctly, can be used as a framework for designing a coherent experience. Each touch point within the framework must now be designed both as an individual instance, and as part of the overall product narrative.

Figure 8: Differing customer entry points into the product envelope model.

We can use the product envelope model to help unpack the customer corridor, and the four pleasures framework to hypothesise user interaction. As the product envelope discussion reveals, the customer co-design experience is revealed not only by the company offering, but what the customer co-designers themselves bring to the experience. Empathic research offers a means of understanding the importance of various touch points, and offers the potential for uncovering latent need, latent touch points, and previously undiscovered or desired co-design experiences; for example how do people use the configurators?; what interactions take place between users and how can a company mediate this behavior to their advantage? As Kelley and Littman (2001) describe "designing new experiences is usually about figuring a way to connect with people". Without input from codesigners it is naive to believe we can understand their co-design experience, but equally it is important to recognise that the co-designer cannot tell us what to design for them. Bill Moggridge of international design consultancy IDEO tells us that "the only way to experience an experience is to experience it" (Suri 2004).


201

User research must be considered as inspirational rather than explicit, and it must be remembered that each customer is unique. The experience matrix offers a systematic way of supporting and informing design decisions when designing a MC customer experience. This tool does not seek to supersede nor eliminate existing design and engineering approaches, but rather seeks to pull the existing disparate threads together to form a complete, coherent tapestry. Each touch point (established from the customer corridor and relevant empathic research approaches) can be placed within its overall narrative, and can be considered in relation to the four pleasures framework. By prioritising the pleasures for each given touch point, this knowledge will enable appropriate design processes to be selected and used for designing each element within an MC customer co-design experience. Figure 9 demonstrates an experience matrix for the generic customer corridor described in Figure 3.

Figure 9: The experience matrix.

The experience framework offers potential both for mapping existing co-design experiences, developing new or improved co-design experiences, and creating guidelines for good practice within industry sectors.

202


Discussion and Conclusions In seeking to develop knowledge and good practice in the generation of MC products and services, this chapter highlights and defines the importance of the customer co-design experience as an integral part of any MC product offering. In designing and/or reflecting upon an MC co-design experience, we believe that the customer corridor offers a valuable tool in mapping both the doorways and the touch points which frame the narrative and experience for the customer codesigner. Further work in the field is needed in establishing and understanding customer corridors; unpacking both tangible and intangible, explicit and latent touch points that construct a co-design experience for a customer. It is these subtle, currently undefined opportunities which may offer real design advantage; empathic research methods offer valuable opportunities in supporting this work. These investigations in to user-centred research approaches appear crucial in enabling co-design to reach its full potential. As an area currently underresearched, we believe that many developments in the co-design experience can be gained from the use of literature and practice in many adjoining disciplines. The product envelope model offers a wider picture, a means of reflecting upon the customer corridor through the use of a four pleasures analysis. This hypothesis generating tool assists in establishing areas where design effort is needed for a particular, company, product or MC experience. When used in conjunction with the product envelope model and customer corridor, the experience matrix then offers a systematic approach to mapping and understanding the requirements for the specific touch points which construct that customer co-design experience. The tools presented here can be used to inform and support an existing design process, enabling appropriate design methods to be selected and utilized; the choice of design method is crucial in generating the desired experience. Further work is required in translating these approaches into industry and our research is currently refining these models, exploring both the mapping of MC customer corridors, and how empathic research methods can be used to effectively identify both explicit and latent touch points within MC co-design experiences. References Apple Store (2008) [Internet]. Available from: <store.apple.com> [Accessed 19 March 2008]. Bardill, A., Herd, K. & Karamanoglu, M. (2007). Product Envelopes: Designing Positive Interplay between Brand DNA and Customer Co-Designers. International Journal of Mass Customization. 2(1/2): 57–75.


203

Bardill, A. & Herd, K. (2006). Maintaining Positive Interplay between Brand DNA and Customer CoDesigners in Mass Customized Products. International Conference on Strategic Innovation and Creativity in Brand & Design Management. 29 November. Seoul, Korea. Bee, O. & Khalid, H. (2003). Usability of design by customer websites. In Tseng, M & Piller, F. (eds). The Customer Centric Enterprise. Advances in Mass Customization and Personalization. Berlin: Springer. Belk, R. (1988). Possessions and the Extended Self. Journal of Consumer Research. 15(2): 139–168. Bedbury, S. (2002). A Brand New World. Harmondsworth: Penguin. Bee, O. & Khalid, H., (2003) Usability of Design by Customer Websites. In: Tseng, M & Piller, F. (eds). (2003). The Customer Centric Enterprise. Advances in Mass Customization and Personalization. Berlin: Springer. Bennett, O. (2003). Set in Emotion, Design Week. 18(50): 22–23. Berger, C., Möslein, K., Piller, F. & Reichwald, R. (2005) Co-Designing Modes of Cooperation at the Customer Interface: Learning from Exploratory Research. European Management Review. 2: 70–87. Berger, C. & Piller, F. (2003). Customers as Co-Designers. IEE Manufacturing Engineer. 82(4): 42–46. Dellaert, B. & Dabholkar, P (2007). Using complementary services to support online mass customization. The 2007 World Conference on Mass Customization and Personalization. Boston, USA. 7–10 October. Demirbilek, O, & Sener, B. (2003). Product Design, Semantics and Emotional Response. Ergonomics. 46(13/14): 1346–1360. Desmet, P. (1999) To Love and Not to Love: Why do Products Elicit Mixed Emotions? In: Overbeeke, C & Hekkert, P. (eds.) Proceedings of the First International Conference on Design & Emotion. 3–5 November 1999. Delft: Delft University of Technology. 67–73. Desmet, P. & Hekkert, P. (2007). Framework of Product Experience. International Journal of Design. 1(1): 57–66. Fiore, A., Lee, A. and Kunz, G. (2004). Individual differences, motivations, and willingness to use a mass customization option for fashion products. European Journal of Marketing. 38(7): 835–849. Flickr (2008) [Internet] Available from: <www.flickr.com> [Accessed 19 March 2008]. Franke, N. & Piller, F. (2003). Key Research Issues in User Interaction with Configuration Toolkits in a Mass Customization System. International Journal of Technology Management. 26(5/6): 578–599. Freitag (2008) [Internet] Available from: <www.freitag.ch> [Accessed 19 March 2008]. Gamble, P., Stone, M., Woodcock, N. & Foss, B. (2006). Up Close and Personal? Customer Relationship Marketing @ Work. 3rd ed. London: Kogan Page Ltd. Giebelhausen, M. (2007). Custom Kicks: Using Metaphor Elicitation to Understand Consumers' Thoughts and Feelings Regarding Online Shoe Customization. The 2007 World Congress on Mass Customization and Personalization. 7–10th October. Boston, USA. Gilmore, D. (2002). Understanding and Overcoming Resistance to Ethnographic Design Research. Interactions. 9(3): 29–35. Gobé, M. (2007). Brandjam. Humanising brands through emotional design. New York: Allworth Press. Hernandez, G., Allen, J. & Mistree, F. (2003) A Theory and Method for Combining Multiple Approaches for Product Customization. 2nd Interdisciplinary World Congress on Mass Customization and Personalization. 6–8 October 2003. Munich, Germany. Huffman, C. & Kahn, B. (1998). Variety for Sale: Mass Customization or Mass Confusion. Journal of Retailing. 74(4): 491–593.

204


Johnson, B. (2003). The Paradox of Design Research. The Role of Informance. In: Laurel, B. (ed) Design Research. Methods and Perspectives. Cambridge: The MIT Press. Jordan, P. (2000). Designing Pleasurable Products. An Introduction to the New Human Factors. London: Taylor & Francis. Kamali, N. & Locker, S. (2002). Mass Customization: On-line Consumer Involvement in Product Design. Journal of Computer-Mediated Communication. 7 (4). Available from <jcmc.indiana.edu/vol7/issue4/loker.html> [Accessed 14 February 2008]. Kaplan, A., Schoder, D. & Haenlein, M. (2007). Factors Influencing the Adoption of Mass Customization: The Impact of Base Category Consumption Frequency and Need Satisfaction. The Journal of Product Innovation Management. 24: 101–116. Karjalainen, T. (2003). Semantic Knowledge in the Creation of Brand-Specific Product Design. [online] 5th European Academy of Design Conference. 28–30 April. Barcelona, Spain. Available from: <www.ub.es/5ead/PDF/14/Karjalainen.pdf> [Accessed 19 May 2005]. Kelley, T. & Littman, J. (2001). The Art of Innovation. Lessons in Creatvity from IDEO, America’s Leading Design Firm. London: HarperCollinsBusiness. Krishnapillai, R. & Zeid, A. (2003) Adaptive Design Customization for Mass Customization. A Framework. 2nd Interdisciplinary World Congress on Mass Customization and Personalization. 6–8 October. Munich, Germany. Kumiawan, S., Tseng, M. & So, R. (2003). Consumer Decision-Making Process in Mass Customization. 2nd Interdisciplinary World Congress on Mass Customization and Personalization. 6–8 October. Munich, Germany. Lewis, D. & Bridger, D. (2004). The Soul of the New Consumer. London: Nicholas Brealey Publishing. Leonard, D. & Rayport, J. (1997). Spark Innovation through Empathic Design. Harvard Business Review. 75(6): 102–113. Mattelmäki, T. (2003) Probes: Studying Experiences for Design Empathy. In: Koskin, I., Battarbee, K. & Mattelmäki, T. Empathic Design. User Experience in Product Design. Finland: IT Press. Mattelmäki, T. (2006). Design Probes [Internet]. Finland: Gummerus Printing. Available from: [Accessed 19 March 2008]. Matzler, K., Hinterhuber, H., Bailom, F., & Sauerwain, E. (1996). How to Delight Your Customers. Journal of Product and Brand Management. 5(2): 6–16. Meyer, C. & Schwager, A. (2007). Understanding Customer Experience. Harvard Business Review. 82(2): 116–126. Moggridge, B. (2007). Designing Interactions. Massachusetts: The MIT Press. Myerson, J. (2001). IDEO. Masters of Innovation. London: Laurence King Publishing. My K-Swiss (2008) [Internet] Available from: <www.mykswiss.com> [Accessed 19 March 2008]. Nagamachi, M. (2002) Kansei Engineering as a Powerful Consumer-Oriented Technology for Product Development. Applied Ergonomics. 33(3): 289–294. Nike iD (2008) [Internet] Available from: [Accessed 19 March 2008] Nike iD studio (2008) [Internet] Available from: [Accessed 19 March 2008]. Norman, D. (2004). Emotional Design. New York: Basic Books. Papanek, V. (1997). Design for the Real World. 2nd ed. London: Thames and Hudson Ltd.


205

Piller, F., Koch, M., Möslein, K. & Schubert, S. (2003). Managing High Variety: How to Overcome the Mass Confusion Phenomenon of Customer Co-design. EURAM. 3–5 April. Milan. Piller, F., Schubert, P., Koch, M. & Möslein, K. (2004). From Mass Customization to Collaborative Customer Co-Design. Proceedings of the 12th European Conference on Information Systems. 6–8 June. Finland. Piller, F. & Müller, M. (2004). A New Marketing Approach to Mass Customization. International Journal of Computer Integrated Manufacturing. 17(7): 583–593. Pine, J. (1993). Mass Customization. The New Frontier in Business Competition. Boston: Harvard Business School Press. Pine, J. & Gilmore, J. (1999). The Experience Economy. Boston: Harvard Business School Press. Porter, S., Chhibber, S., Porter, M. & Healey L. (2005) RealPeople: Making Users' Pleasure Needs Accessible to Designers. Accessible Design in the Digital World Conference. 23–25 August. Scotland. Puma Mongolian Show BBQ (2008) [Internet]. Available from: mongolianshowbbq.puma.com [Accessed 19 March 2008]. Reichheld, F. (2001). The Loyalty Effect. The Hidden Value Behind Growth, Profits and Lasting Value. 2nd ed. Boston: Harvard Business School Press. Reichwald, R., Seifert, S., Walcher, D. & Piller, F. (2004). Customers as Part of Value Webs: Towards a Framework for Webbed Customer Innovation Tools. Proceedings of the 37th Annual Hawaii International Conference on System Sciences. 5–8 January. Hawaii. Rogoll, T. & Piller, F. (2004). Product Configuration from the Customer’s Perspective: A Comparison of Configuration Systems in the Apparel Industry. [online]. International Conference on Economic, Technical and Organizational aspects of Product Configuration Systems, June 28–29. Copenhagen. Available from preview.tinyurl.com/ljx7wj [Accessed 06 January 2006] Rosenthal, S. & Capper, M. (2006). Ethnographies in the Front End: Designing for Enhanced Customer Experiences. The Journal of Product Innovation Management. 23: 215–237. Schmitt, B. 1999. Experiential Marketing. New York: Free Press. Schreier, M. (2006). The value increment of mass-customized products: an empirical assessment. Journal of Consumer Behavior. 5: 317–327. Shen, X., Tan, K., & Xie, M. (2000) An Integrated Approach to Innovative Product Development using Kano’s Model and QFD. European Journal of Innovation Management. 3: 91–99. Slassi, K. (2005). The Sensory Experience. Brand Strategy. (193): 40–41. Slater, D. (1997). Consumer Culture and Modernity. Cambridge: Polity Press. Smyth, S. & Wallace, D. (2000). Towards the Synthesis of Aesthetic Product Form. Proceedings of the Design and Engineering Technical Conferences and Computers and Information in Engineering Conference. 10–13 Sept. Baltimore, Maryland. Suri, J. (2004). Design Expression and Human Experience: Evolving Design Practice. In: McDonagh, D., Hekkert, P., Van Erp, J. and Gyi, D. (eds) Design and Emotion. London: Taylor and Francis. Timbuk2 (2008) [Internet] Available from: <www.timbuk2.com> [Accessed 19 March 2008]. Tseng, M & Piller, F. (eds). (2003). The Customer Centric Enterprise. Advances in Mass Customization and Personalization. Berlin: Springer. Tseng, M., Qinhai, M. & Su, C. (1999). Mapping Customers' Service Experience for Operations Improvements. Business Process Management. 5(1): 50–64.

206


Von Hippel, E. (2001). Perspective: User toolkits for innovation. The Journal of Product Innovation Management. 18(4): 247–257. Von Hippel, E. & Katz, R. (2002). Shifting Innovation to Users via Toolkits. Management Science. 48(7): 821–833. Watkins, H. (2007). Drive Loyalty through Fantastic Customer Experiences. [Internet] August 2007. Available from: www.themarketingleaders.com/articles/aug07/huw_watkins.html. Weightman, D., & McDonagh, D. (2003). People are Doing it for Themselves. Proceedings of Designing Pleasurable Products and Interfaces 2003, DPPI'03. 23–26 June. Pittsburg, Pennsylvania. 34–39. Wright, P., McCarthy, J. & Meekison, L. (2003). Making Sense of Experience. In: Blythe, M., Monk, A., Overbeeke, K. & Wright, P. (eds). Funology: From Usability to Enjoyment. Netherlands: Kluwer Academic Publishers 43–53.

Author Biographies Kate Herd is a Research Student/Tutor within the Product Design and Engineering department at Middlesex University. Her background is in product design, supporting new product development within SMEs. Her time is divided between her PhD "The development of conceptual models for designing for co-design in mass customization", and teaching across undergraduate and postgraduate provision. Her research is investigating the customer co-design experience, developing models and tools to inform the emergent task of designing for customer co-design. Utilizing design probes, her PhD explores the entirety of the customer co-design experience from co-design to receipt of the product and beyond, working with empathic research methods and theories from the field of product design. Contact: www.creativeconversation.net | [email protected] Andy Bardill is a Principal Lecturer in Product Design and the Director of Product Design and Engineering Programs at Middlesex University. In his early career he worked in military avionics, principally with communications, radar, laser and camera reconnaissance systems. He became disaffected with working with things and transferred his energies to working with people; he has worked in design education for the past 18 years. He holds a PhD in virtual learning environments through the development of intelligent hypermedia courseware, supported by system intelligence derived from dynamic user models in task oriented, goal directed, design scenarios. His current work and research interests include: Interaction Design, affecting human behavior through environmental design, mass customization, and the "robot apprentice". Contact: www.creativeconversation.net | [email protected] Mehmet Karamanoglu is a Principal Lecturer in Product Design and the Head of Department of Product Design and Engineering within the School of Engineering and Information Sciences at Middlesex University. He is also a Visiting Professor and advisor on Higher Education Management for Cooperative Education at University of Cincinnati, Ohio, USA. He is the UK Skills Expert for Mechatronics competitions and currently coordinates the University’s EPSRC Collaborating Training Account for postgraduate


207

research training. Mehmet is a graduate of Mechanical Engineering with a PhD in numerical methods. His research interests include mass customization, mechatronics, design engineering, discrete event simulation, manufacturing automation and the use of robotics in the creative industries. Contact: www.creativeconversation.net | [email protected]

2.3

Why Consumers Are Willing to Pay for Mass Customized Products: Dissociating Product and Experiential Value Aurelie Merle Department of Marketing, Grenoble Ecole de Management, France Jean-Louis Chandon Department of Marketing, Université Paul Cézanne Aix-Marseille, France Elyette Roux Department of Marketing, Université Paul Cézanne Aix-Marseille, France

The aim of this paper is twofold. First, we propose to conceptualize the perceived value of mass customization into two components: 1) mass-customized product value and 2) mass customization experience. Second, we test an integrative framework bringing together value components and willingness to pay for mass-customized products. As opposed to previous research, the findings show an indirect effect of mass customization experience on consumer willingness to pay. Furthermore, no direct effect is found.

Introduction More and more brands offer mass-customized solutions for a wide range of products, such as sneakers (Converse, Puma, Nike...), clothing (Timberland, Quiksilver), cameras (Leica) or candies (M&M’s). Some authors distinguish two categories of definitions of mass customization (MC): visionary and practical (Da Silveira, Borenstein, and Fogliatto 2001; Kaplan and Haenlein 2006; Zipkin 2001). The visionary concept was first coined by Davis (1987) and represents "the ability to provide your customers with anything they want profitably, any time they want it, anywhere they want it, any way they want it" (Hart 1995). On the other hand, practical definitions consider MC as "building products to customer specifications using modular components to achieve economies of scale" (Duray, Ward, Milligan, and Berry 2000). Our vision of MC is in keeping with this second way of thinking. Nevertheless, this definition is not specific enough to thoroughly understand the concept from a marketing point of view. In this paper, we define mass customization as an offer that allows customers to: 1) participate in codesign by personally modifying several features of a product, from a predefined 208


209

set, and 2) buy the co-designed product. Thus, they actively become part of the product development process. Our definition does not include body-scanning technology since people do not modify the product themselves. We also restrict our analysis to a business-to-consumer setting. Researchers have come to recognize the perceived value of mass customization as the main key to success (Broekhuizen and Alsem 2002). However, they point to the lack of knowledge about this topic (Franke and Piller 2003, 2004; Hart 1995; Schreier 2006; Zipkin 2001): very few studies have tried to understand the value components of such an offer. Furthermore, one of the objectives of a mass customization strategy is to increase the willingness to pay (Bardakci and Whitelock 2004; Piller and Müller 2004). Numerous papers have investigated this concept. One element that is still missing is an integrative framework that brings together MC value components and willingness to pay for a mass-customized offer (Franke and Piller 2004; Shreier 2006). In other words: why are consumers willing to pay for mass customization? In this context, we develop and test an integrative framework bringing together perceived value and willingness to pay for mass-customized products. This chapter is partly based on Merle, Chandon and Roux (2008). Theoretical Background and Hypotheses The perceived value of mass customization Schreier (2006) theoretically identified several categories of benefits: functional benefits (i.e. a better perceived fit), perceived uniqueness, process benefit of self design (hedonic, experiential benefit), and pride of authorship. Broekhuizen and Alsem (2002) separated instrumental benefits (better fitting products) from hedonic benefits (enjoyable process). Based on these findings, we can argue that MC should be valued on two different dimensions: mass-customized product value and mass customization experience value. The first dimension is associated with the "anticipated consumption experience" (Arnould, Price, and Zinkhan 2002). The second is related to the interaction between the consumer and the product during the co-design stage. Harris, Harris, and Baron (2001) explained that the more a customer is engaged in service production, the more likely it is that he will perceive it as a positive experience. Because mass customization needs the consumer to be actively engaged in co-design, the process should be analyzed as an experience. More empirical support for these dimensions comes from the work of Fiore, Lee and Kunz (2004). They confirm that willingness to use co-design is positively

210


related to two motivations which are creating a unique product and trying codesign as an exciting experience. The literature on perceived value does not take into account the link between value components (Evrard and Aurier 1996; Evard, Aurier, and N'Goala 2004). In the same way, studies on mass customization value hypothesize the independence between these two value components (Franke and Schreier 2006a; Schreier 2006). However, Franke and Piller (2003) called more research on the relationships between these two components. Since a customized product is the direct result of the co-design process, the value of this experience should have an impact on product value, leading to the following hypothesis: H1. Mass customization experience value positively influences masscustomized product value. From mass customization value components to willingness to pay The research on consumer willingness to pay for mass-customized products is an issue of critical importance. Several studies have therefore tried to understand the willingness to pay a premium for such an offer compared with a conventional standard product. (Franke and Piller 2004; Franke and Schreier 2006a, 2006b; Schreier 2006; Schoder et al. 2006). They have reported contrasted results on the number of people who are willing to pay a premium for mass-customized products and on the value increment. According to Kamali and Locker (2002), people are not willing to pay more for a mass-customized t-shirt, whatever the level of customization. On the contrary, Franke and Piller (2004) discovered a 300% value increment for watches designed by users with the help of a MC toolkit. Schreier (2006) found that 88% of respondents are willing to pay more for a mass-customized t-shirt and other products: the value increment is 113% for t-shirts, 207% for cell phone covers and 106% for scarves. He argued that empirical research should bring together the value components and willingness to pay for such an offer. In this context, Franke and Schreier (2006a) examined for the first time the effect of two value sources on willingness to pay a premium and supported the influence of utilitarian value ("preference fit hypothesis") and of hedonic value on the price premium. Thus, one would expect that mass-customized product value and mass customization experience value have an influence on willingness to pay. Accordingly, the following hypotheses can be stated: H2. Mass-customized product value positively influences willingness to pay for mass customization. H3. Mass customization experience value positively influences willingness to pay for mass customization.


211

Figure 1 illustrates our theoretical model of mass customization value. As opposed to previous studies, we also expect a direct and indirect effect of mass customization experience value on willingness to pay.

Figure 1: Theoretical model of mass customization value.

Method Sample and data collection To test these hypotheses, 567 French students took part in a mass customization experience in a laboratory setting. The subjects (48.4% female) were 20 years old on average (SD: 1.94). They were asked to customize their favorite pair of shoes on the Nike Id program (www.nikeid.com), which was at that time, in our opinion, the most advanced mass customization program in a B-to-C context. Before the study, 17.6% of the respondents had visited the Nike Id website and 1.6% had already bought a Nike Id product. After the co-design phase, they filled out a questionnaire. To increase their involvement, a drawing made it possible for participants to win the pair of shoes they had customized. Measurement scale development Customized product value versus customization experience value In order to have a deeper understanding of the value components of the two identified dimensions, we conducted a qualitative study on the Nike Id program. Our research was consistent with the experiential view (Holbrook and Hirschman 1982) since we wanted to understand the value of MC during the customization experience. Therefore, verbal protocol analysis and structured interviews were used. Twenty French consumers aged from 14 to 31 took part in a mass customization experience and were asked to think out loud and were video-taped. After

212


the co-design experience, they watched the video of their customization. Then, a series of questions was asked about their preference for the product they customized. Content analysis was performed with NVivo software. Five usable value components were identified. Three of which: uniqueness value, utilitarian value and self-expressiveness value, can be associated with product value, and two: creative achievement value and hedonic value, can be associated with experiential value. Thus, three facets define the construct of mass-customized product value, and two facets define the mass customization experience construct. Table 1 describes the definition and statement related to each of the five components.

Value components

Definitions

Statements

Uniqueness value

Value acquired from the opportunity to assert personal uniqueness with the masscustomized product

"My goal in doing this is to be the only one to have it. It’s like: I got those shoes, not you"

Utilitarian value

Value acquired from the closeness of fit between product characteristics and individual preferences

"You could find the shape of a shoe interesting but not like the color...So here you can really do what you want"

Self-expressiveness value

Value derived from the opportunity to possess a product that is the reflection of personality

"It’s a nice thing to be able to create your shoes according to your personal taste as opposed to the pre-defined one the brand has created for you based on the wants and needs of the consumer. It’s more personal"

Creative achievement value

Value acquired from the accomplishment related to the creative task of co-designing

"The pleasure of doing it... yes... the satisfaction of doing something"

Hedonic value

Mass customization experience value

Mass-customized product value

Table 1: Product and mass customization experience components: Taxonomy and definitions.

Value acquired from the experience capacity to "mean a need of enjoyment, fun, pleasure" (Lai 1995)

"It’s a play thing...it’s fun, you do your own thing" "I could enjoy myself"


213

An initial pool of 36 items was generated for measuring each of these five value components using the statements of the qualitative study and literature on customer value (e.g. Aurier, Evard, and N'Goala 2004; Babin and Darden 1995; Evrard and Aurier 1996; Holbrook 1999; Mathwick, Malhotra, and Rigdon 2001). After an item reduction process involving several steps (8 expert coding and 12 "think-out-loud interviews" with consumers), 7 items were deleted and several were revised (see Appendix A: measurement items). The scales (7-point Likert scales) were then tested on two samples composed of 228 and 546 students (The first sample included 58% female respondents (age average: 22, S.D.: 4). In the second sample, 21 questionnaires out of 567 were deleted since they had too many missing values). In the first sample, we noted that the respondents generally agreed with the items: all the skewness coefficients were therefore negative (between -.149 and -1.265 for the items that we wanted to use for the second data collection). On average, only 15.34% of the answers were between the first ("strongly disagree") and the third point of the scale ("somewhat disagree"). In order to obtain more discrimination between respondents, we decided to use asymmetric scales for the second data collection (one negative point, one neutral and five positive points). This procedure was beneficial: the skewness coefficients improved for all of the items (between .918 and -.322). Measurements were analyzed for reliability and validity check following the guidelines provided by Anderson and Gerbing (1988). In the first data pool, we conducted a series of exploratory factor analyses (EFA) to purify the measurement models. Principal component analyses with promax rotation were conducted separately on product value items and on experiential value items. Finally, 23 out of 29 items were retained, accounting for 78.4% of the variance for the product scales (14 items) and for 68.4% of the variance for the experiential scales (9 items). In the second pool of data, another series of EFA was performed to examine whether the items could be loaded on their appropriate factors. After the deletion of one unstable item, the results were satisfactory: the axes accounted for 76.8% of the variance for the product scales and for 71.8% for the experiential scales, with each item loading strongly on the appropriate dimension and very little on the others. Next, a confirmatory factorial analysis (CFA) was carried out on the three facets of mass-customized product value and on the two facets of mass customization experience value (first-order confirmatory analyses), to assess the psychometric properties of the scales. After some re-specifications (Based on modification index and standardized residuals, the following items were deleted:

214


Un5, Un1, Uti5, SE1, U4 and U3 for the product value scale, and Hedo5, CA2 and Hedo2 for the experiential value scale.), the results of the measurement models indicated a good fit for product value (χ² /df = 1.96, GFI = .985, AGFI = .969, RMSEA = .042, TLI = .992, CFI = .995) and for experiential value (χ² /df = 2.76, GFI = .992, AGFI = .971, RMSEA = .057, TLI = .987, CFI = .995). Jöreskog’s ρ were all above 0.8, indicating good reliability for the scales. Average variances explained (AVE) were greater than 0.5 (between 0.62 and .81). Thus, convergent validity was assessed. Discriminant validity was established using two criteria. First, AVE exceeded the squared correlation for all of the constructs (Fornell and Larcker 1981) except for creative achievement value and hedonic value (AVE creative achievement = .62/ common variance with hedonic value = .63). However, none of the confidence intervals surrounding the factor correlations contains "1.00" (Anderson and Gerbing 1988). Therefore, discriminant validity was achieved. It should be noted that discriminant validity is difficult to establish when working with multidimensional constructs (Mathwick, Rigdon, and Malhotra 2001). The theoretical discussion suggested mass-customized product value and experience value are operating as second-order factors. The existence of this hierarchical structure was also tested. The loadings of the facets on their corresponding constructs are large, with coefficients all equal or above 0.7, supporting the hypothesized structure. Table 2 presents the measurement models results. Willingness to pay The majority of research on willingness to pay for mass-customized products has only used one evaluation method. However, the literature has suggested that different methods for measuring consumer willingness to pay result in different willingness to pay estimations (Völckner 2006). Franke and Piller (2004) found an average of 29% difference between willingness to pay for a mass-customized watch measured by Vickrey auction and by contingent valuation. Jorgensen, Syme, Smith and Bishop (2004) advocated a latent variable approach for measuring willingness to pay in order to assess the reliability of contingent values. In the same vein, Hauser and Urban (1986) used four indicators resulting from an unobserved "utility". To obtain a more reliable measurement, we also decided to apply three indicators of willingness to pay for mass-customized products. First, we used a variant of the double-bounded dichotomous choice contingent valuation (For further information on this method, see Wertenbroch and Skiera 2002). Contingent valuation is currently one of the most popular methods for measuring hypothetical willingness to pay.30 We began the process by giving the

215


respondent the price for a standard Nike pair of shoes (same model as the one they customized, but bought in a shop without any customization). The same "reference price" was announced to each respondent for practical reasons. It was defined as an average price on the market for a branded pair of shoes (€80). We requested the subjects to make a series of hypothetical buy/do-not-buy choices at two price points between €80 and €240. The presentation order of the prices was counterbalanced across subjects and had no effect on the subjects' responses. Then, they were asked exactly how much they were willing to pay for the shoes they customized. We used the price premium as an indicator of willingness to pay for mass customization which was measured as the difference between the maximum price people are willing to pay for their customized shoes and the reference price of €80 (price for a standard branded pair of shoes). Table 2: Measurement models results. Loading Components and manifest variables

Critical ratio Item

Facet

Mass-customized product value (AVE = 0.72) χ² /df = 1.96, GFI = .985, AGFI = .969, RMSEA = .042, TLI = .992, CFI = .995 (Since there are fewer than four first-order factors, the overall tests of goodness of fit of the models do not test the second order structure (Rindskopf and Rose 1988). They are approximately the same as those of the first order models.) Utilitarian value (AVE = 0.81, Jöreskog’s ρ = 0.83)

.90

14.075

U1

.86

*

U2

.84

21.651

Uniqueness value (AVE = 0.49, Jöreskog’s ρ = 0.88)

.70

*

Un2

.86

23.47

Un3

.82

*

Un4

.86

22.34

Self-expressiveness value (AVE = 0.88, Jöreskog’s ρ = 0.93)

.94

13.890

SE2

.89

34.47

SE4

.94

28.80

SE5

.86

*

216

HANDBOOK OF RESEARCH IN MASS CUSTOMIZATION & PERSONALIZATION Table 2 (Continued ) Loading Components and manifest variables

Critical ratio Item

Facet

Mass customization experience value (AVE = 0.81) χ² /df = 2.27, GFI = .992, AGFI = .976, RMSEA = .048, TLI = .990, CFI = .995

.997

Hedonic value (AVE = 0.99, Jöreskog’s ρ = 0.86)

*

H1

.73

21.35

H3

.86

21.4

H4

.86

*

Creative achievement value (AVE = 0.62, Jöreskog’s ρ = 0.77) CA5. CA6.

.79

*

.75

*

.82

18

Note: The metric for each scale was established by fixing one of the constructs loading to "1" (*). For mass customization experience value, non-standardized loadings between the second-order factor and the first-order factors were fixed to "1" since the second-order model only has two indicators (Rindskopf and Rose 1988).

Second, a conjoint analysis was performed. Two "super-attributes" (Green and Srinivasan 1978) were created: (a) type of product, and (b) price, as realized by Schoder and al. (2006) in the context of a newspaper’s mass customization. The lower price level was €80 (the same price as for the standard branded product in the contingent valuation), whereas the upper price was €160 (except for Franke and Piller’s results, the larger value increment found in the literature was around 200%). Schoder and al. (2006) have called for using more than two price levels even if the relative attribute importance could be affected artificially by the number of intermediate attribute levels (Wittink, Krishnamurthi, and Nutter 1982). Consequently, five price levels were retained. The final attributes and their levels are described in Table 3. A full-profile conjoint analysis was performed: respondents were asked to divide 10 stimulus cards (verbal descriptions) into two packs of five cards, the first one containing the pair of shoes they were most likely to buy, and the second one containing those they were least likely to buy. Then, they were asked to rank the cards according to their preferences. The path-worth utility of the "mass-customized product" level of attribute was used for measuring


217

willingness to pay since it expressed the value consumers place on masscustomized products. Table 3: Attributes and their levels. Attributes

Level

Type of product

Mass-customized pair of shoes Standard pair of shoes

Price

€80, €100, €120, €140, €160

Third, we used a 7-point Likert scale to measure mass customization overall value. Based on Zeithaml’s definition, Aurier, Evrard, and N'Goala (2004) proposed a scale to measure "a consumer’s overall assessment of the utility of a product based on the perceptions of what was received and what was given" (Zeithaml 1988). Thus, two items were created that integrated the trade-offs between the benefits of mass customization and two costs: price and time devoted to specifying product requirements (Items translated into English: (1) Finally, customizing a Nike Id pair of shoes is worth what it costs (in time and money), (2) Customizing a Nike Id pair of shoes is worth the time and the money, which I can devote to it.). The reliability of the scale was assessed (Cronbach’s alpha = .76). The two items were averaged to create a single indicator of willingness to pay for mass customization.

Figure 2: Measurement model of willingness to pay for mass customization.

Figure 2 depicts the measurement model of willingness to pay for mass customization, which is operationalized as a latent construct composed of three indicators.

218


The average variance extracted is 0.21, which is not a good score (Table 4). However, we retained all of measurements in the model since the t-statistics were acceptable (Hauser and Urban 1986). Table 4: Estimation results for the willingness to pay measurement model. Variable

Estimated factor loading (t-statistic)

Square correlation

Double-bounded contingent valuation

.516

0.27

Conjoint analysis

.404 (3.681)**

0.16

Overall value score

.446 (3.348)**

0.20

** p < 0.001

Out of the 546 respondents, 79 did not answer one of the three methods of evaluation. We decided to eliminate them from the analysis. Finally, 467 questionnaires were included in the study. Principal Results Preliminary results With the contingent valuation method, 73.3% (n = 343) of the subjects were willing to pay a premium for the pair of shoes they customized, while 26.6% were not. The average premium was €22.79 by taking into account "negative premium" (a 28.49% value increment, S.D. = €33.41), 37.23€ without. These results were far lower than those found in the literature. This could be explained by the high price of the standard model (€80). In the conjoint analysis, 86.3% (n = 403) preferred the mass-customized product to the standard product. In addition, 42.9% were willing to pay more (a €20 minimum incremental value) for the pair of shoes they customized. With the last measure, the average score of overall value is 4.19 (S.D. = 1.43). Results for H1, H2, and H3 Previous research has hypothesized the independence between mass customization experience value and mass-customized product value, leading to model M2 (Figure 3).

219


Figure 3: M2 Model: independence between value components.

This model was compared with our proposed model (M1, Figure 1) to determine which one fit the data better. The comparison of qualities of adjustment showed the superiority of the M2 model over all of the goodness-of-fit measurements (Table 5). In addition, the chi-square differences indicated that the fit of the M2 model was statistically better than that of the M1 model (∆χ² = 295.50, p < 0.001). Finally, more variance was explained by the M2 model (55.2% versus 44.3% for the M1 model). Table 5: Comparison of goodness-of-fit measurements for the competing models. χ² (df)

χ² /df

GFI

RMSEA

AGFI

TLI

CFI

M1 Model

504.23 (97)

5.20

.902

0.095

.862

.884

.906

M2 Model

211.733 (96)

2.21

.947

0.051

.925

.967

.973

The overall statistics showed that the structural model fit the data well. Table 6 suggests that the results support two out of the three hypotheses in the conceptual model. Thus, mass customization experience value has a strong influence on mass-customized product value, supporting H1 (p= 22)

(>= 25)

(3 x 330)

Product 1

23

25.5

3 x 330

1600

Product 2 *)

25

30

3 x 330

2000

Product 3 **)

24

25

3 x 330

1800

*) Has frequency converter drive, i.e. significantly lower power consumption **) Approved for running in explosion risky zones

In the second example (Table 2), it is shown that alternative safety valves can be selected. Two valve products satisfy the requirements but, as shown, with great difference between the prices. A significant attribute is the delivery time, which may set serious limitations regarding the time for procurement. However, this is dependent on the production location so, if for instance the production location is changed to the East Asia, a dramatic reduction of delivery time and price can be reached. Table 2: Alternative feed water pumps specified with a set of attributes. Set pressure bar(gauge)

Size

Production location

(Requirements)

(19)

(DN50)

(Deliv. location: Finland)

Product 1

19

DN50

Product 2

19

DN50

Delivery time

Price €

Germany

2 days

200

China

30 days

130

Two examples of abstraction by classification can also be presented (see (Jørgensen 2008) for description of the syntax). Example one is about oil fired boilers, where the module type "OilfiredBoiler" is the super-type for two sub-types "MissionOS" and "MissionOL". Two attributes show the decision making, "BurnerType" and "Capacity".

334


type OilfiredBoiler { BurnerType:

{KB,KBO,KBE,KBSA,KBSD};

Capacity :

(1.6 .. 15.5);

} type MissionOS subtypeof OilfiredBoiler { BurnerType :

{KB,KBO} default KB;

Capacity :

(1.6 .. 6.5);

.. } type MissionOL subtypeof OilfiredBoiler {...} Etc.

For oil fired boilers, the burner type can be any of the listed values, while for mission OS boilers only a subset of burners is valid. The capacity for mission OS boilers is similarly narrowed compared to the oil fired boilers in total. Example two regards feed water pump units, where there are two sub-types and where the regulation type differs. type FeedWaterPumpUnit {RegulationType : {OnOff,Modulating}; } type FeedWaterPumpUnitOnOff subtypeof FeedWaterPumpUnit {RegulationType : {OnOff}; } type FeedWaterPumpUnitModulating subtypeof FeedWaterPumpUnit {RegulationType : {Modulating}; }

Both examples show that the super-type modules represent decisions on a higher abstraction level because selection of a general module type establish some degree of specification while remaining decisions are postponed. In contrast, sub-types represent decisions about more precise specifications. In the sales process, it will be possible to assist the customers with decisions about how specific they must be from the beginning. A balance must be obtained. Relatively specific decisions


335

give more precise estimations (cost, required capacity, delivery, etc.) but are most likely subject to changes and, on the other hand, decisions on a more general level will lead to uncertainty about estimations. A key issue in relationship with configuration is to develop models for calculating estimations based on different levels of abstraction in decision making. Conclusions In this chapter, it is underlined that there are some fundamental issues of information modeling, which can be applied to product family modeling. For Product family models, it is important to identify the attributes in the model of the end-products and, because some attributes in models of product families will be assigned values during the configuration process, they must be defined with optional values i.e. domains. It is also characteristic for product family models that relations/constraints must be defined between attributes of the possible endproducts and the attributes of the identified modules/components. There is set special focus on how to develop product family models, which can support product configuration on multiple abstraction levels. First of all, it is proposed that configuration is performed by specification of attributes instead of selection of modules. This means that the structure of end-products is defined indirectly based on the values of attributes. Thereby, configuration is also oriented towards customer needs because attributes are essential in connection with the functional demands from customers. Further, it is proposed that, when modules are selected, it is important to develop classifications of module types and form a taxonomy. Such a structure is well suitable for identification of multiple abstraction levels by classification, where specifications can range from a general level to a more specific level. The aim of developing product family models is that they can be used as a foundation for development of specific product configurator software and the proposed methodology, included in this chapter, is for the moment being used by a particular ETO company, which intend to develop an advanced product family model and a product configurator that can support many organizational functions in the company world wide. Especially, the top-down approach with modeling on multiple abstraction levels are followed very closely and considerable amount of specially designed software modules are being developed.

336


References Andreasen, M. M. (2003). Relations between modularisation and product structuring. In Proceedings of the 6th workshop on Product Structuring – application of product models, MEK-DTU, Denmark: 1–15. Berman, B. (2002). Should your firm adopt a mass customization strategy? Business Horizons. 45(4): 51– 60. Davis, S. (1989). From future perfect: Mass customizing. Planning Review. 17(2): 16–21. Du, X., Jiao, J., and Tseng, M. M. (2000). Architecture of product family for mass customization. In Proceedings of the 2000 IEEE International Conference on Management of Innovation and Technology. Felfernig, A., Friedrich, G., and Jannach, D. (2001). Conceptual modeling for configuration of masscustomizable products. Artificial Intelligence in Engineering. 15: 165–176. Faltings, Boi and Freuder, Eugene C. (Ed.) (1998): Configuration: Getting it right. Special issue of IEEE Intelligent Systems. 13(4). Gilmore, J. and Pine, J. (1997). The four faces of mass customization. Harvard Business Review. 75(1): 91–101. Hvam, L. (1994). Anvendelse af produktmodellering, -set ud fra en arbejdsforberedelsessynsvinkel (in Danish). PhD thesis, Driftteknisk Institut, DTU. Hvam, L. (1999). A procedure for building product models. Robotics and Computer-Integrated Manufacturing. 15: 77–87. Jiao, J., Tseng, M. M., Duffy, V. G., and Lin, F. (1998). Product family modeling for mass customization. Computers & Industrial Engineering. 35: 495–198. Jørgensen, K. A. (1998): Information Modeling: foundation, abstraction mechanisms and approach. Journal of Intelligent Manufacturing. 9(6): 1998. Jørgensen, K. A. (2002): A Selection of System Concepts. Aalborg University, Department of Production. Jørgensen, K. A. (2003): Information Models Representing Product Families. Proceedings of 6th Workshop on Product Structuring, 23rd and 24th January 2003, Technical University of Denmark, Dept. of Mechanical Engineering. Jørgensen, K. A. (2005): Product Modeling on Multiple Abstraction Levels. Proceedings of IMCM'05 Klagenfurt, Austria, June 2–3, 2005. Jørgensen, K. A. (2008): Product Configuration and Product Family Modeling. Working Papers, Aalborg University, Dept. of Production. Lampel, J. and Mintzberg, H. (1996). Customizing customization. Sloan Management Review. 38: 21–30. Männistö, T., Peltonen, H., Soininen, T. and Sulonen, R. (2001). Multiple Abstraction Levels in Modeling Product Structures. Data and Knowledge Engineering. 36: 55–78. Pine, B. Joseph (1993): Mass Customization: The New Frontier in Business Competition. Harvard Business School Press, Boston Massachusetts, 1993. Pine, J., Victor, B., and Boyton, A. (1993). Making mass customization work. Harvard Business Review. 71(5): 108–119.


337

Reichwald, R., Piller, F. T., and Möslein, K. (2000). Information as a critical succes factor or: Why even a customized shoe not always fits. In Proceedings Administrative Sciences Association of Canada, International Federation of Scholarly Associations of Management 2000 Conference. Rosch, E. (1978): Principles of Categorisation. In: Cognition and Categorization. Laurence Erlbaum, Hillsdale, Jew Jersey, 1978. Rumbaugh, J., Jacobson, I. and Booch, G (1999): The Unified Modeling Language Reference Manual. Addison-Wesley 1999. Sabin, D. and Weigel, R. (1998): Product Configuration Frameworks. IEEE Intelligent Systems & Their Appplications. 13(4): 42–49. Silveira, G. D., Borenstein, D., and Fogliatto, F. S. (2001). Mass customization: Literature review and research directions. Int. Journal of Production Economics. 72: 1–13. Skyttner, L. (2005): General Systems Theory. 2nd edition, World Scientific Publ, 2005. Smith, J. M. and Smith, D. C. P. (1977): Database Abstractions: Aggregation. Communications of the ACM. 20(6): 405–413. Smith, J. M. and Smith, D. C. P. (1977): Database Abstractions: Aggregation and Generalization. ACM Transactions on Data Base Systems. 2(2): 105–133. Sowa, J. F. (1984): Conceptual Structures: Information Processing in Mind and Machine. AddisonWesley, 1984.

Author Biography Kaj A. Jørgensen is associate professor at Aalborg University in Denmark, where he leads the research group Information Technology in Industrial Systems. His background is research in information modeling and the current primary research activities are about product configuration, product modeling, product family modeling and building modeling. Contact: personprofil.aau.dk/Profil/100792 | [email protected]

3.3

Market-Based Strategic Platform Design for a Product Family Using a Bayesian Game Seung Ki Moon Department of Industrial & Manufacturing Engineering, The Pennsylvania State University, USA Timothy W. Simpson Department of Industrial & Manufacturing Engineering, The Pennsylvania State University, USA Soundar R.T. Kumara Department of Industrial & Manufacturing Engineering, The Pennsylvania State University, USA

The objective of this chapter is to propose a methodology for strategic platform design in a product family using concepts from game theory to model the situations of uncertain market environments. We identify module-based platform design for a product family and consider a module selection problem as a strategic game with incomplete information. In particular, a Bayesian game is employed to model uncertainty situations regarding market environments. The proposed Bayesian game is used to decide strategic equilibrium solutions for selecting modules in the product family being designed. To demonstrate implementation of the proposed Bayesian game, we use a case study involving a family of power tools.

Introduction For mass customization, companies are increasing their efforts to reduce cost and lead-time for developing new products and services while satisfying individual customer needs. Mass customization depends on a company’s ability to provide customized products or services based on economical and flexible development and production systems (Silveria et al. 2001). By sharing and reusing assets such as components, processes, information, and knowledge across a family of products and services, companies can efficiently develop a set of differentiated economic offerings by improving flexibility and responsiveness of product and service development (Simpson 2004). Product family design is a way to achieve cost-effective mass customization by allowing highly differentiated products to be

338


339

developed from a common platform while targeting products to distinct market segments (Shooter et al. 2005). In global sourcing and manufacturing environments, product development will be highly dependent on knowledge-intensive and collaborative systems for building on specialized knowledge across nations, organizations, and professions to develop customized products for different market segments (Szykman et al. 2001; Lefebvre et al. 2006). Knowledge-intensive and collaborative support has been increasingly important in product development to maintain future competitive advantage (Zha and Sriram 2006). In dynamic and uncertain market environments, however, we only have incomplete or uncertain information regarding market trends, customer’s preferences, production costs, and a company’s strategies for product development. To facilitate customized product design, we investigate strategic module sharing between products for designing a platform in a product family through a game theoretic approach in an uncertain market environment. Game theoretic approaches provide a proper framework for managing and evaluating strategies to achieve players' goals using their complete or incomplete information and knowledge (Gibbons 1992). A Bayesian game is designed to model situations wherein some of the players have incomplete information or uncertain characteristics for the other players (Osborne and Rubinstein 2002; Gibbons 1992). The objective of this chapter is to propose a methodology for strategic platform design in a product family using concepts from game theory to model the situations of uncertain market environments. We identify module-based platform design by introducing unique modules, common modules, and engineering parameter (EP) modules for product family design. We consider a module selection problem as a strategic game with incomplete information that is described by products' market share ratios and customer’s preferences. A Bayesian game is employed to model uncertainty situations regarding market environments and decide strategic equilibrium solutions for selecting modules in the product family being designed. The remainder of this chapter is organized as follows. We review related literature and background about product and service family design as well as game theory. Then we describe the proposed game theoretical approach for determining EP modules to design a platform using a Bayesian game. Next, we give a case study using a family of power tools. Closing remarks and future work are presented in the final section.

340


Background and Literature Review A product family is a group of related products based on a product platform, facilitating mass customization by providing a variety of products for different market segments cost-effectively (Simpson et al. 2005). A successful product family depends on how well the trade-offs between the economic benefits and performance losses incurred from having a platform are managed. Simpson et al. (2001) introduced a method to optimize a platform by minimizing performance loss and maximizing commonality. Gonzalez-Zugasti et al. (2000) designed platform modules to minimize design risk and save costs relating to developing a product family. Moore et al. (1999) used conjoint analysis to help determine a product platform. Siddique and Rosen (2000) described a method to design a platform from an existing group of products by comparing commonalities in assembly processes. Rai and Allada (2003) used a two-step approach to determine a modular platform for a product family, which consists of an agent-based optimal technique and post-optimization analysis using the quality loss function. Johannesson and Claesson (2005) proposed a configurable product platform design process and model using an operative product structure and a hierarchical function-mean tree to capture parameters describing design information such as rules, variants, requirements, and product configuration possibilities. Moon et al. (2006) developed a multi-agent system to identify and configure a module-based platform for a product family using a market-based reputation mechanism that implements a learning algorithm to select stable and reputable modules in an electronic market environment. Thevenot et al. (2007) introduced the design of commonality and diversity method (DCDM) to provide designers with recommendations for both the functional and component levels by the inherent tradeoff between commonality and diversity during product family and platform development. A game is a description of strategic interaction that includes constraints based on players' actions. Game theory provides reasonable solutions for various games and evaluates their properties (Gibbons 1992; Osborne and Rubinstein 2002). According to constraints and the situations of games, game theoretic models can be partitioned into three categories: (1) cooperative and non-cooperative games, (2) strategic and extensive games, and (3) games with complete and incomplete information. In engineering design, game theoretic approaches have been applied to model strategic relationships between designers for sharing design knowledge and solving design problems. Xiao et al. (2002) applied game theoretic approaches and design capability indices to model the relationships between


341

engineering teams that were described as cooperative, non-cooperative, and leader/follower protocols, and facilitate collaborative decision making during a product realization process. Fernandez et al. (2005) proposed a framework for establishing and managing collaborative design spaces by combining elements of cooperative and non-cooperative behavior, and formulating strategic and extensive games with utility theory. Kopin and Wilbur (2005) introduced a Bayesian game to model cost sharing in uncertain and incomplete information that were related to producer and consumer attributes such as nature, production costs, players and information, and preferences. Correia (2005) investigated the representation of incomplete and asymmetric information to model a strategic Bayesian game that was represented by the constraints of a transmission system and player’s strategic reactions to estimate uncertainties. Lewis and Mistree (1998) presented mathematical constructs for modeling a multidisciplinary optimization problem using game theoretic principles and the compromised Decision Support Problem (DSP) in a collaborative, sequential, and isolated design environment. In the next section, module-based platform design and the proposed Bayesian game for product family design are discussed in detail. Strategic Platform Design Based on Game Theory for Product Family Design Module-based platform design and engineering parameter (EP) cost module The basic idea of modular design is to organize products as a set of distinct components that can be designed independently and develop a variety of products through the combination and standardization of components (Kamrani and Salhieh 2000). Modules are achieved by decomposing product functions into functionally independent sub-functions in which interaction or interdependence between subfunctions is minimized (Tarasewich and Nair 2001). The modules make it easier to reuse in different products, allowing development and manufacturing costs to be significantly reduced (Ulrich and Eppinger 2000). Modules can be categorized based on function into: (1) unique, (2) common, and (3) variant or engineering parameter (EP) modules. Unique modules are based on distinctive functions within a product family – components in these modules cannot be replaced by those in different modules to fulfill their task. Different options within the product family can be designed as unique modules to create a variety of products. Common modules are based on common functions within a product family so that components in the modules can be shared. Variant or EP modules are based on common functions but differ in having different EP values. An EP module is a combination of one or several components that vary between products based on its EP values. The EP of a component is a representative

342


engineering parameter that is used for component selection in the market, and each EP component can be provided by a number of suppliers. For example, a motor in an EP module is represented by several EP values (e.g., torque, speed). If EP values that are selected for a platform do not meet the EP values of corresponding EP modules across a product family, additional EP components (i.e., gears with high ratios) are required to improve the EP values in terms of their functional requirements. As such, EP costs include the costs caused by changing EP components to increase the functionality of common EP modules. Let EPik be the EP value of EP module i in product k; let EPCik be the EP component costs at EPik ; let AC ik (α ) the assembly cost of EP module i in product k. The EP costs for EP module i in product k can be formulated at the module level as follows: EP costs = µEPC ik + D(ik , j ) + AC ik (α )

(1)

where α is a degree of cost reduction by commonality in assembly, D( ik , j ) is the EP loss cost in product j by sharing EP module i in product k, and µ is discount ratio for sharing the number of EP modules. For example, if a module is unique, then its EP value is calculated by combining the component cost and the assembly cost. If EPik is selected for a platform, then the EPC ik is reduced to µEPCik . Meanwhile D( ik , j ) is incurred when EPik does not meet EPij . The EP loss cost can be determined by the additional component cost needed to satisfy the functional requirements. Assembly cost depends on the number of EP components. Commonality reduces the number of different EP components to be assembled, resulting in cost reduction ( AC lk (α ) ). Table 1 represents these relationships in a matrix form for three EP (torque) modules of the family: EP11 (5N.m.), EP12 (7N.m.), and EP13 (10N.m.). To set the EP module cost, an industrial case study is necessary to determine the EP loss cost by investigating the relationships between modules in various conditions (da Cunha et al. 2007). A well-defined platform reduces production costs by improving economies of scale and reducing the number of different components that are used. Suppose that a product family consists of unique modules, common modules, and EP modules as illustrated in Figure 1. The platform level is defined as the number of modules in the platform and consists of the common modules and the EP modules. An appropriate platform level for a product family is determined by minimizing the EP costs associated with the EP modules. High levels of the platform (i.e., high commonality of the EP modules) decrease assembly and component costs while increasing EP loss costs. On the contrary, low platform levels (i.e., low common-

343


ality of EP modules) decrease EP loss costs while increasing assembly and component costs. The next section discusses the expected strategy cost for designing a platform. Table 1: EP Costs of an EP module for product family design. EP

EP 11 (5N.m.)

EP 12 (7N.m.)

EP 13 (10N.m.)

Unique EP11, EP12, EP13

EPC11 + AC11 (α )

EPC12 + AC12 (α )

EPC13 + AC13 (α )

Common EP11

µEPC11 + AC11 (α )

µEPC11 + AC11 (α ) + D(11, 2 )

µEPC11 + AC11 (α ) + D(11, 3)

EP12

µEPC12 + AC12 (α )

µEPC12 + AC12 (α )

µEPC12 + AC12 (α ) + D(12, 3)

EP13

µEPC13 + AC13 (α )

µEPC13 + AC13 (α )

µEPC13 + AC13 (α )

Unique modules

Component cost Assembly cost

EP modules

Trade-offs

EP loss cost

Common modules Product A

Product B

Product C

A Product family Figure 1: Trade-off in platform level selection in terms of EP costs.

Platform level

344


Expected strategy cost A platform designer determines a feasible set of strategies for the platform based on his/her design knowledge. The strategies are represented as alternative design methods and can be constructed by combining components in EP modules for a platform. Let S be a set of strategies, EP a set of EP values, L a set of products in a product family, and Q a set of module quantities. S, EP, L, and Q are finite sets. The expected strategy cost, c( si ) , for designer’s strategy s i ( i = 1,.., S ) is estimated by an expected strategy cost function: f i : S × EP × L × Q ℜ . Hence, the real number of f i ( s i , ep, l , q) represents the cost of strategy paying ep at quantity q for performing strategy s i for a platform from l products. For example, the expected strategy cost for s can be determined based on EP costs as: f i ( s i , ep, l , q) = η ×

∑ ( EPC

i

+ AC i + D i )

i∈l

l×q

(2)

where l is a strategy weighting function as follows: 1, if module is unique l=  L, otherwise

(3)

and q is a quantity function as follows: 1, if module is unique q=  χ , otherwise

(4)

where χ is a volume discount factor or market share ratio. For a given set of products, the value of c( si ) varies depending on the strategy for platform design. The expected strategy cost function will be applied to calculate players' payoffs for a game and can be developed by various cost functions based on products' characteristics and/or company’s strategy in product family development. The next section discusses a Bayesian game model for determining a platform design strategy. Bayesian game model for strategic platform design A module selection problem can be considered as a strategic game with incomplete information. The strategic game provides a useful technique for determining a strategy in uncertain environments (Gibbons 1992; Koessler 2004). In this chapter, we employ a Bayesian game to solve the module selection problem in given product family design.

345


Consider the following module selection problem for platform design in a dynamic and uncertain market environment. There are two players: (1) a designer who has module design strategies for a product family and (2) a customer who has prices for a module. The designer provides a module with the cost c, and the customer pays the price v for the module. The designer’s cost and the customer’s price are dependent on the market share ratio of the products and customer’s preference, respectively. The market share ratio and the customer’s preference are assumed to be independently and uniformly distributed based on their market information. The cost and the price are constrained to be non-negative. If the market share ratio is greater than or equal to the customer’s preference, then the module will be produced at a price equal to the average of the module’s cost and the customer’s price; otherwise, the module will be not produced. Finally, let us assume that the players are risk-neutral for their payoffs. Each player knows his or her own payoff function but may be uncertain about the other player’s payoff functions. In order to formulate the proposed scenario as a Bayesian game, we must first identify the action spaces, the type spaces, the beliefs, and the payoff functions (Gibbons 1992). In this chapter, Player 1 is a designer who knows module design strategies for a platform. Player 1’s action is to select a strategy among design strategies that are all possible combinations for EP module design. The set of actions, A1= {a1,1 , a1, 2 ,..., a1, n } , for Player 1 is represented by the design strategies 1

that can be developed by a designer for products. The set of types, T1,= {t1,1 , t1, 2 ,...t1,m } , is the values of the market share ratio for the products including 1

the module, and the values are obtained from a uniform distribution on [0,1] . Because the values of the market share ratios are independent, Player 1 believes that the probability, b1 , is uniformly distributed on [0,1] . Meanwhile, Player 2 is a customer who wants to buy a module with a market price. Player 2's action is to determine the price of the module. The set of actions for Player 2, A2= {a 2,1 , a 2 , 2 ,..., a 2,n } constitutes modules' prices based on the market price. In this 2

chapter, we define the market price as (2× wi c( si ) ), where wi is the proportion of the number of products that satisfy EP constraints by strategy i in a family ( 0 < wi ≤ 1 , i ∈ n1 ). The set of types of Player 2, T2,= {t 2,1 , t 2, 2 ,...t 2,m } , is represented 2

by customer’s preferences, and the values of the preferences can be obtained from a uniform distribution on [0,1] . Because the values of the customer’s preference are independent, Player 2 believes that the probability, b2 , is uniformly distributed on [0,1] . A1, A2, T1, and T2 are finite sets that are defined by the number of n1 , n2 ,

346


m1 , and m2 , respectively. Therefore, Player 1 may be uncertain about the Player 2’s payoff functions, since Player 1 may be uncertain about the types of Player 2, denoted by t −1 . In this game, the probability distribution b1 (t −1 | t1 ) is defined as Player 1’s belief about Player 2’s types, t −1 , given Player 1’s knowledge based on type, t i . According to the proposed scenario, Player 1 and Player 2 can have two possible payoff functions based on their selected types. The two players’ payoff functions are given by: u1 (a1* , a 2* ; t1 ) = (c + v ) / 2 − c , if t1 ≥ t 2

(5)

u 2 (a1* , a 2* ; t 2 ) = (c + v) / 2 − v , if t1 ≥ t 2

(6)

u1 ( a1* , a 2* ; t1 ) = u 2 (a1, a 2 ; t 2 ) =0, if t1 < t 2

(7)

where c is the expected cost based on a1* and is calculated by the expected strategy cost mentioned in Section 3.2, and v is the price of the module based on

a2* and is calculated by (the market price × t 2 ). Formally, this game is denoted by G = { A1 , A2 ; T1 , T2 ; b1 , b2 ; u1 , u 2 } .

In the scenario, Player 1 will try to seek for a module that provides more profits as a platform in uncertain customers' preferences based on minimizing the expected strategy cost in various market share ratios. Otherwise, Player 2 wants to buy a module that provides more functions in a product for maximizing own payoffs. In the proposed Bayesian game, a strategy for Player 1 can be represented by a function a1 (t1 ) specifying the market ratio that Player 1 would choose. In a Bayesian Nash equilibrium, Player 1’s strategy a1 (t1 ) is a best response to Player 2’s strategy a 2 (t 2 ) , and vice versa. Based on the Definition of Bayesian Nash Equilibrium (Gibbons 1992), the pair of strategies (a1 (t1 ), a 2 (t 2 )) is a Bayesian Nash Equilibrium, if for each t y in [0,1] , (y=1,2), a y (t y ) solves: max a y ∈A y

∑u

y

(a1* (t1 ), a 2* (t 2 ); t )b y (t − y | t y )

(8)

t − i ∈T− y

For a given value of a1 , the best response of Player 1 is obtained by: max ( a1

(v + c ) − c) Pr(t1 ≥ t 2 ) 2

(9)

where c is the expected strategy cost and is calculated by:

∑ ( EPC c =η ×

i

+ AC i + D i )

i ∈a1

l × t1

(10)

347


where i is the member of design strategies including EP modules, i=1,2,…, n1 . v is the market price obtained by: (11)

v = 2 × wi c( s i ) × t 2

And, the probability of t1 ≥ t 2 is 1

Pr(t1 ≥ t 2 ) = 1 − Pr(t1 < t 2 ) = 1 − ∫0 FT (t 2 ) f T (t 2 )dt 2 1

2

(12)

t2

where, FT (t 2 ) = ∫0 f T (t1 )dt1 . Here, f T (t1 ) and f T (t 2 ) are a uniform distribution 1

1

1

2

over [0, 1]. For Player 2, we can obtain the best response through the same processes. In this game, strategies for Player 1 represent the various module design methods depending on EP modules in a product family. Therefore, engineering parameters' values for platform design can be determined by selecting strategies in uncertain market environments. In the next section, the proposed Bayesian game is applied to determine an EP module for platform design using a case study involving a family of power tools. Case Study To demonstrate implementation of the proposed Bayesian game, a power tool family consisting of a jigsaw, circular saw, sander, drill, and brad nailer (Figure 2) is investigated. The reason for selecting this particular power tool family is the availability of data in a Design Repository at the University of Missouri-Rolla (UMR) (function.basiceng.umr.edu/repository). Currently, these products have common modules related only to the electrical components (i.e., the battery) at the platform level. Platform design strategies Using the information from the Design Repository, we can develop function structure models for each product based on the bill of materials, assembly relationship, functional flows, and energy flows. Then based on the functional structure models, modules for the products are defined using the heuristic method of (Stone et al. 2000). Table 2 shows modules for the tool family and their target engineering parameters in EP modules. There are 16 modules for developing the new product family: one common module, two EP modules, and 13 unique modules. In the case study, we assume that the engineering parameters of the products are defined as the torque of a motor and the current of a battery.

348


Jigsaw

Circular Saw

Drill

Sander

Brad Nailer

Figure 2: Power tool family. Table 2: Modules for the power tool family. Product

Common module

EP module

Unique module

Circular saw

Electronic

Motor (12 N.m.), Battery (5.0 A)

Blade mounting, Input1

Jig saw

Electronic


Conversion1, Blade mounting, Input2

Sander

Electronic


Conversion2, Sander mounting, Input3

Drill

Electronic


Drill mounting, Input4

Brad nailer

Electronic


Nail Hitter, Magazine assembly, Input5

Based on the information in the Design Repository, we can define components for designing EP modules as shown in Table 3. For example, a motor module consists of three components: a motor, a gear, and a shaft. As shown in Table 3, a designer can obtain information to fulfill module design from a market by an auction (Blecker et al. 2005; Xia et al. 2004). The information includes size, weight, type, cost, and quantity. We generate the numerical data based on unit cost that is

349


depended on components' EP values. Suppose that Table 3 illustrates the result of bidding with suppliers to generate platform design strategies in a market environment. Table 3: The Result of the auction with suppliers for EP modules. Module

Motor Motor

Gear Shaft

Battery

Engineering parameter

Size

Weight

Cost

Assembly cost

A

Torque 3 N.m.

3

5

5

2

B

Torque 5 N.m.

3

6

7

2

A

Gear ratio 2.5

2

2

4

1

B

Gear ratio 4.0

2

4

6

1

A

-

2

3

4

1

A

Current 5.0 A

2

2

4

1

B

Current 7.0 A

2

4

6

1

Component

Battery

Common and unique module design can be determined from an auction based on market mechanisms to minimize total production cost. Since the motor module and the battery module are EP modules, we suppose that they are involved in designing the platform. Therefore, appropriate EP values for EP modules can be determined by a game to design the platform. The designer can configure components to develop design strategies based on the EP values for the game with a customer. Table 4 shows possible design strategies of the motor module and the battery module for a platform. For example, from Table 4, a design strategy for the motor module consists of three components: Motor A, Gear A, and Shaft A. The resulting torque is 7.5 N.m. (torque 3 N.m × gear ratio 2.5), which satisfies the target engineering parameter for the sander. To determine the expected strategy cost, we use the expected cost functions, Equation (2). The strategies also should satisfy the constraints when the strategies are selected. The EP costs can be calculated by the EP cost model given by Equation (1). We assume that the loss cost of EP is defined as an additional component cost, which is 1 unit and 2 units for torque and current, respectively. We assume that a factor of overhead is 2 units. The designer calculates the excepted strategy cost based on the minimum cost of each design method for each product. For example, s m1 has a module cost of 13 (=5+4+4) and an assembly cost of 4 (=2+1+1) for satisfying the constraint of the sander. To satisfy the EP

350


value constraints for the circular saw, the jig saw, the drill, and the brad nailer, additional costs for s m1 are 5, 5, 8, and 8 units, respectively. Therefore, the expected strategy cost for s m1 is 44.4, if the value of market share ratio is 1. For a customer, the market price’s weight coefficients of four strategies can be determined by the proportion of the maximum numbers of satisfying EP value constraints. Table 5 shows four expected strategy costs and the weights of the market price coefficients, when the values of market share ratio (MSR) and customers' preference (CP) are 1, respectively. Table 4: Strategies for EP module design. Module

Strategy

Component

EP value

s m1

A,A,A

7.5 N.m.

sm 2

A,B,A

12 N.m.

sm3

B,A,A

12.5 N.m.

sm 4

B,B,A

20 N.m.

s b1

A

5.0 A

sb 2

B

7.0 A

Motor

Battery

Table 5: Expected strategy costs and market prices (MSR and CP =1). EP module

Motor

Battery

Strategy

Expected strategy cost

Market price coefficient

Market price

s m1

44.4

0.2

17.76

sm 2

40.4

0.6

48.48

sm3

40

0.6

48

sm 4

42

1

84

s b1

13.2

0.6

15.84

sb 2

14

1

28

351

VOLUME 1: STRATEGIES AND CONCEPTS Table 6: A Bayesian game for determining a design strategy. Bayesian game

Player 1

Player 2

Player 3

Player

Motor module designer

Battery module designer

Customer

Action

Select a design strategy

Select a design strategy

Select a market price

Type

Market share ratio

Market share ratio

Customer’s preference

Belief

Uniform probability [0, 1]



Payoff function

Profit (v+c)/2- c

Profit (v+c)/2- c

Profit (c+v)/2- v

Analysis for a bayesian game in platform design The game between a designer and a customer for platform design of this family is defined by the proposed Bayesian game. Table 6 shows the Bayesian game for determining EP modules with three players. In this case study, the Bayesian game focuses on determining an EP module for a platform based on designer’s action. To determine the best response of Player 1 and Player 2, we performed a sensitivity analysis for various market share ratios based on customer’s strategies. Figures 3 and 4 show the two designers' payoffs for platform strategies based on different market price strategies, when the customer’s preference is 0.6. The two designers' payoffs were calculated by Equations (5) through (7). Then, we determined a maximum payoff as Bayesian Nash Equilibrium within given customer’s preference using Equations (8) through (12). In these cases, s m 3 and s b1 are dominated strategies based on the designer’s payoff. Therefore, a motor module with torque 3 N.m. and gear ratio 2.5 and a battery module with current 5.0 A can be designed as a new platform. The results from a sensitivity analysis provide a designer with information for determining a platform strategy in an uncertain market environment. In conclusion, if the customer’s preference is predicted as 0.6, a new platform for five tools can consist of three modules that include an electronic module, a motor module (torque 3 N.m., gear ratio 2.5), and a battery module (current 5.0 A). Comparing this to the current platform for the five products, we can increase the number of common modules based on common functional features.

352

HANDBOOK OF RESEARCH IN MASS CUSTOMIZATION & PERSONALIZATION Market Price Strategy based on Sm1

Market Price Strategy based on Sm2 5

5

0

0.2

0.4

0.6

0.8

1

Sm2

-5

Sm3

Payoff

Payoff

0

Sm1

0

Sm4

-10

Sm1 0

0.2

0.6

0.8

1

Sm2 Sm3

-10

-15

Sm4

-15

Market Share Ratio

Market Share Ratio

Market Price Strategy based on Sm3

Market Price Strategy based on Sm4

5

5 Sm1 0

0.2

0.4

0.6

0.8

1

-5

Sm2 Sm3 Sm4

-10

Sm1

0 Payoff

0 Payoff

0.4

-5

0

0.2

0.4

0.6

0.8

1

-5

Sm2 Sm3

-10

Sm4

-15

-15

Market Share Ratio

Market Share Ratio

Figure 3: The payoffs of a designer for the motor module when customer’s preference is 0.6.

Market Price Strategy based on Sb1

Market Price Strategy based on Sb2

3

3 2

1

Sb1

0 -1 0

0.2

0.4

0.6

0.8

1

Sb2

Payoff

Payoff

2

1 0 -1 0

Sb1 0.2

0.4

0.6

0.8

1

Sb2

-2

-2

-3

-3 Market Share Ratio

Market Share Ratio

Figure 4: The payoffs of a designer for the battery module when customer’s preference is 0.6.

Through the case study, the proposed Bayesian game was demonstrated to determine the EP-value of an EP module to select appropriate modules for the platform. Therefore, the Bayesian game can facilitate product family design in various dynamic market environments. Closing Remarks and Future Work In this chapter, we have investigated strategic module sharing between products for designing a platform in a product family through a game theoretic approach in an uncertain market environment. Module-based design was introduced to explain trade-off in platform level determination for product family design using unique modules, common modules, and engineering parameter (EP) modules. We


353

considered a module selection problem as a strategic game with incomplete information that was represented by product market share ratio and customer’s preference. A Bayesian game was employed to model uncertainty situations regarding market environments and decided strategic equilibrium solutions for selecting module design strategy based on the expected strategy cost in product family design. We have applied the proposed Bayesian game to determine platform design strategies using a case study involving a family of power tools. Through the case study, we demonstrated that the Bayesian game can be used to determine the EP-value of an EP module that was described as a design strategy for platform design. Therefore, we expect that the Bayesian game can help to facilitate product family design in dynamic and uncertain market environments. To improve the Bayesian game, we need to develop a method that can identify modules based on the designers' knowledge and customers' requirements for establishing design strategies effectively. Since an expected strategy cost function is sensitive to parameters in the mathematical model, the parameters should be determined based on product characteristics, company's and customers' preferences, and a market environment. For a large-scale product family, an effective search algorithm is needed to generate a set of feasible strategies in a game. To explore the best response of players in Bayesian games, we should consider computationally intensive numerical integration such as Markov-chain or MonteCarlo simulation methods for the market share ratios and customer’s preferences. Future research efforts will be focused on improving the efficiency of the Bayesian game, developing design strategies for various product family environments, and expanding its application to developing a negotiation mechanism for web-based product family design. Acknowledgments This work was funded by the National Science Foundation through Grant No. IIS0325402. Any opinions, findings, and conclusions or recommendations presented in this chapter are those of the authors and do not necessarily reflect the views of the National Science Foundation. References Blecker, T., Friedrich, G., Kaluza, B., Abdelkafi, N. and Kreutler, G. (2005) Information and Management Systems for Product Customization, New York, YN, Springer Science Business Media Inc. Correia, P. F. (2005) Games With Incomplete and Asymmetric Information in Poolco Markets. IEEE Transactions on Power Systems. 20(1): 83–89.

354


da Cunha, C., Agard, B. and Kusiak, A. (2007) Design for Cost: Module-Based Mass Customization. IEEE Transactions on Automation Science and Engineering. 4(3): 350–359. Fernandez, M. G., Panchal, J. H., Allen, J. K. and Mistree, F. (2005) Concise Interactions and Effective Management of Shared Design Spaces: Moving beyond Strategic Collaboration Towards Co-design. ASME International Design Engineering Technical Conference & Computers and Information in Engineering Conference, September, 24–25, Long Beach, CA, Paper No. DETC2005-85381. Gibbons, R. (1992) Game Theory for Applied Economics, Princeton, NJ, Princeton University Press. Gonzalez-Zugasti, J. P., Otto, K. N. and Baker, J. D. (2000) A Method for Architecting Product Platforms. Research in Engineering Design. 12(2): 61–72. Johannesson, H. and Claesson, A. (2005) Systematic Product Platform Design: A Combined FunctionMeans and Parametric Modeling Approach. Journal of Engineering Design. 16(1): 25–43. Kamrani, A. K. and Salhieh, S. M. (2000) Product Design for Modularity, Boston, MA., Kluwer Academic Publishers. Koessler, F. (2004) Strategic Knowledge Sharing in Bayesian Games. Games and Economic Behavior. 48(2): 292–320. Kopin, V. and Wilbur, D. (2005) Bayesian Serial Cost Sharing. Mathematical Social Sciences. 49(2): 201220. Lefebvre, E., Lefebvre, L. A., Hen, G. L. and Mendgen, R. (2006) Cross-Border E-Collaboration for New Product Development in the Automotive Industry. Proceedings of the 39th Hawaii International Conference on System Sciences, Jan. 4–7, Kauai, HI. Lewis, K. and Mistree, F. (1998) Collaborative, Sequential, and Isolated Decisions in Design. Journal of Mechanical Engineering. 120(4): 643–652. Moon, S. K., Kumara, S. R. T. and Simpson, T. W. (2006) A Multi-Agent System for Modular Platform Design in A Dynamic Electronic Market Environment. ASME Design Engineering Technical Conferences & Computers and Information in Engineering Conference,Philadelphia, PA, September 10– 13, ASME, Paper No. DETC2006/CIE-99286. Moore, W. L., Louvier, J. J. and Verma, R. (1999) Using Conjoint Analysis to Help Design Product Platforms. Journal of product innovation management. 16(1): 27–39. Osborne, M. J. and Rubinstein, A. (2002) A Course in Game Theory, Massachusetts, MA, MIT. Rai, R. and Allada, V. (2003) Modular product family design: agent-based Pareto-optimization and quality loss function-based post-optimal analysis. Int. Journal of Production Research. 41(17): 4075–4098. Shooter, S. B., Simpson, T. W., Kumara, S. R. T., Stone, R. B. and Terpenny, J. P. (2005) Toward an Information Management Infrastructure for Product Family Planning and Platform Customization. International Journal of Mass Customization. 1(1): 134–155. Siddique, Z. and Rosen, D. W. (2000) Product Family Configuration Reasoning using Discrete Design Spaces. ASME Design Engineering Technical Conferences Proceedings, Baltimore, MD, ASME, Paper No. DETC00/DTM-14666. Silveria, G. D., Borenstein, D. and Fogliatto, F. S. (2001) Mass Customization: Literature review and research directions. International Journal of Production Economics, 72(1), 1-13. Simpson, T. W. (2004) Product Platform Design and Customization: Status and Promise. Artificial Intelligence for Engineering Design, Analysis, and Manufacturing. 18(1): 3–20. Simpson, T. W., Maier, J. R. A. and Mistree, F. (2001) Product platform design: method and application. Research in Engineering Design. 13(1): 2–22.


355

Simpson, T. W., Siddique, Z. and Jiao, J. (2005) Product Platform and Product Family Design: Methods and Applications, New York, YN, Springer. Stone, R. B., Wood, K. L. and Crawford, R. H. (2000) A heuristic method for identifying modules for product architectures. Design Studies. 21(1): 5–31. Szykman, S., Sriram, R. D. and Regli, W. C. (2001) The Role of Knowledge in Next-Generation Product Development Systems. Journal of Computing and Information Science in Engineering. 1(1): 3–11. Tarasewich, P. and Nair, S. K. (2001) Designer-Moderated Product Design. IEEE Transactions on Engineering Management. 48(2): 175–188. Thevenot, H. J., Alizon, F., Simpson, T. W. and Shooter, S. B. (2007) An Index-based Method to Manage the Tradeoff between Diversity and Commonality during Product Family Design. Concurrent Engineering: Research and Applications. 15(2): 127–139. Ulrich, K. T. and Eppinger, S. D. (2000) Product Design and Development, 2nd, Boston, MA, McGrawHill/Irwin. Xia, M., Koehler, G. J. and Whinston, A. B. (2004) O.R. Application: Pricing combinatorial auctions. European Journal of Operational Research. 154(1): 251–270. Xiao, A., Zeng, S., Allen, J. K., Rosen, D. W. and Mistree, F. (2002) Collaborating Multidisciplinary Decision Making using Game Theory and Design Capability Indices. 9th AIAA/ISSMO Symposium on Multidisciplinary Analysis and Optimization, September 4–6, Atlanta, Georgia. Zha, X. F. and Sriram, R. D. (2006) Platform-based product design and development: A knowledgeintensive support approach. Knowledge-Based Systems. 19(7): 524–543.

Author Biographies Dr. Seung Ki Moon is currently a Postdoctoral Research Associate of Mechanical Engineering at Texas A&M University. He joined Texas A&M in July of 2008. He received his Ph.D. from the Pennsylvania State University, University Park in 2008. He received the B.S. and M.S. degrees in Industrial Engineering from Hanyang University, South Korea, in 1992 and 1995, respectively. He was a Senior Research Engineer at the Hyundai Motor Company, South Korea. His research interests focus on family and platform design for products and services; universal design; strategic and multidiscipline design optimization; agent-based decision-making; engineering knowledge engineering; and intelligent information system and management. Contact: www.tamu.edu | [email protected] Dr. Timothy W. Simpson is a Professor of Mechanical and Industrial Engineering and Engineering Design at the Pennsylvania State University. He received his Ph.D. and M.S. degrees in Mechanical Engineering from Georgia Tech in 1998 and 1995, and his B.S. in Mechanical Engineering from Cornell University in 1994. He is the Director of the Learning Factory (www.lf.psu.edu) and the Product Realization Minor at Penn State. His research interests include product family and product platform design, mass customization, and data visualization to support complex systems design. He is an active member of ASME, AIAA, and ASEE. He is the Chair of the AIAA Multidisciplinary Design Optimization (MDO) Technical Committee and the Past Chair of the ASME Design Automation Executive Committee. Email: edog.mne.psu.edu/ | [email protected]

356


Dr. Soundar Kumara is the Allen, E., and Allen, M., Pearce Chaired Professor of Industrial Engineering at The Pennsylvania State University. He holds a joint appointment with the department of Computer Science and Engineering, and an affiliate appointment with the College of Information Sciences and Technology. He serves also as an Adjunct Professor of C R Rao Advanced Institute of Mathematics, Statistics and Computer Science (AIMSCS), University of Hyderabad, India. His research interests are in intelligent systems design, complex networks and sensor networks. He has won several awards including the Penn State Engineering Society Premiere Research Award, and the Penn State Faculty Scholar Medal- the highest research award at Penn state. He is also the recipient of PSU Graduate Faculty Teaching award. He is an elected Fellow of the Institute of Industrial Engineers and the International Academy of Production Engineering (CIRP).

3.4


Hans Johannesson Product and Production Development, Chalmers University of Technology, Sweden Stellan Gedell Saab Automobile AB, Sweden

Product platform development faces problems with inefficient knowledge management and reuse, as well as configuration strategies leading to intolerable part number growth. In order to address these problems a system oriented and abstract knowledge based approach is proposed to define and to describe configurable product platforms. A modeling procedure and a new fully configurable platform model concept, consisting of linked fully configurable generic and autonomous sub-systems, have been developed. The model has been implemented as a separate platform configuration (PFC) system being the base for system configuration. The proposed platform model has been partly verified and validated in cooperation with the industrial partners participating in the referred research project.

Background The main driving force for platform based development and manufacturing is the possibility to combine customization with economies of scale. The means to achieve this is reuse of common resources in multiple customized product variants. This is in practice described today in terms of "carry over" or "commonality" and means utilization of common parts in different customized variants. By doing this it is possible to create new product variants without having to develop all of its contained parts — just the ones that are variant specific. The rest can be "carried over" from already existing products or from a common core of parts in a product family or in different involved families of different brands — the product platform. Configuration of product variants is thus achieved by combining the parts in the platform with variant specific parts. Although much have been gained with this strategy it has its limitations, and it needs to be further developed in order to prevent the amount of part numbers to be managed in a developing and manufacturing company to grow out of hand and the reuse of common resources from restraining flexible configuration and further development. For a supplier delivering a customized sub-system solutions to different OEM companies or 357

358


other higher tier customers, a platform based product development approach becomes as important as it is for an OEM. Both need a product platform that can be configured to deliver customized variants to different customers. For the OEM the delivery is customized product variants to end customers, while for the supplier, with a B2B relationship to an OEM, it might be customized sub-systems that require further customization of the OEM to fulfill its end customers' needs. To deliver customized product variants by means of a product platform based on already designed and approved parts has drawbacks. An example, as indicated above, is that the customization in itself requires new customized parts to be developed and new customized systems and sub-systems configurations to be handled. This adds to the total amount of part numbers to be managed and maintained. Another, may be more important drawback, is that such a part based platform is inflexible to reuse in further development of the platform, as it becomes heavily constrained of already decided part design solutions. More efficient variant management and reuse abilities are regarded by the industry as important benefits to strive for. Present drawbacks could be overcome by adopting a more abstract and knowledge based platform description that comprises both the platform design rationale and the knowledge needed to compose customized product variants. The objective of this paper is to show how such a knowledge based configurable product platform could be realized. Product Platforms Product platforms and modular product development have grown in importance in recent years in order to lower cost (Huang and Kusiak 1998) by using the same unit in several products, and by having clear interfaces. The platforms enable companies to rapidly produce follow-on products or new variants (Meyer et al. 1997 and Halman et al. 2003), and to respond to market changes. This facilitates more customer oriented offers. As mentioned previously platforms enable economy of scale benefits in production and efficient utilization of resources of an enterprise. Building a platform can be seen as an evolutionary process where companies continuously have to renew their product families and eventually their platform to adapt to changing market needs (Meyer et al. 1997). There is not one unified way of describing what a platform is. Different researchers have different definitions and descriptions of how platforms relate to adjacent topics like product families, modules and brands (Halman et al. 2003). Simpson et al. (2001) defines a product platform to be a set of parameters, features, and/or components that remain constant from product to product within a given product family. Product families are defined as groups of related products that share

359


common features, components, and subsystems and satisfy a variety of market niches. Some parts are constant such as platform components and some vary in order to configure variants. This description of platforms is suitable for much of the platform development found in industry. This kind of traditional, part based platform descriptions have however shortcomings. Inclusion of other resources like manufacturing and organizational assets, as well as methods and IT tools, in such platform definitions is difficult. Robertsson and Ulrich (1998) give a wider description of the concept, describing platforms as a collection of assets, components, processes, knowledge, people and relationships that are shared by a set of products. Many companies define a product platform as the common resources within a single product family. Other companies like for instance in the automotive industry define platforms that carry multiple product families across different brands as shown in Figure 1. Other ways of looking at platforms are:

As an architecture which can support more then one product characterized by common structures, scaled variables and variable structures, controlled by design (Gershenson et al. 2006) This description expresses the need to exchange parts or components and also scale products to suit certain customer segments.

A basic architecture comprised of subsystems or modules and the interfaces between these modules (Meyer et al. 1997). This description of platforms addresses the need for interfaces between interacting systems.

Family A

Family A Family B

Brand 3

Family C

Platform C

Family A

Platform B

Family C

Platform A

Brand 2

Family B

Brand 1

Figure 1: Product platforms carrying multiple product families of different brands.

Platform development can, according to Jiao et al. (2006), be achieved in two different ways, design for functional variety or technical variety. The first aims to

360


satisfy diverse customer needs while the second aims at reducing the in-house variety. The two approaches require different strategies which address the two different advantages searched for in platform development, i.e. variety or reduction of unique parts. There is furthermore a difference in developing platforms for industrial products compared to consumer products. Industrial buyers often have deep knowledge about the product and the focus is more on facts like performance and cost instead of more vague criteria like aesthetics or consumer satisfaction. In the industrial case it is also easier to get a grasp of the market needs since the stakeholder group is a lot smaller (Jiao and Tseng 1999). In a supply-chain there is often an OEM on top producing for the consumer market. This OEM is at the same time an industrial customer buying components and sub-systems from its suppliers. The multi-brand platform approach, indicated in Figure 1, is today most often based on product descriptions composed of variants of part lists. Here the common parts of the platform are combined with the family variant and brand specific parts to achieve variety. Although reuse of common platform parts withholds the growth of unique parts when the variety is increased, the increasing demand for more variety still makes the growth of part numbers to manage a problem. For an OEM producing for the consumer market, a broader platform carrying all families and brands, as indicated in Figure 2a, would be the goal to aim for. Similarly a supplier, delivering variants of basically the same components and sub-systems to different customers, would benefit from the same platform approach (Figure 2b).

Family A

Customer 3

Family C

Family A

Family A

Family B

Family A Family B

Supplier platform

Customer 2

Family A

Family A

Customer 1

Family B

Brand 3

Family C

Brand 2

Family C

Brand 1

Family C

Platform ABC

(b)

Family B

(a)

Figure 2: A single platform carrying multiple product families for (a) different OEM brands, or (b) a suppliers' different customers.


361

Platform description in a PLM context Computer based PLM (Product Lifecycle Management) tools are today commonly used within product developing companies. Such tools have grown in importance due to their ability to handle product information like part descriptions, part structures and related documents (Saaksvuori and Immonen 2005, Stark 2005 and Xu and Liu 2003). Existing PDM (Product Data Management) tools capabilities are however mostly limited to handling single products and not complete product platforms. This problem, as well as the indicated problems with the conflict between reuse and commonality versus increased demands on product variety, call for new approaches for platform descriptions and IT support. What is needed is an integrating tool that handles all knowledge related to the whole platform system as well as to its contained sub-systems and components, the relations between contained items and the rules governing the use of the contained knowledge for different purposes during the platform lifecycle. The content of the platform and how this content is developed and described is crucial for the sake of maximizing both reuse and product variety simultaneously, and to this end a platform based on ready designed parts has severe shortcomings due to its poor flexibility. More abstract and easily configurable platform descriptions have therefore been proposed. van Veen (1990) proposes a generic bill-of-material (BOM) to provide possibilities to describe large varieties of product types and structures. The idea is to define product data on a level of “sets of product types” instead of defining "individual product types". This concept has been applied and explored further by Erens (1996) when developing product families and synthesizing product variants. Mannisto et al. (2001) propose a strategy related to the idea of a generic BOM. They describe a "master BOM" which is a generic description for many product variants that can be defined and manufactured based on the platform. From the "master BOM" different "order BOMs", defining products in an individual customer order, can be instantiated. A new strategy for handling the problems has been proposed by Claesson (2006). The proposed strategy is based on a more abstract and knowledge based platform definition consisting of linked system structures of configurable autonomous subsystems. With input values on its variant defining parameters such a configurable system structure automatically generates the variant defining information (i.e. part numbers, instantiated CAD models, material specifications, valid reference documents etc.) needed for preparation of production. To function as stated, each configurable sub-system must carry three kinds of knowledge about itself: (i) knowledge about its origin, (ii) knowledge about its interactions and (iii) knowledge about how to compose variants.

362


A platform definition based on this kind of knowledge carrying sub-systems provides much more configuration flexibility than a part based defined platform, and it does not generate new part numbers to be managed for each new configuration. In this work a platform is defined as "a common knowledge based configurable system model containing system design rationales (including requirements, generic design concepts and decision history) and rule based models for variant instantiation plus common resources used by the configurable system model". Such a configurable and knowledge based platform is much more robust as reuse of configurable sub-systems instead of reuse of parts makes it possible to have the complete product knowledge contents available for redesign in order to meet new demands. Research Approach The research presented in this paper is based on an abductive approach. Problems, identified in industrial interview studies and discussed in literature, have been addressed by proposing a new platform description concept for product platform development. It is aimed at supporting company in-house platform development as well as development collaboration in the supply-chain. The concept is based on systems engineering principles and design theory (Hubka 1997 and Andreasen 1998). It has been implemented in a software demonstrator which has been tested in a case study together width two of the industrial project partners. Interview studies The interview studies performed in this work have been made at two Swedish OEM companies active in the transportation and automotive industries and at a second tier sub-supplier in the automotive industry. The OEMs have several suppliers and their core competences are described in terms of domain specialization and system integration. The sub-supplier, which has a unique competence which is lacking at its customers, has many OEM and first tier customers. The interviews have been made with some 25 employees working with development at different departments in the three companies. Development managers, design engineers, PLM managers, process developers and marketing people involved in supply chain collaboration were purposefully selected for the interviews. The main theme for the interviews were how the investigated companies work and collaborate in platform development today, and which areas they think are necessary to improve in order to be more efficient, both in internal platform development and in buyer-supplier collaborations when developing platforms in the future. The main interview topics were: platform development, platform


363

definitions/descriptions, supplier collaboration, information exchange and IT support. Two different kinds of interviews, with different purposes were conducted. Semi-structured interviews, based on a pre-defined questionnaire, where the participants were allowed to speak freely around the topics and given questions were performed in order to get an over-all picture of the topics. The results from these interviews formed the foundation for the chosen PLM system architecture approach. The second kind of interviews was deep interviews with product specialists which provided in depth knowledge and information about the products, how they are developed and why they are designed the way they are. This input has formed the base for the knowledge based modeling of the test implementation. These deep interviews were of an iterative nature where the modelled results of our understandings were discussed, changed and complemented in a number of iterations. Theoretical approach The more abstract and knowledge based platform definition, consisting of linked structures of configurable sub-systems that was referred to above, was proposed by Claesson (2006) in his doctoral thesis. In his work the configurable autonomous sub-systems are called "configurable components (CCs)", and a product or a product platform is described by a linked set of such "configurable components" (Figure 3). Each CC object is a configurable system that has relations to other CCs, and it can be instantiated by setting values on its variant parameters (VPs) through a "variant parameter interface (VPI)". Composition of variants is defined by the "composition set (CS)", and the result of a composition is transferred to other used CCs or external applications like PDM and CAD systems. An instance of a product or a sub-system that shall be manufactured is specified by its variant defining parameters. With this input the system of involved CCs automatically generates the variant defining information (i.e. part numbers, instantiated CAD models, material specifications, valid reference documents etc.) needed for preparation of production. The described functionality can be achieved if each CC has the following necessary knowledge about itself:

Knowledge about its origin, i.e. what it should do and be, how this is realized and why the chosen solution is what it is.

Knowledge about its interactions with the external environment.

Knowledge about how to compose variants of its design solutions by means of its internal resources or by using other CCs. Knowledge item 1 is realised by describing the CCs design rationale (DR). This is done by using an enhanced function-means tree (Andersson et al. 2000) which is

364


reflecting the functional requirements (FRs), derived from the product specification, and shows the design solutions (DSs) chosen to fulfil these functional criteria and relevant design constraints (Ca, Cb, …Cn). Each FR, C and DS is described with variation ranges (bandwidth). This reflects the required variety of the design solutions implemented by the configurable component. Note that each design solution (DS) is an "organ" in the sense of the "theory of domains" (Andreasen 1998), i.e. it can be finally realized by physical hardware components or software components as

one single component or sub-system,

an integrated part of a complex component or sub-system

or a result of the interaction between two or more components and/or subsystems. Knowledge item 2 is realised with special interface design solutions (I/F) handling the external interactions. Finally the knowledge item 3 is realised with a configuration rule set (CRS) and a composition set (CS). The CRS containins programmed procedural and/or inferential rules linked to the CC object and its relations. So far only explicit knowledge has been modelled in the test implementation described below. There is however no restrictions to also handle tacit knowledge that can be formulated as design rules.

Modeling considerations and consequences The customized instantiation of a CC variant is realized by a number of composition elements (CE) in the composition set (CS) (Figure 4). Each CE is an implementation of (has an iaio-relation to) a design solution (DS). It uses variant parameter values (VPV1) from the variant parameter interface (VPI) and the configuration rule set (CRS) to generate the information defining an instantiated designed variant of a DS and its interfaces (I/F). A composed variant can consist of other used instantiated CCs. To use another CC the CE calls the CC to be used (applies the icu-relation) with generated variant parameter values (VPV2) as input to the VPI of the called CC. To use parts (HW or SW) that are already defined and handled by external application is done in the same way. The CE will call a CC that represents that part but this CC may only contain methods to generate parameters (x, y, z) that are needed to retrieve stored items or to instantiate parametric models in the application system (CAD, CAE, PDM, RM, etc) data bases. Similar methods can also be used by anyone of the CCs internal objects (i.e. FR, DS, C) or the CC itself to link to different models in external applications needed for description of those objects.

365


CC

Var def parameters CCA

CC

VPI

DR

Ca, Cb, …Cn

CCAa

DS

I/F FR

FR

FR

DS

DS

DS

CCAb

CCBa

CCBb

Var manuf data

FR

CRS

CCB

CS

Figure 3: Configurable components.

An important consequence of creating and defining the total customized system by letting instantiated CCs use other CCs for its composition is that the structure of a product instance is a result of the instantiation. It does not exist explicitly beforehand, just implicitly as different possibilities depending on different CCs composition abilities. A prerequisite to consider when making it possible for one CC to use another CC, is that the two CCs must be compatible. This means that

there are variant parameters with values (VPV2) in the using CCs CEs = variant parameters with values (VPV2) in the used CCs VPI

top level FRs with DSs in the used CCs function-means model correspond to matching FR-DS pairs in the using CC

design constraints imposed on and modeled in the using CC, that also should be met by the used CC, must be properly partitioned and modeled in the used CC. Besides imposing constraints from using to used CCs, constraints may be imposed within a CC when developing its function-means model as a consequence of the design decisions made. Those constraints can appear both as top level (CC overall) constraints and as dedicated constraints imposed on lower level DSs in the function-means model. A constraint (C) which is imposing a lower level DS is

366


indicated in the function-means model by a ipmb-relation ("is partly met by"). In the function-means decomposition process such a DS is either

assigned a version of the imposing constraint that is more detailed, or

decomposed further by formulating a sub-FR that drives a solution (sub-DS) that contributes to the fulfillment of the C.

CC1

VPI VP(VPV1i)

CC3

VPI VP(VPV3i)

DR

icu

CRS

I/F

DR

Ca, Cb, …Cn

I/F

FR

ipmb

iaio CS CEn(VPV2i), CEn+1(VPV3i)

CRS DS

C

FR

FR FR

FR

DS

DS DS

DS

Meth(x,y,z)

CS

icu

CC2

VPI VP(VPV2i)

DR

Ca, Cc, …Cm

FR

CRS

DS

I/F

CS

FR

FR

FR

DS

DS

DS

External application (CAD, CAE, PDM, RM, …..)

Figure 4: Composition: External and internal CC compatibility.

A product platform model must have "bandwidth" in order to handle the required product variety. For the CC based platform model this means that the platform defining objects in the CCs must be able to handle variety. These objects are functional requirements (FR), design solutions (DS), constraints (C) and variant parameters (VP). There are different kinds of variety expected from these different objects:

FR: Ranges of functional properties, alternative discrete functional properties, optional functional domains

DS: Ranges of solution characteristics and properties, alternative discrete solution characteristics and properties, optional design solutions

C: Ranges of constrained characteristics and properties, alternative discrete constrained characteristics / properties, optional constrained solution spaces.


367

VP: Ranges of variant parameter values, alternative discrete variant parameter values, design option defining variant parameter values.

To handle these different kinds of variation, the varying objects contain variation defining parameters. These parameters can be given values of the different types described above. Product Configuration in a PLM Context A previous attempt has been done (Claesson 2006) to implement configurable CC structures into commercial PDM software. That was done with without including the design rationale (DR) part of the integrated CC models. Within the project referred to in the present work a second attempt to follow the same approach, but including the DR part, failed due to modeling limitations in the PDM tools that were tried. A decision was therefore taken to develop a separate platform configuration system based on the configurable component approach. See Figure 5 where the platform configuration system is shown in a PLM environment. The implementation of the platform configuration system is made in the AKM software (AKM 2007). This is an object-oriented software for knowledge modeling that provides the high degree of modeling flexibility needed for the task at hand. With the chosen approach the complete product platform description will be carried by a number of software tools integrated in a PLM environment:

The platform configuration (PFC) system

A PDM system

CAE systems like CAD, FEM, MBS, etc.

A requirement management (RM) system

The PFC and the PDM systems are mandatory parts of the architecture. The PFC system, which is the user interface system for product variant configuration as well as for platform design and development, contains the configurable CC structure of a platform. In order to fully define a configured instance, the CC models need to link relevant instance related part and document identities in the PDM systems to their contained objects and relations. This is done with methods linked to the objects and relations in the CC structure. The AKM software provides the ability to create such methods that facilitate communication with other software systems.

368

HANDBOOK OF RESEARCH IN MASS CUSTOMIZATION & PERSONALIZATION Platform configuration system AKM

RM system

CC

VPI

DR

Ca, Cb, …Cn

FR

CRS

RM model

Integrated f-mCC models

Parts Models

PDM system

Documents

DS

I/F FR

FR

FR

DS

DS

DS

CS

Models

CAD/CAE systems

Figure 5: Proposed PLM architecture for product platform descriptions.

The role of the PDM system is to be a carrier and manager of the parts, documents and information belonging to the product platform. It is furthermore a carrier and manager of the information identifying all product models in different PLM systems belonging to the platform. Those are for example CAD models, FEM models, MBS models but also RM and PFC models. Finally the PDM system has its role as process work-flow manager. The design rationale models in the CC structures are carrying the knowledge about the origin of each CC. They explain what the sub-systems configured by the CCs should do and be, how the solutions are realised and why they are realised the way they are. The functional requirements (FRs) and constraints (Cs) contained in these models originate from the product platform specifications. If these are available as RM models in requirement management systems, appropriate links should be established between the RM models and the PFC models for traceability reasons. Such links could be realized by using AKM methods linking FR and C objects in the CC structures to corresponding requirement items in the RM system models. In the present work no such RM models have been available and no such links have been established. The necessary input to the created design rationale models has been elicited from requirement documents and interviews with experienced designers. In order to configure product instances that contain for example hardware components that are configured in size and shape, the PFC system must refer to CAE systems containing configurable hardware models. Parametric CAD


369

models are such examples. Here generic geometry models, with rules governing the geometric configuration, are stored in CAD system database. The administration of the CAD models is handled by the PDM system. By using a method, an object in the PFC systems CC structure can identify a model in the PDM system and call it in the CAD system. The instantiated parameters, that have been generated in the PFC system and govern the configuration of the parametric CAD model, are then transferred to the CAD system and the parametric CAD model is instantiated. Results: The Car Seat Platform Model A platform configuration system based on the integrated configurable component structure approach has been implemented using the AKM software. The chosen case is a car chair platform, and the implementation has been done in collaboration with a Swedish car manufacturer and a Swedish supplier of seat heating systems to the automotive industry. An initial verification and validation of the platform configuration system has been performed simultaneously with the modeling by adapting the model and the modeling procedure to the needs and requirements of the industrial participants in the referred project. The main purpose of the test implementation of the car seat platform is to validate the proposed platform modeling approach. The car seat example was chosen because of its suitable complexity and multi-technological contents. It is also a technical system of great interest of the industrial partners in the project. Although system complexity initially was considered to be suitable, it turned out to be very complex when the real modeling was done. It was therefore decided to model the overall structure but not to fill all parts of the model with detailed information to begin with. The detailed modeling was in this work focusing the parts that were of interest to validate the general CC modeling approach and its usefulness for the involved industrial companies. This is indicated in Figure 6 where the "Seat" system, its sub-system "Seat heat" and the sub-sub-system "Heat element", linked with "is_composed_using" (icu) relations, are shown in some detail, whereas other sub-systems to "Seat" just are indicated. The CC "Seat" is a system owned by the OEM. If there would have been a first tier supplier of seats involved in the project, this CC would have been an OEM configured CC from a more generic seat CC owned by the first tier seat supplier. In the present work no such first tier seat supplier was involved. The CCs "Seat heat" and "Heat element" are OEM configured CCs of more generic corresponding CCs owned by the second tier supplier of the seat heating system. To demonstrate how the model and modeling procedure work, modeling

370


of "Heat elements" that are used in "Seat heat" systems which in turn are used by "Seats" is described below. The configuration and composition procedure is illustrated in Figure 7. The CC "Seat" is configured by assigning values to twelve overall variant parameters (VPs) in its variant parameter interface (VPI). Four of these, "Seat heat" (Y/N), "Seat vent" (Y/N), "Seat cover" (Textile/Leather) and "Styling theme" (A, B, C), have some kind of influence on the "Seat heat" system that, if used, will be used in the "Seat" configuration. CC

VPI

DR

Ca, Cb, …Cn

Seat FR

CRS DS

I/F FR

FR

FR

DS

DS

DS

CS

CC

CC???

CC???

VPI

DR

Ca, Cb, …Cn

Seat heat

CC???

FR

CC???

CRS DS

I/F FR

FR

FR

DS

DS

DS

CS

CC

VPI

DR

Ca, Cb, …Cn

Heat element FR

CRS DS

I/F FR

FR

FR

DS

DS

DS

CS

Figure 6: Configurable car seat platform CC structure.

The function-means tree in the DR, defining what the system "Seat" shall be, do and how this is realized, has a number of top functional requirements. One is "Provide a place to sit". The DR also has a number of over-all design constraints like "Environmental friendly", "Fire resistant", "Crash safe", etc. Following a function-means decomposition procedure the top FR is decomposed in three steps, and then a fourth level sub-requirement "Avoid cooling" is found. A design solution, "Electric seat heating", is assigned to fulfill the FR. Relevant design constraints as "Environmental friendly" and "Fire resistant" apply on this design solution. To realize an instantiated variant of the DS "Electric seat heating" a composition element (CE), in the composition set (CS), uses the VP values, the relevant rules


371

in the configuration rule set (CRS) and the relevant interface design solutions (I/Fs) to compose a configured "Electric seat heating" instance. In order to do that the CE calls the CC "Seat heat" and transfers the appropriate variant parameters to its VPI.

Figure 7: Composition procedure.

The CC "Seat heat" is configured by assigning values to six overall VPs. Four of these, "Seat element size" (Shape, Dim), "Backrest element size" (Shape, Dim), "Nominal power" (x W) and "Surface layout" (Layout 1/Layout 2), have some kind of influence on the "Heat element" system that will be used in the "Seat heat" configuration. The function-means tree in the DR, defining what the system "Seat heat" shall be, do and how this is realized, has a number of top functional requirements of which one is "Provide uniform temperature distribution". Over-all design constraints of the DR like "Environmental friendly", "Fire resistant", etc. also apply on "Seat heat". Following the function-means decomposition procedure the top FR is decomposed one step. Then two second level sub-requirements "Heat seat" and "Heat backrest" are found. Design solutions, "Heat element", are

372


assigned to fulfill both FRs. The over-all constraints also apply on these design solutions. To realize instantiated variants of the DSs "Heat element" for the seat and for the backrest two separate composition elements (CEs) in the composition set (CS) use the VP values, the relevant rules in the configuration rule set (CRS) and the relevant interface design solutions (I/Fs) to compose configured "Heat element" instances — one for the seat element and one for the backrest element. In order to do that the CEs call the CC "Heat element" and transfer the appropriate variant parameters to its VPI. The CC "Heat element" is configured by assigning values to one overall VP, "Element shape" (Shape 1, 2, 3 and 4). The function-means tree in its DR, defining what the system "Heat element" shall be, do and how this is realized, has three top functional requirements of which one is "Heat seat surface". Relevant over-all constraints also apply on "Heat element". After two steps of functional decomposition the FR "Generate heat" is found. The design solution "Resistive wire", is assigned to fulfill this FR. To realize an instantiated variant of a DS "Resistive wire" a composition element (CE) in the composition set (CS) uses the VP value, the relevant rules in the configuration rule set (CRS) and the relevant interface design solutions (I/Fs) to compose a configured "Resistive wire" instance. In this case no other CC is called as no other CC is used for the composition. The instantiated variant of the "Resistive wire" is fully defined by the internal knowledge carried by the "Heat element" CC and the values of its governing VP. Discussion and Conclusion Case studies performed as part of this work have revealed that both OEM companies and suppliers want to reuse their product knowledge as much as possible in order to optimize economy of scale versus product variance. More efficient variant management and reuse abilities are regarded as important benefits to strive for. The paper addresses important issues concerning product platform descriptions and their consequences for customization and reuse of product knowledge in such industrial settings. Problems with existing platforms, based on common ready designed parts and related to part number growth and inflexible reuse, are focused. The objective of the paper is to show how a more abstract, knowledge based, configurable product platform, that can handle these problems, could be realized. A feasible solution — the proposed configurable product platform model and the modeling procedure — has been presented. The corner


373

stone of the solution is that each CC, being an autonomous system that can use other CCs, contains easy traceable and reusable knowledge giving

descriptions of the CCs offer to its environment, i.e. information about how it should be used in order to provide instantiated customized variants, and

a complete description of each represented systems origin, external interfaces and composition abilities, i.e. the systems complete design rationale.

In a sales-to-order, or a build-to-order, scenario this allows for definition of a new system product variant by means of a set of configuration parameter values, without creating new part numbers to manage. In an engineer-to-order scenario, as well as when extending an existing platform or when developing a new platform generation, the proposed platform description will be a rich, flexible and powerful source for reuse of not only ready designed parts, but also the complete design rationale with its requirements, design concepts, decision history and defined relations between interacting systems. The configurable platform model concept and the modeling procedure have been verified by implementing parts of a car chair platform in the AKM software. The functionality of the CC structure and the object contents, VPI, f-m model and CS, have so far been partly verified. Limited implementations, but still successful, have so far also been done of CRSs and I/Fs. Further implementations and tests will be done for full verification. Design engineers from one of the participating OEMs and from the participating supplier company have taken part in the implementation of the chair platform model for validation purposes. They have strongly influenced the information and knowledge contents of the model as well as the modeling of relations between different information and knowledge items. All involved have expressed their appreciation for the structured description and the holistic as well as detailed view of the platform that it provides. Its potential, both as a tool for efficient product customization and as a knowledge reuse base for internal platform development, has been regarded as high. The main driving force for platform based development and manufacturing is the possibility to combine customization with economies of scale. This is in practice achieved by "carry over" or reuse of common parts, i.e. the platform, in different customized variants. Configuration of product variants is achieved by combining the parts in the platform with variant specific parts. This strategy has its limitations as it drives an increase of part numbers to be managed and maintained in a developing and manufacturing company to grow out of hand. It furthermore restricts the reuse, in engineering-to-order and platform development situations, to

374


already designed parts, which can have severe negative consequences for new interacting design solutions. In order to address these problems a new more system oriented and abstract knowledge based approach is proposed to define and describe configurable product platforms in the present work. A modeling procedure and a new fully configurable platform model concept, consisting of linked fully configurable generic and autonomous sub-systems, have been developed. The model has been implemented as a separate platform configuration (PFC) system being the common base for system configuration. The proposed platform model has been partly verified and validated in cooperation with the industrial partners participating in the research project. The overall conclusions so far are that the platform model and modeling approach have the potential

to enable more efficient product customization without driving growth of part numbers to be managed and

to provide more efficient means for reuse of product knowledge for platform development This is all in line with the needs expressed by the industrial partners participating in the research project.

Acknowledgments The authors want to thank SSF, The Swedish Foundation for Strategic Research, and VINNOVA, the Swedish Governmental Agency for Innovation Systems, for funding this work, and the automotive OEM and supplier companies as well as Active Knowledge Modeling AS for their participation in the performed case studies. References Andersson, F., Nilsson, P. and Johannesson, H. (2000). Computer Based Requirements and Concept Modeling: Information Gathering and Classification. Proc. 12'th International Conference on Design Theory and Methodology ASME DETC2000/DTM-14561, September 10–13, Baltimore, Maryland. Andreasen, M. M. (1998). The Theory of Domains. The EDC Workshop on Understanding Function and Function to Form Evolution Cambridge, UK. Claesson, A. (2006). A Configurable Component Framework Supporting Platform-Based Product Development. PhD Thesis. Chalmers University of Technology, Göteborg, Sweden. Erens, F. J. (1996). The synthesis of variety: developing product families. PhD thesis. Eindhoven University of Technology, Eindhoven, the Netherlands.


375

Gershenson, J. K., Khadke, K. N. and Lai, X. (2006). A Research Roadmap for Robust Product Family Design. Report Dept. of Mechanical Engineering-Engineering Mechanics, Michigan Technological University, Houghton, Michigan, USA. Halman, J., Hofer, A. P. and van Vuuren, W. (2003). Platform-Driven Development of Product Families: Linking Theory with Practice. Journal of Product Innovation Management. 20: 149–162. Huang, C-C. and Kusiak, A. (1998). Modularity in Design of Products and Systems. IEEE Transactions on Systems, Man, and Cybernetics, Part A: Systems and Humans. 28(1): 66–77. Hubka, V. (1997). Principles of Engineering Design. Zurich, Switzerland: Heurista. Jiao, J., R. and Tseng, M. M. (1999). A methodology of developing product family architecture for mass customization. Journal of Intelligent Manufacturing. 10(1): 3–20. Jiao, J. R., Simpson, T. W. and Siddique, Z. (2006). Product Family Design and Platform-Based Product Development: A-state-of-the-Art Review. Journal of Intelligent Manufacturing (Special Issue on Product Family Design and Platform-Based Product Development). Mannisto, T. Peltonen, H., Soininen, T. and Sulonen, R. (2001). Multiple abstraction levels in modeling product structures. Data and Knowledge Engineering. 36(1): 55–78. Meyer, M. H., Tertzakian, P. and Utterback, J. M. (1997). Metrics for Managing Research and Development in the Context of the Product Family. JSTOR. 43(1): 88–111. Robertson, D. and Ulrich, K. (1998). Platform Product Development. Sloan Management Review. Saaksvuori, A. and Immonen, A. (2005). Product Lifecycle Management. Second ed, Springer. Simpson, T. W., Maier, J. R. A. and Mistree, F. (2001). Product Platform Design. Research in Engineering Design. 13: 2–22. Stark, J. (2005). Product Lifecycle Management, 21st Century Paradigm for Product Realization. Springer. van Veen, E. A. (1990). Modeling Product Structures by Generic Bills-Of-Material. PhD thesis Eindhoven University of Technology, Eindhoven, The Netherlands. Xu, X. W. and Liu, T. (2003). A web-enabled PDM system in a collaborative design environment. Robotics and Computer Integrated Manufacturing. (19): 315–328.

Author Biographies Dr. Hans Johannesson is chair professor in Engineering Design at Chalmers University of Technology in Gothenburg, Sweden, where he is heading the Product Development division. His research is carried out in close cooperation with Swedish automotive industry within the framework of the Wingquist Laboratory, where he also is responsible for the research focus area "Systems Engineering and PLM". Contact: www.chalmers.se/ppd & www.wingquist.chalmers.se | [email protected] Stellan Gedell is an industrial PhD candidate at Saab Automo bile AB, Sweden.

3.5

Change Prediction for Mass Customized Products: A Product Model View René Keller Department of Engineering, Cambridge University, United Kingdom Claudia M. Eckert Department of Engineering, Cambridge University, United Kingdom P. John Clarkson Department of Engineering, Cambridge University, United Kingdom

In the automotive industry, engine companies produce a highly complex product, with a need to satisfy stringent legislation emission targets while at the same time they are faced with a highly fragmented market of different customers demanding highly customized products. A main driver to be competitive in such a situation is to manage changes effectively. Change requests from new customer requirements even late in the process must be analyzed for potential costly knock-on effects to other, not necessarily connected components. This paper describes how the Change Prediction Method (CPM) can be applied for the assessment of knock-on change risks which supports companies in planning for changes before being implemented and allows for an improved planning towards mass customization.

Introduction Customers demand highly customized products that suit their individual needs as well as shorter delivery times and lower cost (Feitzinger and Lee 1997). This is a result of the "mass-market breakdown" (Hart 1995) leading to fragmented markets with diverse customers and a need for greater product variety and customization. Product life-cycles have become shorter, and it is difficult for both customers and manufacturers to keep up with technological change. At the same time, technology allows for flexible manufacturing and information technologies that enable industry to deliver a higher variety of products at lower cost (Da Silveira et al. 2001). Since the late 1980s, mass customization systems have been beginning to provide the ability to supply customers with such customized products and services in certain domains. These mass customized products and services are tailored towards individual customers. So far mass customization works best in domains

376


377

where a relatively simple product can be adapted to individual needs according to a standardized and well understood rules, for example jeans or shoes. However, complex products such as diesel engines also need to be produced in a way that they fit individual customers' needs (Eckert et al. 2003). Here, companies aim to design the product in an as flexible way as possible and then offer wide ranging option packages with different configurations for non-core parts such as the fuel pump or sump, hoping that each customer can find what they need within this offering. However, configuration management (Sommerville 2006), can not always provide the required product, and changes to generate an option might have to be made. Often specific target uses require functionality or product geometry not offered by the option package. In the defence industry, for example, it is common that weapon systems such as helicopters are highly customized towards customers' requirements to meet particular mission requirements or standardise across the fleet. Each customer is then able — when purchasing the product — to choose from a large variety of options or to modify the product to suit individual needs. This makes a flexible, agile manufacturing process necessary (Gunasekaran 1999) and also challenges traditional methods of designing as the core products must allow for such options. One key strategy to achieve this is to "master change" (Da Silveira et al. 2001). Changes occur throughout the life of a product, and most new designs in fact result from changing previous versions of the product. New customer requirements, legislation changes and the need to prevent and solve errors also make changes necessary. Designing a version or an option of an exiting product to fulfil individual customer needs — as in mass customization — is another source for changes. Eckert et al. (2004) argue that the process with which a change is decided on and carried out is similar for all changes, regardless of the cause of the change — for example, see (Jarratt 2004) for changes due to variants of existing products, (Keller 2007) for the design of a new generation of products based on changing existing ones. Changes are constraint cost through product and production properties, and designers need to be able to select change options which enable them to carry out the requested change at a reasonable cost. Managing changes, however, is a problem in industry. A study by the AberdeenGroup (Brown 2006) showed that the majority of changes cause "scrap, wasted inventory, and disruption to supply and manufacturing". Nevertheless, the report also shows that companies do not properly assess the consequences of changes. Only 11% of all companies were able to "provide a precise list of items affected by a change" in the development of a single product, while only 12% were able to assess the consequences of changes on the life cycle of the product. Another

378


survey of 50 German manufacturing companies (Deubzer et al. 2005) supports these findings. It is pointed out that managing engineering change is an important issue in manufacturing industry. 56% of all changes made to a design happen after the initial design phase and of these, 39% are said to be avoidable. Also, the later a change is implemented, the more costly it is. Clark and Fujimoto (1991) suggested a "rule of 10" meaning the cost of a design change grows by the factor of ten with each passing design phase. In light of mass customized products, where changes can occur very late — at the time a customer purchases the product — it is important to design a product in a way that it can then be used to produce variants, within limits, and versions to meet individual customer requirements (Eckert et al. 2003). A product that lies within the options offerings of a configurator can be produced relatively quickly and cheaply, whereas a product that requires a specific change is often delayed, as designers need to be pulled out of ongoing design activities to make the modifications. Design for Changeability (DfC) (Fricke and Schulz 2005) is centred around building flexible products and tries to incorporate change considerations early into the design. In this context — when changes can occur very late in the design — is also important to know the effects of possible changes due to customization early in the design and have a core product that prevents changes from propagating. For example, a common change request of engine customers is to change the oil filter position because of better accessibility. If such a change is anticipated early in the design, with all possible knock-on effects to other components, i.e., the Cylinder Block, such a change can easily be implemented. In all cases the key to offering customized engineering products is to anticipate potential changes at the beginning of the design process. This chapter introduces a case study in an automotive company that produces highly customized off-road diesel engines. It shows how the Change Prediction Method (CPM) for assessing change risks in complex design (Clarkson et al. 2004) can be applied to designing the next generation diesel engine and how it can be used to plan for customization. Decision Support for Managing Change Customization requests from customers are frequently dealt with through configuration systems, which include all the options that have been generated and the rules that govern their conflict. In this case change propagation has anticipated and work out in the rules that govern the compatibility of options. Changes are


379

required if an option is requested that is not part of the configuration systems offering. The implementation of a change can be seen as a normal design process, albeit on a smaller scale. Maull et al. (1992), for example, describe the change process as a six-stage process consisting of: 1) filtration of change request, 2) development of solution; 3) assessment of solution; 5) decision on which solution to adopt; 4) authorisation of change; and 6) implementation. In light of mass customization, a number of possible solutions must be generated, assessed and then decided upon. Also, each change can have knock-on changes that then have to be investigated (Figure 1).

Filtration of Change Request

Development of Development of Solution A Development Solution B of Solution Z Assessment of Assessment of Solution A Assessment Solution B of Solution Z

Decide on Solution

Authorisation of Change

Implementation

Figure 1: Stages of the engineering change process.

Decision support tools for change management focus on the early stages of the change process where the change and potential solutions are analyzed and the impact on the product is determined. In terms of mass customization, decision

380


support tools for managing change processes can help in the early design stage as they allow the identification of change risks later in the development process. "RedesignIT" (2004) is a tool that provides qualitative guidance for engineers based on models of the key components and attributes of a product. The software presents the user with all possible solutions to a change request. Then analysing all possible side effects of these solutions supports the designer in deciding on the best solution. This tool was used to model the changes made to a diesel engine that should be changed in order to have a larger output torque based on specific customer requests. C-FAR (Change FAvorable Representation, Cohen et al. 2000) also aims at identifying the effects of changes. However, the underlying data structure, which tries to capture all relevant aspects of the design artefact, limits its use to simple and small products. However, for complex products, an exhaustive assessment of all possible solutions for a change request can be highly difficult, especially when several options or variants of the same product are available. Earl et al. (2005) argue that in order to successfully predict changes, a higher level of abstraction is needed, because of the number of potential change paths and the fact that designers need to make decisions at any point in a chain of changes, which might well be sub-optimal from the perspective of the particular change. They therefore advocate a probabilistic modeling approach. CPM method The CPM method as described by Clarkson et al. (2004) is based on the calculation of indirect change propagation risks between components. The development of the method was triggered by the need of a UK aerospace company to have support for tendering, when decisions on the feasibility of new customer requests, as the changes required for new versions of helicopter are substantial (Clarkson et al. 2004). The basic assumption is that if one component changes, this change can have knock-on effects on connected components, meaning that there exists a probability that adjacent components change in response to the initiating component. These components can then in turn cause changes to adjacent components, so that change spreads through the system. The CPM method aims at identifying these "hidden" indirect change dependencies between components and drawing the attention of the design engineers and managers responsible towards high-risk connections. It was shown (Jarratt et al. 2004) that the results obtained through

381


this method match the expectations of experienced designers and that the method was also able to predict past cases of change propagation. Input Experts’ Knowledge

Change Case

Action

Output

Build Product Model

Product Connectivity Model

Stage 1: Build Model

Compute Combined Risks

Combined Model

Stage 2: Compute Risks

Analyse Risks

Change Risks

Stage 3: Analyse Risks

Figure 2: The Change Prediction method.

The change management methodology (Figure 2) using this method consists of three stages: 1) building a product model, 2) computing combined risks, and 3) analysing risks. The first stage involves the creation of a product linkage model. This model captures the components of a product and models linkages between them. This model is then further refined into a probabilistic model that also captures the likelihood and impact of a change propagating between each connected pair of components. This data is then used in the second stage to compute combined risk values. In the analysis stage (stage 3) this data is visualized in such a way that high-risk connections can be easily identified and acted upon. The CPM method has been applied in a several companies ranging from automotive to aerospace and has been implemented into a software tool (Keller 2007). Case Study Companies in the diesel engine market face the problem of having to produce highly complex products meeting stringent emission legislation, which also have to be mass customized for individual customers — in this case, off-highway equipment manufacturers purchasing these engines. The case study reported here was conducted at an international leading UK automotive company producing offroad diesel engines. The high diversity of customers demanding engines for a large variety of applications including agriculture, construction, marine and power

382


generation results in a large number of different variants of one particular engine family which is one of the key drivers for the development of engines. Each engine comes in a variety of versions to allow for mass customization and different customers — for example, approximately 250 different options for the four cylinder engine (Jarratt 2004) are available. A second main driver for the development of new engine generations is emission legislation (Jarratt et al. 2003) and the development of new engines has to be synchronised with new legislation standards. The first emission standard by the U.S. for off-road diesel engines, Tier 1, was phased-in from 1996 to 2000. The more stringent Tier 2 and Tier 3 standards are in effect since 2000. In 2008, the new legislation standard Tier 4, to be phased from 2008 to 2013, will require engines to have significantly less diesel particle and NOx output, which both have to be reduced by approximately 90%. To achieve such reductions, a number of technologies can be used, including advanced exhaust gas after treatment, similar to the standards for highway diesel engines. The company is currently in the conceptual design phase of new engines satisfying these new Tier 4 legislation standards. Case study methodology The case study was split into two phases. In the first phase, 20 interviews with design engineers were conducted focussing on change in general (Jarratt 2004). This first stage, which took place while the company was concerned with customizing engines for customer needs, identified that the diversity of the product range — with the different options of the different engine series — is seen as a strength as well as a weakness as it requires more design effort. In the second stage, 12 visits in total at the company’s main site were conducted focusing on the design of the next generation engine. In total, 10 semi-structured interviews with a Senior Engineering Manager and 3 group meetings with a design team were held and transcribed afterwards. The main topics were engineering freeze and the development of a change software tool. During three of these interviews, the CPM software tool, an implementation of CPM as described in Section 2, was used to build a model of the next generation diesel engine in interaction with the Design Engineer. On several occasions, the CPM software tool was presented to a wider audience at the company. In total, approximately 15 design engineers and managers got an impression of the look and feel of the software, and the provided feedback was used for further improvement of the software.

383


Cost data in change risks assessment A major shortcoming of the CPM method – which is based on change risks – was raised by some design engineers stating that especially managers are used to assess cost data, rather than risks to individual components. It was mentioned that besides the original risk data in the CPM tool, "The two other ones that I thought of were key ones, were time and cost". The following simple cost model was developed together with the company to incorporate available costing data into the CPM method and the software tool that was used to build the product model in Section 4. The basic assumption is that the cost of a component i (ci) can be divided into two parts. One is the cost that can be directly assigned to the component (cci), which is captured in conventional cost systems that capture procurement or material and manufacturing costs and can potential calculate life cycle cost. The new contribution of this tool is calculating the "hidden" dependency cost which results from propagation from changes to other components dci (see Equation 1).

ci = cci + dci

(1)

The term cci can be broken down into the terms shown in Equation 2. fixi denotes the fixed costs associated with component i, Ei represents the expected number of changes initiated by i, and reworki is the expected cost for redesigning the component.

cci = fixi + E i × rework i

(2)

The "hidden" dependency cost dci of a component is calculated as the product of the rework cost and the change risks induced by other components (see Equation 3). reworki is defined as previously, r(j→i) is the risk of component j affecting component i and Ej represents the expected number of changes initiated by component j. #comps

dci = rework i

∑ r ( j → i) E j

(3)

j =1 j ≠i

These cost estimates allow the calculation of the actual cost of the entire program by summing up the individual component costs. One can also investigate how much hidden cost is created due to change propagation in comparison to the actual cost of the product. The cost estimates from above can be simplified if the risk that a component affects itself with a change r(i→i) is set to 1 (see Equation 4).

384

HANDBOOK OF RESEARCH IN MASS CUSTOMIZATION & PERSONALIZATION # comps

ci = fixi + rework i

∑ r ( j → i) E j

(4)

j =1 j ≠i

Steering the design and versioning of the next generation diesel engine The CPM method, in combination with the cost model described in the previous section, was used to steer the design of the next generation diesel engine. The aim was to identify risky component connections that could lead to change propagation in order to be able to assess those in terms of component cost. This risk is intrinsic to the product and does not consider time considerations or planning risk, however it could guide the designers in looking timing information up in planning systems. Product description During the case study interviews, a DSM model of a diesel engine (refer to Figure 3 for an illustration the air and exhaust flow modelled in the DSM) aimed at lowpower market (90-130kW) satisfying the Tier 4 exhaust legislation was created. The core engine is displayed with its 6 cylinders in the centre of the diagram. One can see that the approach to reduce NOx output is to reinsert the exhaust back into the combustion.

Exhaust is transmitted back into the engine

Exhaust Pipe

Figure 3: Low-power diesel engine model.


385

This happens in two ways. On the one side, air is taken from the exhaust and transmitted to the engine via a Jacket Water Cooler. The second loop leads from the Variable Geometry Turbine (VGT) via the Aircharge Cooler (ATAAC) to the engine air intake and is similar to the normal cooling cycle of engines. A Diesel Particular Filter (DPF) removes particles from the exhaust. This model is used to analyse the potential change knock-on effects to two core engine components — which might have to change due to error prevention — and one component that is likely to change due to customer requests. Model building In order to reduce the model building effort, the core components and their interactions of a previous diesel engine model (Jarratt et al. 2004) were used and extended by the specific components that make the engine fit for the legislation laws. The linkages between components and the corresponding change likelihood and impact values were elicited in a session with a company representative, who has been part of a number of model-building exercises in the past. For model building, the new CPM software tool was utilized and the tool showed its capability in easily adding components and links using multiple representations. The resulting model contains 24 components with 110 links between them resulting in a density of 19.9%. The components can be classified into a) components that model the machine outside the engine, b) emission control components and c) core engine components. The resulting linkage model can be seen in Figure A-1 in the Appendix, the shading of the components reflects the classification from above. Initial analysis This section describes the standard visual analysis of the diesel engine for Tier 4, as, for example, shown in Keller (2007) and Keller et al. (2008) for the model of the previous generation engine model. The combined risk model can be seen in Figure A-2 (Appendix). One can see that the core engine components cause lots of change propagation to other engine components and that only few changes are initiated by the additional exhaust reduction components. The highest change risks are between the Cylinder Block and the Cylinder Head and from the Fuel Injection Assembly to both, the Cylinder Block and the ECM. Especially the link from the Fuel Injection Assembly to the Cylinder Block is interesting, as there is no direct link between those two components. This is a case where the CPM method identified a high-risk link that otherwise would not have been predicted

386


because it is an indirect change resulting from change propagation. This model was verified with company representatives. Of the additional components modeling the exhaust treatment, the VGT, the DPF and the Control Valve cause the most changes to other components. Components in the top left quadrant are propagation multipliers (i.e. they cause change propagation to other components), while components in the lower right quadrant are propagation absorbers (i.e. are highly affected by other component changes). Components located in the top right have high incoming and outgoing change propagation risk. The Propagation Absorber/Multiplier Portfolio (Keller 2007) shown in Figure A-3 classifies components depending on the change risks. Components in the top-left represent components that are highly affected — in terms of change risks — by changes to other components but in term have little effect on other parts of the product. Components in the bottom right cause changes to propagate to other components but are only minimally affected by change propagation. One can see the core components of the engine — Cylinder Head and Cylinder Block — are the highest risk producer and receiver and that the exhaust treatment components do not add significantly to the change risks — and thus are located in the bottom left corner. The ECM and the Wiring Harness are components that are mostly only affected by changes, while the Fuel Injection Assembly is a component that, when changed, causes lots of change risks to other components. The cost and change data collected indicated that two components of the core engine are likely to change as an initiating component: the Piston and the Fuel Injection. As a third change case, the Transmission was chosen as this is a component that is on the interface of the engine and is very likely to change due to varying customer requests when an engine is mass customized. As an example, the change case of changing the Fuel Injection will be discussed here in detail and it will be shown how the interactive diagrams can help in analysing this change case; similar analyses can be carried out for the other Change Cases. The results for all three change cases are summarized in Table 1. This table shows the components that have the highest risk of being changed. For the first two cases, the most risky components are the Cylinder Head, the Cylinder Block and the ECM. The case where the Fuel Injection is changed results in highrisk for the Cylinder Block which as shown beforehand, is not directly connected to the Fuel Injection Assembly. Thus, an indirect high-risk component interaction was identified. In the case of a change to the Transmission, the Wiring Harness and the ECM are the most risky components. Figure A-4 in the Appendix shows the resulting Risk Network — where the distance to the central component is inversely proportional to the risk that it is

387


affected by a change to the component in the centre — for changing the Fuel Injection Assembly. In this display, the Cylinder Block is highlighted as well as all connections to and from the Cylinder Block. One can see easily that there is no direct link between the Cylinder Block and the Fuel Injection, but the close proximity to the Fuel Injection indicates a high risk. One can also see that this display incorporates both connectivity and risk as it is still possible to observe components that are directly connected to the Fuel Injection Assembly (see highlighted links). Table 1: Change Cases for the Tier 4 diesel engine. Initiating Change

Affected Components

Piston

1. Cylinder Block Assembly 2. ECM 3. Cylinder Head Assembly 13. VGT 14. DPF

Fuel Injection Assembly

1. ECM 2. Cylinder Head Assembly 3. Cylinder Block Assembly 13. VGT 14. DPF

Transmission

1. Wiring Harness 2. ECM 3. Crank Shaft 11. DPF 12. VGT

A Propagation Tree (Keller 2007) as shown in Figure A-5, where all occurrences of the Cylinder Block are highlighted, verifies that there is a high risk of a change to the Fuel Injection, which affects the Cylinder Block. In this display, all occurrences of the Cylinder Block are highlighted. One can see that there is no direct connection between the two components; however, there is a high number of propagation paths between both components resulting in an overall high change likelihood. Figure A-6 gives the Case Risk Plot with the Cylinder Block, the Cylinder Head, the ECM and the two exhaust treatment components VGT (Variable Geometry Turbine) and DPF (Diesel Particular Filter) highlighted. This plot verifies the findings from the previous visualizations, showing that these components are most

388


likely to be affected by a change to the Fuel Injection Assembly. One can also see that of the new exhaust treatment component, the Variable Geometry Turbine and the Diesel Particular Filter are most likely to be affected by changing the Fuel Injection Assembly. To summarize, the analyses showed that for the first two given change cases that resulted from changing core engine components, the same components were likely to be affected, which were the Cylinder Block, the Cylinder Head and the ECM. The high-risk connection from changing the Fuel Injection to the Cylinder Block was especially interesting as it was an indirect one. The exhaust treatment components all only added small risks to the entire model and did not have a high risk of being affected by the Change Cases. Of these, the DPF and the VGT were the most risky ones. The third Change Case of changing the Transmission due to a customer request resulted in changes to the Wiring Harness and the ECM, new exhaust treatment components were not affected. Cost analysis With the help of the company interviews and documents, cost and rework data were elicited. The motivation for this research was mentioned in one interview when the engineer stated that he would be "expecting it (the new legislation) to double the cost of the engine". The expected number of initiating changes regarding different components was based on his experience and reflects how often a component is usually changed due to new customer requirements or design errors as change initiators. Based on these costs and the cost model established in the previous section, the hidden costs for each component were computed. Figure 4 shows a bar chart of the component costs. For each component it gives the direct costs as well as the dependency costs. One can see that the DPF is an especially expensive component on its own. Due to indirect change propagation, this cost is further increased by 40%. Other components, such as the Wiring Harness, are even more affected by indirect changes. This verifies the previous analysis and the cost of the Wiring Harness, for example, grows by 240%, but due to its little initial cost, the overall effect remains quite small. One can also see, that some of the core components of the engine, such as the Cylinder Head and the Cylinder Block, have an exceptionally huge cost increase due to indirect change propagation and, on average, these components are twice as expensive when considering indirect changes. The resulting costs for the three Change Cases established previously can be found in Table 2. This table lists the total costs for each of the cases and also breaks down which component changes add the most cost. The real numbers are omitted for confidentiality and given in relation to changing the Piston; however, one can see that changing the Fuel Injection is approximately

389


25% more expensive than changing the Piston and equally expensive as changing the Transmission.

Figure 4: Direct and dependency costs for all components of the low-power diesel engine. Table 2: Cost estimations for the three change cases. Initiating change

Piston

Fuel Injection Assembly

Transmission

Total cost (from initiating component)

Components (cost contributed)

1 (14%)

1. Cylinder Block Assembly (26%) 2. DPF (18%) 3. Piston (14%) 4. Transmission (13%) 5. Cylinder Head Assembly (10%)

1.25 (4%)

1. Cylinder Block Assembly (25%) 2. DPF (21%) 3. Transmission (14%) 4. Cylinder Head Assembly (11%) 5. ECM (8%)

1.24 (71%)

1. Transmission (71%) 2. DPF (8%) 3. Cylinder Block (8%) 4. ECM (3%) 5. Crankshaft (2%)

390


The main finding is that the expected cost of a change only partly depends on the cost for the initiating component. For example, changing the Piston including all possible knock-on effects is expected to cost approximately 7 times the cost of just changing the Piston in isolation. One can also see that despite the small risks of affecting the DPF as seen in Figure 4, for all Change Cases, the costs contributed by changing the DPF are quite high. The other component that is not in the core engine is the Transmission, which is very costly to change in the first two cases. In case the Transmission is changed due to a customer request, most of this cost will be contained within this component and only 29% of the costs are due to change risks to other components. This is an important point as it shows that the engine is flexible enough to accommodate changes to the Transmission without too many implications to the rest of the engine. Discussion The CPM model analysis can also provide support in the latter stages of the conceptual design process. The analysis of the overall change risks helped in searching for working principles and working structures, as high-risk component connections were made visible and allowed to design engineers to made decisions accordingly, such as by providing enough margins so that changes can be handled locally. The analyses also allowed combining and firming up into concept variants, as different scenarios — e.g., where certain component connections are forbidden — can be analyzed. The cost calculations allowed evaluating the concept against (technical and) economic criteria, as it clearly showed which changes might result in knock-on changes to very costly components and it also provided a cost estimate for all components of the product. The main message to the company based on the above analyses is that changes propagating to the expensive Diesel Particular Filter (DPF) should be avoided at all cost as its high cost is even increased by changes propagating from other components. It also has a significant effect on other component changes due they could result in a knock-on change to the DPF. In the case of the three Change Cases, which resembled standard change requests — be it from customers or to correct errors in the design — the method described here highlights that most of the cost associated with a change result from the risk of propagating changes. It also gives the opportunity to choose the most cost effective strategy to implement a change. If, for example, a higher power request from a customer can be achieved by either changing the Piston or the Fuel Injection, the Change Case analyses have shown that it might be beneficial to change the Piston — although changing the Piston is more expensive in the first place (0.14% of 1) — rather than the Fuel


391

Injection Assembly (4% of 1.25), because changing the Fuel Injection Assembly carries a higher risk of causing high costs due to change propagation. The cost data analysis also allowed the identification of costly components that could be changed without causing too many risks to other components in the engine. For example, the Transmission is a part that is likely to be changed by customers as it is at the interface of the engine. However, in contrast to the two Change Cases where core engine components were changed, the costs of the component change of the Transmission is mostly contained within the Transmission and change propagation to other parts only adds a small amount to this costs: 29% instead of almost 96% in case of the Fuel Injection. Conclusions This paper showed how a technique for predicting the effects of change propagation can be applied for supporting the design of products for mass customization. The model is probabilistic. It does not indicate which components needs be to redesigned, rather it can help identify components within a product that can accommodate different customer needs that could lead to changes to components. For example, the Transmission in the diesel engine is a component that can — and is likely to — be changed due to late customer requests. However, it is designed in such a way that changing it requires mostly rework within the Transmission, preventing changes from propagating to other components. On the other hand, changing a core engine component such as the Piston will lead to expensive change propagation to other costly components. The use of the CPM method also supports the identification of components that should be prevented from being changed from change propagation resulting from late changes. The Diesel Particular Filter, for example, only has a low risk of being changed, however, in case it is changed, extensive costs can be expected. All in all, the use of the CPM software tool and method has shown to be beneficial to an automotive company planning for the next generation highly customized diesel engine. Acknowledgments This research is funded by the EPSRC. The authors would like to thank Perkins Engines Co. Ltd. and in particular Richard Weeks for the help with this paper.

392


Appendix

Figure A-1: Linkage model of the Tier 4 low-power diesel engine. The colors indicate that type of linkage between components.


393

Figure A-2: Combined Risk Plot of the Tier 4 diesel engine. The squares indicate the risk of change spreading from columns to rows, where the horizontal axis shows likelihood and the vertical axis shows impact.

394


Figure A-3: Propagation Absorber/Multiplier Portfolio for the Tier 4 Diesel Engine.


395

The colored rings indicate risk bands, the closer a component is drawn to the initiating component located in the centre, the more it is affected by change propagation.

Figure A-4: Risk network of changing the Fuel Injection.

396


This plot shows the resulting propagation paths from changing the Fuel Injection Assembly.

Figure A-5: Changing the Fuel Injection: Propagation tree.


397

Figure A-6: Changing the Fuel Injection: Case risk plot. The x-axis represents change propagation likelihood, the y-axis change impact. Thus, components located in the top-right carry a high risk of being affected by change propagation from a change to the initiating component (which is the Fuel Injection Assembly in this case).

398


References Brown, J. (2006). Managing Product Relationships: Enabling Iteration and Innovation in Design. The Managing Product Relationships Business Value Research Series. Boston, Massachusetts, USA, AberdeenGroup. Clark, K. B. and T. Fujimoto (1991). Product Development Performance: Strategy, Organization and Management in the World Auto Industry. Boston, Massachusetts, USA, Harvard Business School Press. Clarkson, P. J., C. Simons and C. M. Eckert (2004). Predicting change propagation in complex design. Journal of Mechanical Design. 126(5): 765–797. Cohen, T., S. B. Navathe and R. E. Fulton (2000). C-FAR, change favorable representation. ComputerAided Design 32(5): 321-338. Da Silveira, G., D. Borenstein and F. S. Fogliatto (2001). Mass customization: Literature review and research directions. International Journal of Production Economics. 72(1): 1–13. Deubzer, F., M. Kreimeyer, B. Rock and T. Junior (2005). Der Änderungsmanagement Report 2005. CiDaD Working Paper Series. 1(1): 2–12. Earl, C., C. M. Eckert and P. J. Clarkson (2005). Predictability of change in engineering: a complexity view. ASME 2005 Design Engineering Technical Conferences, Long Beach, California, USA. Eckert, C. M., P. J. Clarkson and W. Zanker (2004). Change and customization in complex engineering domains. Research in Engineering Design. 15(1): 1–21. Eckert, C. M., U. Pulm and T. Jarratt (2003). Mass customization, change and inspiration: changing designs to meet new needs. International Conference on Engineering Design (ICED 03), Stockholm, Sweden, CD-ROM. Feitzinger, E. and H. L. Lee (1997). Mass customization at Hewlett-Packard: The power of postponement. Harvard Business Review. 75(1): 116–121. Fricke, E. and A. P. Schulz (2005). Design for changeability (DfC): Principles to enable changes in systems throughout their entire lifecycle. Systems Engineering. 8(4): 342–359. Gunasekaran, A. (1999). Agile manufacturing: A framework for research and development. International Journal of Production Economics. 62: 87–105. Hart, C. W. L. (1995). Mass customization: conceptual underpinnings, opportunities and limits. International Journal of Service Industry Management. 6(2): 36–45. Jarratt, T. (2004). A model-based approach to support the management of engineering change. PhD Dissertation, Engineering Department. Cambridge, UK, University of Cambridge. Jarratt, T., C. M. Eckert and P. J. Clarkson (2004). Development of a product model to support engineering change management. Tools and Methods of Competitive Engineering (TCME 2004), Lausanne, Switzerland. Jarratt, T., C. M. Eckert, R. Weeks and P. J. Clarkson (2003). Environmental legislation as a driver of design. International Conference on Engineering Design (ICED 03), Stockholm, Sweden. Keller, R. (2007). Predicting Change Propagation: Algorithms, Representations, Software Tools. PhD Dissertation, Department of Engineering. Cambridge, United Kingdom, University of Cambridge. Keller, R., C. M. Eckert and P. J. Clarkson (2008). Using an engineering change methodology to support conceptual design. Journal of Engineering Design (in press).


399

Maull, R., D. Hughes and J. Bennett (1992). The role of the bill-of-materials as a CAD/CAPM interface and the key importance of engineering change control. Computing & Control Engineering Journal. 3(2): 63–70. Ollinger, G. A. and T. F. Stahovich (2004). RedesignIT: a model-based tool for managing design changes. Journal of Mechanical Design. 126: 208–216. Sommerville, I. (2006). Software Engineering, 6th Edition. Harlow, Addison Wesley.

Author Biographies Dr. Rene Keller is a Research Associate at the Engineering Design Centre located within the Department of Engineering at the University of Cambridge. He holds a diploma in applied mathematics and economics from the University of Augsburg. He joined the EDC in October 2003 and finished his Ph.D. in Change Management and Prediction in 2007. His Ph.D. research in the EDC, under the supervision of Professor John Clarkson and Dr. Claudia Eckert, concentrated on the role of Visualization of design processes and the prediction of change propagation in complex designs. Currently he is working on the NECTISE project as a Research Associate. There, his focus is on how to assess the effects of changes on complex system design. Contact: www-edc.eng.cam.ac.uk | [email protected] Claudia M. Eckert was a Senior Research Associate and a former member of the Engineering Design Centre located within the Department of Engineering at the University of Cambridge. Contact: www-edc.eng.cam.ac.uk | [email protected] Prof. John Clarkson is Director of the Cambridge Engineering Design Centre, located within the Department of Engineering at the University of Cambridge. He also is Professor of Engineering Design and a Fellow of Trinity Hall. Before entering his current position in spring 1995, he worked with PA Consulting Group’s Technology Division where he was Manager of the Advanced Process Group. At PA he gained wide experience of product development with a particular focus on the design of medical equipment and high-integrity systems. His research interests are in the general area of engineering design, particularly the development of design methodologies to address specific design issues, for example, process management, change management, healthcare design and inclusive design. As well as publishing over 400 papers, he has written a number of books on medical equipment design and inclusive design. Contact: www-edc.eng.cam.ac.uk. | [email protected].


4

Making Manufacturing & Supply Chain Management for MCP Work

401


4.1

An Agility Reference Model for the Manufacturing Enterprise: The Example of the Furniture Industry Riadh Azouzi Industrial Chair on Engineered Wood Products for Structural and Appearance Applications, Laval University, Canada Sophie D'Amours Research Consortium in E-Business in the Forest Products Industry, Laval University, Canada Robert Beauregard Industrial Chair on Engineered Wood Products for Structural and Appearance Applications, Laval University, Canada

There is an extensive amount of research literature about the concept of agility, describing its drivers and capabilities, and even suggesting methodologies to develop agility. However, most of these efforts remain vague with respect to the characteristics and the expected contributions of the technologies involved or required. This paper proposes an agility reference model; a unifying conceptual representation of agility in terms of the necessary capabilities needed by every process involved in the enterprise seeking for agility. Agility is described using three capabilities which are believed to be the sources of competitive advantages; flexibility, responsiveness, and autonomy. It is shown that each capability addresses some specific issues and can only be thoroughly developed if the technologies used are characterized with some specific attributes or properties. The idea behind the proposed agility reference model was to derive a typology framework that emphasizes the taxonomy of the market interaction strategies for furniture products, and the competitive priorities that should be targeted by furniture enterprises aiming to be agile. Accordingly, the issues related to the different agility capabilities were discussed in the context of the furniture enterprise of the future. Then, the suitability of the proposed model for the derivation of the typology was explored based on case studies on two furniture manufacturing enterprises. The case studies analyze the context in terms of competitive priorities and customization strategies and investigate the agility properties of the technologies in use.

403

404


Introduction The Quebec furniture industry is an important manufacturing sector of value added products. Up to 60% of its production is exported, 97% of which is exported to the United States (Lihra et al. 2006). One major threat to the furniture industry in Quebec, and in Canada in general, is the fast growing furniture imports to the United Stated from China (Lihra et al. 2005; Schuler and Buehlmann 2003). The Chinese furniture has always been identified as being low cost commoditystyle furniture mass produced using low cost labor and frugal manufacturing approaches (Cao et al. 2004). It is priced up to 40% lower in comparison to similar furniture made in North-America. Recently however, Chinese furniture manufacturers started to break into niche markets and up-market furniture, a market traditionally occupied by Canadian manufacturers. In order to remain competitive, Canadian companies need to innovate. In Québec, this critical situation has led to the creation of the PARIM (PArtenariat de Recherche pour l'Industrie du Meuble)– a partnership between university research chairs/consortium, and private and government research institutions, interrelated with local industries, dedicated to the technological progress of the Canadian furniture industry. At the PARIM, one major project is the "furniture enterprise of the future" project. This project aims at taking hold of the different facets of the furniture enterprise of the future; its business models and the marketing, the networking and the value chain, the competencies, and most importantly, the technologies. The main goal is to pave the way to Canadian enterprises to gain flexibility, responsiveness and efficiency to continuously evolve and adapt to their markets, be innovative and capture new markets. According to Vernadat (1999), this is known as agility. As a matter of facts, agility goes hand in hand with mass customization, a business strategy introduced by Pine (1993) which has been recently associated with the future of the furniture industry in North-America (Lihra et al. 2008; Lihra et al. 2005; Schuler and Buehlmann 2003). The focus of this paper is the technological aspect of the agile enterprise. For furniture firms seeking new ways to compete in the future, it is essential to adopt the right technologies. If the technology is not an enabler for agility, then changes in the enterprise will be much slower, ultimately affecting its competitiveness. Then, the following questions arise: What are the capabilities that should be sought by furniture enterprises? Which properties define agility and how a technology can be an enabler for agility? In reality, these questions are not specific to the furniture manufacturing sector.

405


This paper establishes an agility reference model reflecting the basic agility capabilities that should be displayed by the main processes taking place in the enterprise. Then, the key agility properties of the technologies used to develop these capabilities are defined. Based on this model, a typology framework, integrating the technologies and the manufacturing strategies, is derived. The agility reference model and the typology are based on a thorough literature review and on conceptual reasoning by the authors. Together, they should serve as supporting tools to help managers make strategic decisions in their pursuit of agility. The suitability of the proposed reference model for the derivation of the typology is explored through the case studies of two furniture manufacturing enterprises. The case studies analyze the context in terms of competitive priorities and customization strategies and investigate the agility properties of the technologies in use. The issues related to the different process capabilities and technologies agility properties are discussed in the context of the furniture enterprise of the future. Accordingly, in the next section, we present our view of the furniture enterprise of the future. Then, in Section 2, we briefly review the technical literature on agility and its capabilities. This is followed by a presentation of the agility reference model we propose, and a discussion of three main hypotheses related to this model. The typology derived using the proposed model is described in Section 5. Finally, Section 6 presents two case studies with two furniture companies. Point of customer involvement Organizational strategy Market Interaction strategy Popularizing Adjusting Varietizing Accessorizing Configuring Monitoring Tailoring Collaborating

Retailing

Distribution Packaging Finihsing Assembling

Component Supply Manuf.

Product Engin.

Product Dev. & Design

Material handling Order taking/ coordination Human Resources

Planning & Control Logisitics Order fulfilment realization Information system

Quality Accouting Marketing Management /Finance

Order fulfilment management Post-Order activities

Resources Management

...

Management & support services

Figure 1: Spectrum of activities and market interaction strategies for the furniture enterprise.

406


The Furniture Enterprise of the Future Figure 1 shows the entire spectrum of processes that are typically involved in every furniture manufacturing enterprise. The processes at the strategic level are mainly dealing with the orientation of the company and the manner in which it differentiates itself from the competition. At the operational level are found the different processes that are conducted in order to produce or carry furniture products to customer, from order taking and coordination to order fulfilment realization, going through product development and design, product engineering, and order fulfillment management. These processes can span over several companies. The network in which a given company operates is determined by decisions made at the strategic level. Finally, at the support level, are found all the services that are not directly involved in the delivery of the manufactured products, and all the activities that follow the completion of an order, such as maintenance and warranty claims. The same processes described above will still be needed by the furniture enterprise of the future but with agility properties. Ideally, agility properties should be displayed by every process in the enterprise model depicted by Figure 1. In reality, however, any enterprise would prefer to develop its agility as needed. Since agility is mainly enabled through the introduction of a new technology or process, then the market interaction strategy pursued by a given firm will determine the required technology and the pace at which these new technologies should be introduced. The market interaction strategy is the result of a strategic decision that every enterprise should make. It encompasses the manufacturing approach used, the variety of products produced, and the market served by the enterprise. At this level is specified the position along the customer order fulfillment process at which a product is allocated to a customer. The focus of this paper is on the technological infrastructure needed for the implementation of the market interaction strategies for mass customization. Mass customization begins when products are not made-to-stock. Table 1 depicts an eight options market interaction framework for mass customization, originally proposed by Montreuil and Poulin (2002), then adapted for the furniture sector by Lihra et al. (2005). The primary differentiator between these options is the design process. This suggests that the agility requirements for the design process are translated into demand on other processes (such as the order taking, order fulfilment realization, order fulfilment management processes) for agility. As will be shown later, the design taxonomy will be very useful for the derivation of our typology.

407

VOLUME 1: STRATEGIES AND CONCEPTS Table 1: Market interaction strategies for MC of furniture products. Strategy

Description

Design

Popularizing

A Limited number of products match a wide variety of customer needs, for those who want off-the-shelf products. Focus on evolving the popular product mix in line with evolving customer needs.

Standard

Adjusting

Product adjusted to customer needs after usage. Distributed information systems capture customer feedback.

Modular

Varietizing

Extensive mix of products to satisfy all customer needs. Retailers pick those they want to offer off-the-shelf and rely on quick delivery from distribution network for fast delivery.

Modular

Accessorizing

A limited set of core products matched with a wide array of accessories. Final assembly of accessorized products performed to order either by the customer, the retailer, or the manufacturer.

Modular

Configuring

Customers define the desired product through the setting of parameters and the selection of options. He is guided through the specification process by the retailer or the manufacturer.

Modular

Monitoring

Product is replaced by more adequate product as the customer needs evolve, ensuring continually a best-fit product. This involves regular and interactive customer feedback.

Modular

Tailoring

Product designed/engineered to customer needs. The customer is closely involved in the product realization process.

Custom

Collaboration

Customer is viewed as a collaborator with an open dialogue. Expert field systems interact with customers, seeking to continually optimize customer return.

Custom

Agility and its Capabilities: An Overview There is an extensive amount of research in the literature about the concept of agility, defining it, presenting it as a strategic response to the new economy of the manufacturing world, discussing its components and the different perspectives from which it can be viewed, determining its enablers, and even attempting to measure it (Sanchez and Nagi 2001; Zhang and Sharifi 2000; Ji-Hai et al. 2003; Yusuf et al. 1999; Kidd 1995). However, a review of the related literature reveals that though there has been considerable research on the subject of agility,

408


insufficient attention has been devoted to the development of a comprehensive method for designing and building agile manufacturing solutions, especially with regards to the technological specifications. Furthermore, it is not rare to see some authors confuse agility with flexibility. The latter is another concept that has been regarded as a way to achieve new forms of competitive advantage, and for which there is no predominant definition. In fact, some authors who are more advocate to flexibility, have concentrated their efforts on categorizing various types of flexibility (Slack 2005; Kara et al. 2000). This led to the expansion of its scope to overlap with agility, creating confusion and ambiguity. In fact, flexibility is more and more identified as one among other capabilities of agility. At this point, it is necessary to define agility. The definition given by Christian et al. (2001) for manufacturing agility will be adopted: "The ability to respond to, and create new windows of opportunities in a turbulent market environment driven by individualising customer requirements cost effectively". On an ideal level of agility, one could think of an enterprise in which "Economies of scale have been eliminated so that the cost of production is the same for 10,000 units of one model, as for one unit each of 10,000 different configurations of a single product. As if that was not enough, the customer’s unique needs must be met instantly with a product exactly as desired" (stated by Meredith and Francis 2000). From these statements, it is implicit that what makes agility different from concepts such as flexibility is the set of capabilities an agile system should have in order to adapt rapidly and cost-effectively to allow future unplanned products to be manufactured and sold. The term capability implies the potential of readiness of the manufacturing system to respond to changes (Ramasesh et al. 2001). More than a decade ago, Kidd (1996) collected the keywords linked with the agility concept. He found that the most commonly used are: (1) quickness/speed which is the ability to carry out activity in the shortest possible time; (2) adaptability which is the capability to change direction with ease, for example, to enter completely new markets or product areas; (3) robustness which is the ability to avoid and withstand variations and disturbances, for example, products that lose market appeal owing to changes in customer preferences; (4) virtual corporations which is the ability to combine talents between companies through (short term) joint ventures; (5) reconfiguration which is the ability to very quickly reconfigure enterprise structures, facilities, people, organization and technology to meet unexpected and short lived market opportunities; (6) dynamic teaming which is the ability to actively look for and build creative and innovative talents of other team members; and (7) transformation of knowledge which is the ability to explicitly transform raw ideas into a range of capabilities which are then embodied in both products and services. A


409

more recent normative review of the literature reveals that the so-called agility capabilities consist of four principles (Yusuf et al. 1999; Zhang et al. 2000; Christian et al. 2001): (1) responsiveness which is defined as the ability to identify changes and respond quickly to them, reactively or proactively, and recover from them; (2) competency which is the ability to efficiently and effectively reach enterprises' aims and goals; (3) flexibility/adaptability which is the ability to process different processes and achieve different goals with the same facilities; and (4) quickness/speed. Furthermore, underpinning these four principles is a methodology to integrate them into a coordinated, interdependent system, and to translate them into strategic competitive capabilities (Sharp et al. 1999). In fact, all the above mentioned keywords and principles represent key capabilities that should be considered during the design of each building-block of the enterprise. Each capability can be sought from more than a single area of the enterprise. This, among other things, depends on the mechanisms used to develop the capability. In general, the researchers agree that the agility capabilities must be sought from three major areas of the enterprise: organization, people and technology. However, they are not very specific as to how to develop or assess the capabilities in the organization, people or technology. In particular, with respect to the technologies, no reference can be found in the literature about how the agility capabilities can be perceived in technology. In practice, the decision to adopt a given technology is driven by qualitative judgements and discussions within the entrepreneurial and management team (Rangone 1998). If a formal model or a strategic decision approach is used in the selection of the best technology, then tangible factors such as cost or time saving are considered but the intangible benefits (in terms of capabilities such as flexibility, responsiveness or competency) that can result from the technology are overlooked (Ragavan and Punniyamoorthy 2003). Technologies, similarly as products, are characterized by their specifications. These are derived from the underlying processes. Specifications may be too much to provide an explanation of what causes or hinders agility capabilities in the enterprise. In fact, there is a need to constrain these specifications to a good sound set of concepts (elementary and very specific and concise attributes) that most people would agree upon and that would support enterprise agility. The authors argue that it is necessary to develop a framework that can be used in order to assess the agility properties in any technology. This would be very helpful in order to determine if a given technology is appropriate for or compliant with the agility capabilities sought by an enterprise. The reference model presented in the next section is meant to be a step forward in this research direction. Under its current form, the proposed model is expected to help understanding the relation-

410


ships among the technologies, and between the technologies and the agility capabilities. A more formal form of this model will be explored in the near future. Agility Reference Model The reference model proposed in this section is a framework for the development of consistent standards supporting agility as far as technology is concerned. It provides a unifying conceptual representation of agility in terms of the required agility properties of the technologies needed by the agile system. Agility is described in Figure 2 using three capabilities which are believed to be the sources of competitive advantage in the furniture industry. These are flexibility, autonomy, and responsiveness. In Figure 2, the agility level of a given process is represented as a 3-dimensional function.

Figure 2: Agility reference model.

For each dimension is given one simple, concise and clear definition. This is in spite of the different aspects under which the flexibility, autonomy and responsiveness of every process in the enterprise might appear.

Flexibility. A process is flexible if it is able to out live the products for which it was originally built.


411

Responsiveness. A process is responsive if it operates effectively and efficiently in a timely manner.

Autonomy. A process is autonomous if it is self-sufficient and able to operate by its own rules with no need for on-line assistance or direct control by an operator or manager.

Assuming the definitions given above, three hypotheses are stated to be analyzed: Hypothesis 1: The technologies in a highly flexible process are fully computerized with large information content. It is assumed that the overall flexibility of an enterprise is determined by the flexibility of its processes. From the definition of flexibility given previously, it is possible to recognize flexibility through the observation of the range of products or parts an enterprise can deliver to its customers. Then, the more its environment is uncertain and its offering in terms of products, parts or services is variable, the more the enterprise needs to improve the flexibility of its processes. In fact, variability and uncertainty have been identified as the factors generating the need for flexibility in manufacturing since many years ago (Correa and Slack 1996; Kara and Kayis 2004). However, most of these studies did not provide clear ideas about how to face these factors. From a theoretical point of view, variability represents the diversity in a population of parameters, and can be reduced by a more detailed model formulation. On the other hand, uncertainty is due to the lack of knowledge about the system’s environment, and can be reduced through improved measurements and model formulation. In both cases, it is essential to have a computerized technology. Then, an increasing information content of processes guarantees a cost efficient and individualized production (Piller 2004). Notice that information content can be enhanced through developments in electronics and software. It is recognized that in all industry sectors, computer technology (hardware and software) has had the largest impact on enterprise flexibility. As we move forward in the implementation of computerized technologies, emphasis is placed more on specialized software than on specialized hardware. Most of the systems for manufacturing automation implemented today make use of computers. Although automation preceded computer technology, it is computer power that made automation effectively flexible in applications other than mass production. The trend today is towards the computerization of all other processes or activities taking place in the enterprise. The integration of the different computerized and information rich processes will result in completely new prospects for enterprise flexibility.

412


Hypothesis 2: The technologies in a highly responsive process are important sources of standardization and time compression. In this paper, responsiveness is taken to be process-related. Every process in the enterprise should be designed, configured and linked to other processes in a way to permit high speed of reaction and maximum effectiveness as demand changes unexpectedly. Keeping in mind that the responsive process needs also to be computerized and rich with information, then the only way to conciliate these requirements is to focus on standardizing and time compressing technologies. Indeed, standardization has been identified as fundamental for enterprises in transition to agility (Anderson 2004). Probably, its main benefit is complexity reduction. This can be linked to the adoption of a modularized-design approach, standardized raw material, and/or uniform information and knowledge management. For instance, in the furniture industry, it is possible nowadays to find a number of software systems that include the tools to enable the user to rapidly select product modules and options from pre-configured libraries or to create new modules and options and add them to existing libraries. These software systems generally automate the design of modular product by integrating tools for CAD (ComputerAided-Design), CSM (Component & Supplier Management), and DFM/A (Design for Manufacturability/Assembly) as well as for CAM (Computer-Aided Manufacturing). With such tools, the user can create the design libraries that meet his design principles and checklists, and the constraints specified by his suppliers (e.g. for hardware such as hinges, knobs and pulls). Then, CAM tools can be utilized to generate the necessary instructions to drive automated machine tools (by generating CNC codes), which can be path-traced using CAD tools. In addition, these software systems assemble the necessary data to support the manufacturing operations; they dynamically generate the bills of materials and parts drawings and, in most cases, will integrate seamlessly with other modules for quotation creation, order entry, planning and scheduling, shop floor execution or reporting. Many of the above-mentioned concepts became conceivable in some furniture industrial sectors (e.g. case-goods and office furniture) after that the main raw material was standardized under the form of wood-based composite panels. Using automated panel-processing equipment, new products could be manufactured with no additional setup, anchoring mechanisms, or cutting tools. Hypothesis 3: The technologies in a highly autonomous process are valuable sources of awareness and knowledge engineering and management capabilities.


413

It is clear that a flexible and responsive system cannot maintain high centralization; the circuits of information and decision making required to be very short. This is where the notion of autonomy intervenes. Autonomy as defined previously is not ignorance of the critical interdependencies between the different processes or between the different technologies used in the same process. In fact, the implementation of flexibility (using computerized technologies and technologies for increasing information content) and responsiveness (using standardization and lead compressing technologies) drives production processes towards more tightly coupled systems where processes are more and more interdependent. This leads to a greater need for coordination in decision making. Such coordination is impossible without awareness capabilities (Klein 1991). Awareness, as defined by Kritchanchai and MacCarthy (1999) is the process knowledge and recognition of the stimuli that occur or may occur, and the preparation and responses necessary to address them, whether they emanate from customer needs, environmental uncertainties, competitors or market conditions. The stimuli could be any internal or external changes that can be perceived by the process as relevant to its operation. Moreover, if these stimuli can be proactively identified by the process, the latter is then provided with more time to gather data or information and react accordingly to the problem or opportunity that will present itself. Such behavior cannot be considered if the process is not provided with the necessary means for knowledge engineering and management. The latter refers to the tools for storing and communicating information or data and the tools oriented towards knowledge, supporting activities such as knowledge creation, mapping, retrieval, and use activities (Shadbolt and Milton 1999). Typology Framework As it was discussed previously, product design appears to be a key factor as to fulfill the agility requirements in furniture enterprises striving to increase their competitiveness and their reactivity. The existence of three product design modes (standard, modular, or custom design) (Table 1) suggests the existence of three corresponding agility levels. Each agility level corresponds to unique patterns of process capabilities, and thus, to unique patterns of agility properties applicable for the set of technologies needed throughout the enterprise. As shown in Figure 3, the necessary/sufficient/competitive agility is linked with the minimal/average/ high levels for the capabilities of flexibility, responsiveness, and autonomy that an enterprise needs to develop throughout its processes in order to achieve some competitive priorities. Lead time, quality, variety, innovativeness and profit margin have been identified as the main priorities (Table 2).

414


Figure 3: Derivation of the typology.

Assuming that the agility properties of the technologies used for any given process in the enterprise can be assessed using the agility reference model, then the proposed typology can be thought of as a mean to visualize the synergetic effect that any set of technologies could have on the relevant competitive priorities. The three agility levels were defined as follows: Table 2: Competitive priorities for furniture enterprises. Strategy

Description

Lead time

The time required to process a customer request for a product, from customer preferences elicitation to product deliver going through order taking.

Quality

The quality or performance standards of the delivered product.

Variety

The number of products or product configurations that can be delivered.

Innovativeness

The ability to introduce new products and processes

Profit margin

The ability to make a gain out of the delivered product unit.

Necessary agility: This level supports furniture products that are produced in high unit volumes and narrow variety, and for which short delivery time is critical. The design process should be flexible enough to allow a small number of different models to be designed. The autonomy level is delimited by the designer’s market knowledge (trends and new fashions). Since the introduction of new products or styles comes at a relatively low pace, then it is acceptable that the design process takes enough time so that the needed responsiveness can be qualified as low.


415

Sufficient agility: This level supports furniture products of modular design. Modularity has been conceptualized by Bourke and Kempfer (1999) as follows: "Modularity is a concept of building smaller sub-systems, designed independently, and able to function properly when assembled and tested as an end-item." It is emphasized here that not only is the design process not linked to a specific customer order, it goes hand in hand with engineering. The designer is constrained to develop standardized and interchangeable modules while observing a complex design layout and maximum reusability. Then the users can configure a range of possible variants of the pre-designed modules (Hermansky and SeelmannEggebert 2003). Variants can take the form of different colors, surface materials, lengths, widths, depths, etc. The finished product will be unique and can be delivered in short period of time. Competitive agility: A competitively agile enterprise is one that has successfully configured and devoted its manufacturing resources in terms of technology, people, and organization so that it remains innovative and able to deliver a large variety of good quality products, in short lead times, while being profitable. Ideally, the customer should be directly involved in the design phase. The salesperson and the customer might need to go through laborious back-and-forth iterations before an agreement is reached. The former needs to elicit the customer’s preferences and personal needs, and in the mean time, to provide the customer with the estimated overall cost and delivery date. The deal is more likely to be closed if the salesperson can allow the consumer to examine the product and evaluate the overview. Within this typology, it is assumed that the agility requirements for the design process are translated into demand on other processes for agility (such as the order taking, order fulfillment realization, order fulfillment management processes). This implies that all the processes in the enterprise should be characterized with more or less equivalent flexibility, responsiveness, and autonomy capabilities. A direct consequence of this capabilities equivalence is that, in an ideal world, all the processes in a "necessarily agile" enterprise should be configured technologically so that the resulting capabilities are as low as for the design process. Thereby, the capabilities of the processes in a "sufficiently agile" enterprise should be at a substantial step above the levels required from the processes of a "necessarily agile" enterprise. Finally, all the processes in a "competitively agile" enterprise should be ideally highly flexible, highly autonomous, and highly responsive.

416


Case Studies The following cases studies were designed based on the case study design methodology proposed by Yin (1994), and covered the five core processes at the operational level in the furniture enterprises; order taking/retailing, design, engineering, order fulfillment management, and order fulfillment realization (refer to Figure 1). Data was collected using a structured questionnaire that contained three main sections. Basically, the respondent was asked to give his appreciation of (1) the targeted and achieved competitive priorities, (2) how developed are the agility’s capabilities of the processes, and (3) how developed are the properties of the technologies. A ten-point scale (with 0=not at all important, to 3=ideal level of importance for a manufacturer of standard products, to 6=ideal level of importance for a manufacturer of modular products, to 9=ideal level of importance for a manufacturer of custom products) was used. The companies in the following two case studies are located in the province of Quebec, eastern Canada. A summary of their main characteristics is presented in Table 3. The results are depicted in Figs. 4, 5 and 6. Notice that the names of the enterprises are fictitious to preserve confidentiality. Table 3: Case studies results. HouseCo

OfficeCo Main characteristics

Solid wood household furniture

Office furniture systems

Number of employees

120

927

Year started interest in customization

Since 20 years

Since more than 35 years

Main Products

Design mode(s) Products of standard design

100%

0%

Products of modular design

0%

90%

Products of custom design

0%

10% Market interaction strategies

Strategies in place

Varietizing Accessorizing Tailoring

Configuring Monitoring Tailoring


417

Case 1: A contemporary furniture manufacturer The enterprise, referred to here as "HouseCo" manufactures household furniture systems mostly made of solid wood. It is a family owned enterprise, which includes two sites, a manufacturing site and a rough mill. Due to significant reduction in last year’s sales, the enterprise had incurred, within the last 8 months, nearly 50% reduction in work force, passing from 250 people down to 120 people, 80 of which are machine operators and craftsman. A team of four people is responsible for designing and prototyping new products. HouseCo foregrounds the quality and the styling of its hardwood and veneered wood constructed furniture. During the last 10 years, the company invested heavily in advanced technologies, including sophisticated computer-assisted equipment. The enterprise uses advanced technologies to control its machining and sanding processes, but relies on traditional cabinet-making techniques for furniture assembly and finishing. The case individual was the vice-president of operations, a shareholder who has been overseeing production for more than 15 years.

Figure 4: Comparison of target, achieved, and ideal competitive priorities.

Since its beginning, HouseCo has been producing standard products only. Early in the 1980s, HouseCo started however to offer more variety with a more designoriented approach and world-class outlook. Nowadays, its catalogue features a number of product lines or collections including over 300 different products that can be offered in 5 different colors. Also, accessories such as knobs and hinges can be varied in compliance with the client preferences. Note that, while its promotional strategies (via advertisement and participation in major tradeshows)

418


target both potential retail distributors and potential consumers, HouseCo does not have channels for direct sales. The customer could buy products only through pre-approved retail distributors (including conventional furniture retailers, specialty stores, and mass merchants). Large retail distributors are involved in the development of a new product.

Figure 5: Actual vs. targeted vs. ideal agility properties of the technologies.

As shown in Figure 4(a), the five competitive priorities appeared to be very important for HouseCo with special emphasis on the lead-time, profit margin and innovativeness. Yet while the priorities of innovativeness, quality and variety appeared to be fairly well achieved, concern about the profit margin and lead-time was very deep. In fact, the situation became alarming when the sales of the enterprise went down by nearly 50% during the last year. The enterprise used to have a strong distribution network enabling it to reach high-end markets.


419

However, the increasing value of the Canadian dollar weakened the distribution network of HouseCo, which among other things was accentuated by the bankruptcy of some of its important distributors. Significant market share, mainly in Canada and the US, was lost to competitors from low labor cost countries such as China making similar styles or even products imitation.

Figure 6: Estimated actual capacities of the processes.

On the other hand, HouseCo had established for itself competitive priorities that, according to the typology, are astonishingly close to the competitive priorities that should be targeted by the "ideal" enterprise making modularized products but certainly not standard products. Under such ambitious manufacturing strategy, each of the manufacturing processes at HouseCo was expected to present an agility level higher than typically found for such processes at an enterprise manufacturing only standard products. From Figure 6(a), it is obvious however that the targeted strategic goals (or competitive priorities) were not matched with the necessary processes capabilities of flexibility, responsiveness, and autonomy. Figure 6(a) shows also that the capabilities of the design and engineering processes appear to go beyond the ideal levels for a manufacturer of standard products, yet the capabilities of the other processes could barely go with the ideal capabilities for such a company, as established theoretically by the typology. The translation of these capabilities in terms of the agility properties of the technologies should be visible from Figure 5. It shows the deviations between the agility properties of the technologies at HouseCo and their ideal counterparts. It can be easily seen that the technological infrastructure at HouseCo could not support its ambitious strategic goals. Probably, the best technologies are employed in the design and engineering processes as well as the order management process. The

420


agility properties of computerization and information content of these technologies exceed the ideal levels for standard products. This may be in agreement with the established strategic goals; however the technologies employed in the other processes are not sufficiently flexible; in other words, the technologies used in these processes are not sufficiently computerized and their information contents are very limited. Nonetheless, the standardization, lead-time compression, engineering and management knowledge, and awareness properties for all the technologies employed at HouseCo rate poorly in general. In conclusion, it appears as if HouseCo was rooted in the "necessary agility" type (since it produces exclusively standard products and had a technological infrastructure that hardly enables agility in most of its processes), it was competing more with enterprises making “styled modular furniture” than with enterprises making furniture of standard design. Case 2: An office systems furniture manufacturer The enterprise, referred to here as "OfficeCo" manufactures office furniture systems, mostly made of composite wood based panels. Although it is part of a larger corporate since the last eight years, OfficeCo is still operating as a single strategic business unit with its own competitors and a manager accountable for operations. It currently employs over 900 people distributed over five sites (four factories and one consolidation center) located within 70 Km from each other. OfficeCo promotes itself as an enterprise that is willing to take the extra step to provide the highest quality service and complete customer satisfaction. In order to gain a competitive edge over competition, OfficeCo had during the last six years made important investments in new manufacturing equipment and software systems. It created a central group for new products developments which employs more than 40 staff. The head of this group, an engineer, was designated to be the case individual. He has been with OfficeCo for the last nine years and started as a factory director. When asked about when his enterprise started to take interest in mass customization, the case individual swiftly answered "more than 35 years!" In fact, he wanted to stress on the fact that office furniture is of modular-design by nature (with its movable and reconfigurable components), and very often tailored to specific customer needs. Within 10 days lead-time, OfficeCo can deliver a wide variety of products (including desks, panel systems, filing and storage systems, etc…). Its on-line catalog featured some planning tools that provide customers with access to a wide variety of layouts by solution, price, size, and product. Like in the previous case, OfficeCo did not have channels for direct sales. In fact, its merger with the


421

large corporation gave OfficeCo access to showrooms located all over the word, and to a large distribution network. The name of this corporation was strongly promoted. The distributor could be closely involved with the customer in configuring and/or tailoring the product solution to the customer’s needs. Figure 4(b) shows that the five competitive priorities appeared to be very important for OfficeCo, with the quality and innovativeness priorities even exceeding the levels set up for the "ideal" enterprise making modularized products. All the priorities appeared to be achieved quite well. In particular, the resulting variety and quality surpass the goals of the enterprise, which themselves surpass ideal levels (for a company making modular products). As a matter of fact, this was in perfect agreement with the image that the enterprise was trying to uphold or to promote. And it was succeeding pretty well. Indeed, OfficeCo has known growth rates of near 20% per year during the last five years. During this same period of time, the number of employees more than doubled, not to mention all the investment the enterprise did in its technology infrastructure and in research and development (The enterprise was indeed very active in its partnership program with universities and other research institutions). The performance reported above can be explained to some extent using the processes capabilities portrayed by Figure 6(b). This figure reveals that the capability disparities among the processes were not considerable. The design & engineering, and order fulfilment realization processes were particularly flexible. This may result from the focus at OfficeCo on the competitive priorities of variety and quality. This is clearly corroborated by the particularly high level technology properties of computerization and information content at each of the two processes (see Figs. 5(a), 5(d)). According to Figs. 5(b) and 5(c), the other two processes (order fulfilment management, and retailing & order taking) should also be sufficiently flexible (as their technologies appear to have computerization and information content levels equal or even exceeding the ideal levels needed by an enterprise producing modular products). However, Figure 6(b) does not confirm this presumption. In fact, according to the respondent, the flexibility of each of these two processes was much lower that the ideal level, an inconsistency that could be attributed to the fact that the measures were exclusively perceptual. Discussion In order to replicate as much as possible the theoretical ideal types stated by the typology, it was preferable if a third case study could be carried out. Several constraints prevented this from happening on time. Yet, through the cases studied, the authors can confidently assert that it was possible to corroborate the main lines

422


of the conceptual reasoning they have formulated. In fact, the absolute validity of the three hypotheses stated about the agility reference model will always remain a matter of investigation. Nevertheless, from the cases studied, it was possible to explain to some extend the observed capabilities of the processes using the properties of the technologies. In general, it was obvious that performance is maximized if the customization strategy put in practice is in conformity with the available technology. More evidence is demonstrated by the following. It was not necessary to have the processes capabilities estimated by the case individuals. This is because these capabilities could have been deduced from the estimates provided for the technologies properties depicted in Figure 5. In fact, the two sets of estimates (direct and indirect) were used for cross-validation purposes. It was easy to note that the two sources of data were concurrent, thus improving the authors assurance regarding the agility reference model. When asked about the possible actions their enterprises are willing to undertake as far as technology is concerned, the respondents gave answers that were verifiable using the spider graphs of Figure 5. Both individuals stressed on the same issue: compressing the throughput time, however their expectation were slightly different. While OfficeCo was expecting better service to customers, HouseCo, on its part, was expecting to see its profit margins raised. Typically, throughput time corresponds to the period of time required to perform all the operations necessary to complete an order, including the transport, queue, setup, and processing times. The most efficient way to compress throughput time consists in improving the standardization, integration, and control of the complete process of bringing a new product to market. The respondent from OfficeCo was very clear. His enterprise was looking to invest in information technology, especially, downstream in its supply chain. OfficeCo wants to improve the flexibility in its IT links and connections (software and hardware). It intends to offer a wider variety of information to end users (e.g., multimedia and order tracking), On the other hand, the respondent from HouseCo was not clear about the specific actions his company intends to undertake, yet he recognized that the standardization of their activities is a major concern. With the results shown in Figs. 6 and 5, it was possible to corroborate these future actions or investment directions. Information technology, with all its ramifications, is known to be directly linked to the autonomy of the processes which appeared to be a big drawback with almost all the processes in the two enterprises. On the other hand, the standardization of the technologies in place especially at HouseCo rated very low. Finally, it is worthy to mention both respondents had in general very positive and encouraging comments with regarding the agility reference


423

model and the typology framework. On their part, the authors acknowledge that the lack of triangulation of multiple data sources for more confidence about the results or the typology framework itself remains a limitation in this study. Conclusions & Future Work This paper introduced agility as a strategic response for Canadian furniture enterprises seeking to adapt to their markets and be innovative, focusing on technology. It proposed a reference model reflecting the key properties that should characterize the technologies needed by the different processes or activities taking place in the furniture enterprise of the future. Basically, this model suggests that, for any given process to have the best contribution to the overall agility of the enterprise, it needs to have the capabilities of flexibility, responsiveness, and autonomy, each at its maximum level. Also, the proposed model identifies the agility properties that should be displayed by the technologies used to implement the needed capabilities. The proposed model played a key role in the derivation of a typology linking the manufacturing strategy of the enterprise with its technology usage. This typology focused on the strategic and technological characteristics for enterprise competitiveness and did not explicitly include performance measures and methods in the model. The appropriateness of the proposed framework including the agility reference model and the typology was explored based on two case studies conducted in two furniture enterprises. The two cases replicated to a certain extent two out of the three theoretical ideal types stated by the typology, and they were very effective in corroborating the main lines of the conceptual reasoning by the authors. Each time, it was possible to position the studied company against the proposed framework, and to identify directions and options for future development work regarding the technologies. Future work is being undertaken to complete additional case studies. Also, a more formal form of the reference model will be explored in the near future.

References Anderson, D. M. (2004). Build-to-Order & mass customization, the ultimate supply chain and lean manufacturing strategy for low-cost on-demand production without forecasts or inventory. Cambria: CIM Press, 1–520. Bourke, R. and Kempfer, L. (1999). Achieving success with mass customization: the vital contributions of engineering. Computer-Aided Engineering. 18(10): 42–52. Cao, X., Hansen, E. N., Xu, M. and Xu, B. (2004). China’s furniture industry today. Forest Products Journal. 54(11): 14–23.

424


Christian, I., Ismail, H., Mooney, J., Snowden, S., Toward, M. and Zhang, D. (2001). Agile manufacturing transitional strategies. 4th Stimulating Manufacturing Excellence in Small and Medium Enterprises (SMESME) International Conference Aalborg University: Denmark (May 14–16). Correa, H.L. and Slack, N. (1996). Framework to analyse flexibility and unplanned change in manufacturing systems. Computer Integrated Manufacturing Systems. 9(1): 57–64. Hermansky, J. and Seelmanaa-Eggebert, R. (2003). Manufacturing Postponed. IEE Manufacturing Engineer. 82(4): 38–41. Ji-Hai, L., Anderson, A. R. and Harrison, R. T. (2003). The evolution of agile manufacturing. Business Process Management Journal. 9(2): 170–189. Kara, S. and Kayis, B. (2004). Manufacturing flexibility and variability: an overview. Journal of Manufacturing Technology Management. 15(6): 466–478. Kara S., Kayis B. and O'Kane S. (2000). The role of human factors in flexibility management: A survey. Human Factors and Ergonomics in Manufacturing. 12(1): 75–119. Kidd, P. T. (1996). Agile manufacturing: a strategy for the 21st century. IEE Colloquium on Agile Manufacturing. 1–6. Klein, Janice A. (1991). A re-examination of autonomy in light of new manufacturing practices. Human Relations. 44(1): 21–38. Kritchanchai, D. and MacCarthy, B. L. (1999). Responsiveness of the order fulfilment process. International Journal of Operations & Production Management. 19(8): 812–833. Lihra, T., Buehlmann, U. and Beauregard, R. (2008). Mass customization of wood furniture as a competitive strategy. International Journal of Mass Customization. 2(34): 200–215. Lihra, T., Beauregard, R., D'Amours, S., Dessureault, Y., Lagacé, D. and Blanchette, J. (2006). Connecting research to the furniture industry: PARIM. Forest Products Society 60th International Conference. Newport Beach: California (June 25–28). Lihra, T., Buehlmann, U. and Beauregard, R. (2005). Mass customization of wood furniture: Literature review and application potential. Proceedings of the 3rd Interdisciplinary World Congress on Mass Customization and Personalization. Hong Kong. Meredith, S. and Francis, D. (2000). Journey towards agility: the agile wheel explored. The TQM Magazine. 12(2): 137–143. Montreuil, B. and Poulin, M. (2002). Demand and supply network design scope for personalized manufacturing. International Journal of Production Planning & Control. 16(5): 454–469. Piller, F. (2004). Mass Customization: Reflections on the State of the Concept. International Journal of Flexible Manufacturing Systems. 16(4): 313–334. Pine II, B. J. (1993). Mass Customization: The new frontier in business competition. Boston: Harvard Business School Press. Ragavan, P., Punniyamoorthy, M, (2003). A Strategic Decision Model for the Justification of Technology Selection. The International Journal of Advanced Manufacturing Technology. 21(1): 72–78. Ramasesh, R., Kulkarni, S. and Jayakumar, M. (2001). Agility in manufacturing systems: an exploratory modeling framework and simulation. Integrated manufacturing systems. 12(7): 534–548. Rangone, A. (1998). On the applicability of analytical techniques for the selection of AMTs in smallmedium sized firms. Small Business Economics. 10(3): 293–304. Sanchez, Luis M. and Nagi, R. (2001). A review of agile manufacturing systems. International Journal of Production Research. 39(16): 3561–3600.


425

Schuler, A. and Buehlmann, U. (2003). Identifying future competitive business strategies for the US residential wood furniture industry: Benchmarking and paradigm shifts. USDA Forest Service General Technical Report GTR-NE-30415, 1–17. Sharp, J. M., Irani, Z. and Desai, S. (1999). Working towards agile manufacturing in the UK industry. International Journal of Production Economics. 62(1-2): 155–169. Slack, N. (2005). The flexibility of manufacturing systems. International Journal of Operations & Production Management. 25(12): 1190–1200. Shadbolt, N. and Milton, N. (1999). From knowledge engineering to knowledge management. British Journal of Management. 10(4): 309–322. Vernadat, F. B. (999). Research agenda for agile manufacturing. International Journal of Agile Management Systems. 1(1): 37–40. Yin, R. K. (1994). Case Study Research: Design and Methods. Thousand Oaks: Sage. Yusuf, Y. Y., Sarhadi, M. and Gunasekaran, A. (1999). Agile manufacturing: The drivers, concepts and attributes. International Journal of Production Economics. 62(1-2): 33–43. Zhang, Z. and Sharifi, H. (2000). A methodology for achieving agility in manufacturing organizations. International Journal of Operations & Production Management. 20(4): 496–512.

Author Biographies Riadh Azouzi (Ph.D., Université Laval, Canada) is a research associate at the Université Laval affiliated with the Industrial research chair on engineered wood products for structural and appearance applications (CIBISA) and the Research consortium in ebusiness in the forest products industry (FOR@C). In 1997-2005, he worked for Baan, Invensys plc then SSA Global, leaders in the provision of manufacturing solutions. He was developing smart solution techniques that allow complete control of manufacturing systems through the integration of aspects of design, planning manufacturing, distribution, and management. His research interests include enterprise agility and the enterprise of the future. Contact: www.forac.ulaval.ca | [email protected] Sophie D'Amours (Ph.D. École Polytechnique de Montréal, MBA Université Lava, Canada) is a professor at the mechanical engineering department of Université Laval. She holds a Canada Research Chair in planning value creating network. She is also the general director of the FOR@C Research Consortium dedicated to supply chain issues within the forest product industry. She has been intensively involved in many applied research projects studying agent-based advanced planning and scheduling, integrating simulation with operational research, as well as studying collaborative planning and emerging business models. Contact: www.forac.ulaval.ca | [email protected] Dr Robert Beauregard is dean of the Faculté de foresterie et de géomatique from the Université Laval. He was between 2003 and 2008 holder of the Industrial Research Chair on Engineered Wood Products for Structural and Appearance applications (CIBISA). His area of expertise is the modeling of manufacturing systems for the forest industries. He

426


develops comprehensive approaches to the design of business models taking into account the interactions between the wood resource, process development and innovative products for better business performance. In 1995-97, he was research scientist with the New Zealand Forest Research Institute. From 1997 to 2000, he has been at the Eastern Laboratory of Forintek Canada Corp. where he was instrumental in the creation of the Department for Value Added Wood Products.

4.2

Overcoming Configuration Process Complexity of Highly Customizable Components Erik Oestreich Technische Universität Chemnitz, Germany Tobias Teich Chair of Computer Science in Economics, Westsächssiche Hochschule Zwickau, Germany

The benefits of product configuration systems are twofold. On the one hand side, they support the creation and management of configuration knowledge. On the other hand side, configurators are tools that enable a company to use its expertise for best possible fulfillment of the customers. A basic requirement for using a product configuration system is the existence of a complete product model. In the area of material customization of standard components, this condition can often not be fulfilled. For this reason, it is impossible to use conventional configuration systems to support the information elicitation process. This paper presents a novel idea for product configuration that can be applied to overcome complexity in the area of material customization. In the centre of interest is a flexible configuration model that supports the generation of free definable descriptions and configuration dialogues, which can be used to identify individual components instead of using unique item numbers. The paper will also consider the fact that in many situations there is also an additional demand for standard parts, needed for the final assembling of an individual component. In order to support the procurement processes of such parts, it is necessary to transfer the contents of the configuration model directly into a bill of materials. Therefore this paper presents the basic requirements as well as an appropriate algorithm to accomplish the transformation process.

Introduction The automotive industry is characterized by a fierce competition, caused by saturated and stagnating markets. These factors, in combination with the expansion of new Asian car makers (KPMG 2005) lead to a steady increase of production overcapacities around the world (Becker 2007). Responding to the needs of customers, manufacturers are aiming for mass customization (Pine 1993; Davis 1987; Tseng and Jiao 2001; Piller 2003) in order to be able to fulfil nearly every customer wish with near mass production efficiency (Tseng and Jiao 2001). Because information can be regarded as one of the most important factors (Piller 427

428


et al. 2000; Blattberg and Glazer 1994; Da Silveira et al. 2001), a successful implementation of a mass customization strategy in practice depends mainly on the ability to handle a huge amount of information by the use of suitable tools (Blecker et al. 2005). In this context, product configurators play an important role. As stated by Bourke (2000), configurators can be considered to be "key enablers" for mass customization. Caused by competition, car makers are forced to find new ways to differentiate their products from those of their competitors. While most of the Western European car makers have already put mass customization into practice by offering a lot of selectable optional features, some of them try to offer even more exclusive and individual packages (Hamprecht 2006). In contrast to the traditional optional equipment, these packages offer the customer the opportunity to determine many more characteristics of a product according to his personal needs (Tiihonen and Soininen 1998). With this kind of customization, car makers try to extend their often limited offers with respect to color, trim and materials to avoid the creation of singular visions of their products (Chin 2005). Furthermore, customers are also often willing to pay a premium price to get exactly the product they want (Bermann 2002; Broekhuizen and Alsem 2002; Piller et al. 2004). Another advantage is the reduction of inventory of work-in-progress and finished goods (Tiihonen and Soininen 1998; Zipkin 2001). In practice, a prerequisite for an effective implementation of such an offer is the ability to deal with the enormous internal and external complexity (Blecker et al. 2005) caused by the huge number of available product variants. While offering conventional optional equipment already leads to a high number of possible variants of a car (Rosenberg 1996), offering individual packages leads to a nearly infinite number of variants for one single component. Especially in the area of material customization of cars it is therefore often necessary to create additional documents, that describe the individual items in a very detailed way instead of using unique item numbers that clearly identify the needed parts. The reason for this procedure is the almost unlimited diversity of variations that exists only for one individual component (e.g. an individual cover of a head rest). Following the ideas of Kneppelt (1984) "… where multiple or complex operations are required to integrate the options for a unique product configuration, some creative generation of shop documentation is necessary", Vollmann et al. (1988) "… the industrial engineer needs other means to say how the dashboard is to be assembled and from what components" and Günthner et al. (2006), it is necessary to develop methods to deal with product complexity. Suitable instruments, which can be used to describe individual components are construction notifications. The generation


429

of such documents can be carried out nearly without any user interaction by a specific configuration system. In addition to individual components, in many cases standard parts are needed for the assembly of the final package. The demand for these standard parts depends closely on the content of a certain construction notification. For this reason it is necessary to transform a given construction notification into a bill of materials in order to support the procurement processes for standard parts. The remainder of this paper is organized as follows. Section 2 gives a brief overview on product variety and complexity in the automotive industry. Furthermore, this section presents a simple example to outline the meaning of the term individualization as it is used in this work. In this context it is shown also, how construction notifications can be used to overcome product complexity. Section 3 presents the basic idea behind the configuration model. It shows how individual components can be identified uniquely by the use of flexible document structures. Furthermore, Section 3 presents an example of a reference implementation of the configuration model. Section 4 presents the process that can be applied to transform a construction notification into a bill of materials. Therefore, this section firstly describes the main difference between the configuration model used to draw up construction notification versus the technical product model used to describe bills of materials. Secondly, two formal models are presented to describe variant or generic bills of materials, as well as order specific bills of materials. Finally, Section 4 shows a basic algorithm which can be used to perform the transformation process. Section 5 concludes this paper. Mass Customization Strategies in the Automotive Industry Among other classifications of mass customization strategies (Lampel and Mintzberg 1996; Duray et al. 2000; MacCarthy et al. 2003), Alford et al. (2000) put forward three distinctive strategies of customization specifically in the context of the automotive industry, namely core, optional and form customization.

Core customization means involving the customer in the design process of the car. This approach is connected with high additional costs and is only available for low-volume models. The basic idea behind mass customization — offering customized products at prices comparable with standard products (Tseng and Jiao 2001) — is missed.

Optional customization means integrating the customer into the manufacturing process. First of all, a basic model and the color are chosen by the customer. It is impossible to change the design in any way. In the next step the

430


car can be adapted to the demands and needs of a customer by selecting options of features offered by the manufacturers at a price premium. At the moment, optional customization can be considered as a common, costeffective approach to offer individuality in a high-volume manufacturing environment.

Form customization can be applied to adapt to the ordered car according to individual customer needs in cooperation with a distributor. This includes adding new parts to the car or changing existing standard parts. Additionally, special services can be offered by the distributor to customize the sales process.

The possibilities to change the design of some parts by the use of form customization in cooperation with the distributor are limited. But there is also a looming trend towards even more individualized cars (Hamprecht 2006). This trend is focused on the design of certain components e.g. an individual leather seat upholstery where a customer can select exclusive materials and colors out of a provided range of options. Especially western European premium car makers picked up this trend and offer special individual packages. In contrast to optional customization where only a predefined package is selected, the customer is now able to determine additional characteristics of the chosen package. On that ground, this trend can be considered as a combination of optional and form customization. Order specification The configuration of a car is carried out on the level of order codes (Sinz 2006; Haag 1998; Herlyn 1990). Each option represents a feature that a customer is able to choose. Order codes that refer to mutual exclusive options are organized in order code groups (order code families). Basically, a customer is only able to select one order code of each group during the configuration process. Order code groups can be divided into two different classes: mandatory and optional groups. While most of the options in the mandatory groups are determined automatically by the selection of an appropriate car model (e.g. engine, drive train), optional features can be used to adapt the car according to the wishes of an individual customer. Figure 1 illustrates a small part of a configuration model of a car, using the notation principles of the Marrakesch data model (Hümmer et al. 2004; Dietrich et al. 2004). The figure shows that there could be multiple relationships between options of different groups. The available types of steering wheels depend for example on the kind of the selected drive train of the car.

431

VOLUME 1: STRATEGIES AND CONCEPTS CAR

TRANSMISSION

GEAR BOX

AUTOMATIC

AND

STEERING WHEEL

3-SPOKE + SHIFT PADDELS

OR

COLOR

3-SPOKE LEATHER

SILVER

BLACK

4-SPOKE + SHIFT PADDELS OR

AND

Figure 1: Using order codes to represent options and dependencies.

Product variety in the automotive industry Offering a high number of optional equipment in combination with different exterior and interior colors, as well as different model types leads to a huge number of possible final product variants. This fact is illustrated by Rosenberg (Rosenberg 1996) by a simple example. Only through the combination of a small number of available features it is possible to create more than 8.9 billion different cars. The result of this example can also be validated in practice. In 2004, Daimler Chrysler produced about 1.1 million Mercedes A class at the production plant in Raststatt. Only two of these cars were completely identical (o.V. 2005). The statement, coined by Henry Ford (1922) "Any customer can have a car painted any color that he wants as long it is black" is not valid anymore. Material customization as a complexity driver The aforementioned variety is a result of the combination of different order codes. In spite of the fact that the number of possible variants is almost unlimited, customers are often not able to order their "dream" car. The main reason can be seen in the limited choice of available colors and materials. Car makers basically offer only a small number of different interior colors and materials (Chin 2005), because they are considered to be one of the most important complexity drivers. In combination with different technical variants of a certain component, colors as well as materials have a deep impact on the total number of item numbers needed to represent all variants of a component.

432


Alders (2006) illustrates this problem by a simple example. If a component with 26 possible technical variants is offered in just one color, 26 item numbers are needed to represent all variants. If the number of available colors increases up to five, the number of required unique item numbers increases to 130. Consequently the demand for a certain variant over the complete product life cycle fluctuates from 20 up to 300.000. Similar results are for example also presented by Anderson (1997). The main problem of a high number of product variants is the hidden costs (Saeed and Young 1998). The costs of complexity are not transparent and could have a negative impact on the competitive advantage of the enterprise. Complexity costs for every single product variant are for example caused by the development department that must detail the design or through procurement and logistic processes. Case study – Designing an individual headrest cover As shown in the previous section, color and material customization can lead to a myriad of possible product variants. For this reason, car makers basically offer only a small range of interior colors and materials, in order to be able to deal with complexity. Additionally, some car makers extend their offer by exclusive optional packages to fulfill nearly every customer wish regarding to the design of certain components. An illustrating example is shown in Figure 2. It outlines the meaning of the term individualization as it is used in this paper and explains the trend towards even more customized and individual components. Figure 2 shows an individual cover of a headrest that can be adapted by a customer according to his individual demands and wishes. To achieve this, the mass customizer offers several options to the customer which supports the specification of all relevant characteristics of the headrest cover. The selection is only restricted by a catalog which defines the available materials and colors (Salvador and Forza 2007; Piller 2003). A first scenario considers the situation where a customer is able to determine three essential characteristics of the cover: the material, the color and the thread used for stitching. Proceeding on the assumption that the customer can chose between ten materials, fifteen colors and thirty different threads, it is possible to design 1.125 different headrest covers. The second scenario considers the situation where a customer can decide separately on the mentioned characteristics for the central piece, as well as for the side pieces of the cover. Based on the assumption made above, a customer is now able to design more than twenty million different kinds of headrest covers. Both

433


given scenarios consider only the variation and complexity necessitated by some design specific characteristics. The example doesn't care about different technical variants as for example the kind of seat. Including technical variants will lead to a sharp increase of possible combinations. Furthermore, it could happen that a certain technical variant introduces a new design specific characteristic that can also be determined by the customer, which in turn leads to an even more increased number of feasible combinations. catalogue (individual options) leather

color

stitching

leather

defin ed by

color stitching

1 i ndivid ual varia nt out of 20.250.00 poss ible vari ants

central piece: 10 materials 15 colors 30 threads

side pieces: 10 materials 15 colors 30 threads

Figure 2: Designing an individual headrest cover. construction notification

Pseudo item number

individual part

1. XXXX - Non series headrest cover Central piece

: Nappa leather in green

Stitching

: Thread in black

Side pieces

: Nappa leather in yellow

Stitching

: Thread in green

Location

: Driver & passenger seat

HRC 000 000 xx

Figure 3: Using construction notifications and pseudo item numbers to identify individual items.

Construction notifications as an instrument to cope with complexity The conclusion of the nearly infinite number of possible variants for only one technical variant of a headrest cover is that the exclusive use of item numbers is not adequate for identifying highly individualized components. The use of pseudo item numbers is also impossible. In contrast to Mather (1982), who defines a pseudo item number as a set of physical products which are always needed in the assembly process, this work associates a pseudo item number with precisely one item. Thereby it defines only the kind and the function of a part, but it doesn't predict something about the design and the appearance of the described item. In

434


point of fact, it seems to be even more appropriate to use an additional layer, the so called construction notifications (Teich and Oestreich 2006). These documents act as a mediator between the basic configuration engine and the product model for any individual package and is used for a structured and detailed textual description of the individual item. By using construction notifications in connection with pseudo item numbers it is possible to shift complexity from the product model towards the configuration model (Figure 3). A Configuration Model for Construction Notifications Product configuration systems Beside the benefits of customizable products, there are also challenges for suppliers and manufacturers, if they are moving into mass customization. One important point is the elicitation process to gather all information needed for individualization (Bermann 2002; Broekhuizen and Alsem 2002; Piller et al. 2004; Zipkin 2001). As stated by several authors, this process can be a complex task that can also lead into mass confusion (Huffman and Kahn 1998; Desmeules 2002). A further challenge is the increased amount of information that must be transferred, as well as an increased number of information flows (Bermann 2002; Broekhuizen and Alsem 2002; Piller et al. 2004; Zipkin 2001). Bourke (2000) defines the term product configurator as "software with logic capabilities to create, maintain and use electronic product models that allow complete definition of all possible product option and variation combinations". A configurator is used to support a company in the product configuration process. According to Forza and Salvador (2002) this is "the process through which the customer’s needs are translated into the product information needed for tendering and manufacturing". Using configuration systems to assist employees, retailers or even customers during the configuration task offers a lot of benefits. First of all, configurators have the ability to reduce or even eliminate product specification errors. Due to the fact that they are also able to carry out complex feasibility checks, they also reduce effort needed for the configuration process (Barker and O'Connor 1998; Aldanondo et al. 2000). Therefore, the sales department is able to sell even more complex and customizable products. A further benefit of configurators is the generation of standardized specifications as a result of the configuration process (Forza and Salvador 2002). But there are also challenges with configurators. One of the most important issues is the continuous updating and maintaining process of the knowledge base


435

(Blecker et al. 2005; Barker and O'Connor 1998; Tiihonen and Soininen 1998). In the case of complex products it is often necessary to acquire a lot of information about product characteristics and dependencies between products. Furthermore, testing of gathered information is also a difficult but essential task (Tiihonen and Soininen 1998). Intuitive interfaces for product configuration systems One important challenge in the field of research concerning product configuration systems is the development of flexible and intuitive interfaces (Fleischhanderl 2005; Felfering et al. 2006; Roach et al. 2005). The goal of this research is an even deeper involvement of the customer into the information elicitation process. First successful approaches to support the generation of adaptive configuration dialogues have already been proposed. The configuration systems developed within the scope of the CAWICOMS project (Ardissono et al. 2003) as well as the Marrakesch research project (Hümmer et al. 2004; Dietrich et al. 2004) support the generation of custom user interfaces on the basis of powerful configuration models. The elicitation of the needed information for highly customizable products as shown in Figure 1 would also be possible by the use of the Marrakesch configuration model. The reasons for the development of a new configuration model are twofold. On the one hand side, it is highly expensive to model the complete individual offer with all relations and restrictions by using the Marrakesch configuration model. On the other hand side, the configuration dialog, generated by Marrakesch seems not suitable to support the information elicitation process in an appropriate manner. Instead of presenting a final configuration dialog with all relevant product characteristics of an individual package, Marrakesch only supports the information elicitation for one product characteristic within a single configuration step. To collect all needed information, it is necessary to pass through a sequence of several configuration dialogs. Requirements towards a flexible configuration model This section points out some basic requirements for a configuration system for material customization. The main purpose of this system is the generation of flexible document structures that can be used to describe individual components on any arbitrary level of detail. Elicitation of individualization information: The first point refers to the way how to capture all needed information for individualization. Conventional configuration systems support the capturing of individual customer wishes through the use

436


of simple and unstructured free texts (Holthöfer and Szilagyi 2001). The main disadvantage of this procedure is the fact that it is impossible to process information outside the configuration system. To avoid this, it is necessary to define a flexible data structure that can be used to elicit all needed information according to a given strategy. In order to be able to gather information for all types of individual components, the data structure must guarantee a maximum of flexibility. Templates: The configuration model must support the generation of templates for different classes of individual items. A template is a document that describes all possible facets of an individual item. Furthermore it is necessary to use rules and constraints to determine whether a certain part of the template is needed for an order specific description or not. Based on these assumptions, a construction notification consists of a set of adapted templates. The configuration system must be able to identify all needed templates for an order specific description and adapt them automatically to the given order specification. Ease of use: The generation of templates must be carried out through employees of the responsible competent department. The process must be arranged in a way that users without having specific IT skills are able to draw up extensive and complex templates too. Elimination of specification errors: The configuration system should be able to reduce or eliminate specification errors during the configuration process. The main source of error in the area of material customization is the selection of materials or colors that are not available for specific components. Thus, the configuration system must offer elements that support the selection of options from predefined option groups. Furthermore it must be possible to set up relationships or dependencies between selection elements to avoid the selection of options, which are not available in combination with other options. Such automated checks reduce the need for technical or product experts and allows less skilled users, like employees, retailers or customers, to draw up complete and correct configurations. Separation of configuration and product model: The configuration model should only be coupled loosely to a product model. Thereby it becomes possible to draw up a template without having a counterpart (an item or package) in the product model. On the one hand, construction notifications contain mainly design specific information of items, which are not part of the product model. On the other hand, the separation of both models supports the integration of the configuration system into an existing system topology. In order to be able to transfer a generated construction notification into a bill of materials, it is only required to specify a


437

lean interface between both models. The generation of a bill of materials may be sometimes necessary to support procurement processes for additionally needed standard items (Oestreich and Teich 2006). Basic configuration model Prerequisite for the generation of construction notifications is the existence of a powerful configuration model. It is used to model a set of freely definable characteristics of an individual component outside of the product model. By applying this strategy it is possible to reduce the complexity of the product model significantly. As a result, the expenses for preparation and maintenance of master data of the product model can be reduced remarkably. As it was shown in the previous section, the process of creating construction notifications should be based on existing templates. While a template describes a certain individual component in all of its facets, a template adapted to a specific order contains only those parts of the original template that are relevant for this order. Nevertheless, the fundamental structure of both descriptions is identical. The decisive difference between both structures is only related to the content of the documents. This makes it is possible to use a homogenous configuration model for templates as well as for construction notifications. The general structure of the configuration model is illustrated in Figure 4. The represented hierarchical model encompasses four different levels. The first three levels are used to reduce the complexity of the model by subdividing the general document structure into different kinds of containers. The elements at the fourth level represent the underlying contents of the document. The first level of the model is called "package". It serves as a general container that is able to incorporate all elements that belong to a description of an individual component. This container can be used later as an entry point to get access to all information related to a package. While an order specific construction notification may consist of more than one individual package, it is possible to assign one element of this type to each individual package of the order. In turn, a template is associated precisely with one element of this type. The second level of the model is called "module". By the use of elements of this level it is possible to create a coarse content-related structuring of the document, where a module describes a delimited part of the complete package. Following the meaning of the package level, any module acts also as a container element that is able to incorporate all elements that are related to a module. The third and last level that is used for structuring the contents of the document is called "building blocks". A building block acts like a package or module as a

438


container, designed to incorporate further elements. In contrast to both containers mentioned above, a building block doesn't contain any other container element. Instead of this, a building block contains all elements that are responsible for the visual representation of the contents of a document. The use of different kinds of building blocks supports the development of a homogenous and well comprehensible overall document structure.

PA C K A GE

M O D U LE

C OMPOSED

P L A IN B U IL D IN G B L O C K L A B EL

C on ta ine r

TE X TB O X S IM P LE ... SP E C IF IC AT IO N E L E M E N T

S E LE C TIO N B a s ic s tru c tu re o f th e C o n fig u ra tio n m o d e l

R e fin e d c o n fig u ra tio n m o d e l

Figure 4: Basic configuration model.

All elements that are responsible for the visual representation of the content of a document are called "specification elements". They are classified into different groups, depending on their visualization task. The standardization of all elements is carried out by the use of an interface that must be implemented by every element. The first group contains such elements that are used for visualization of simple texts or for capturing of free text entries only. Well known members of this group are labels and textboxes. The modeling of special dependencies between those elements is not necessary. In contrast to the elements of the first group, the second group contains elements that can be used for the selection of options of features. With respect to the intended purpose, those elements offer a considerably higher functionality than those of the first group. On the one side, it is possible to assign a certain feature to such an element. As a result of this, the element allows only the selection of these options that are related to the specified feature. On the other side it is possible to determine multiple relationships between elements. This procedure facilitates the

439


avoidance of erroneous user inputs in situations where the range of available options of a feature depends on the selected option of another feature.

##

Non series headrest cover

XXXX Central piece: Primary stitching: Secondary stitching: Side pieces: Stitching: Assemble at:

Drawing Area

Document-Structure Editor V-0000001 VM-0000001 VB-0000001 VS-0000001 VS-0000002 VS-0000003 VB-0000002

Property Editor Property Font Size Bold Italic Name Text Type

Value 12 True False VS-0000002 XXXX Label

VS-0000003

Toolbar (extract) Moduls Building blocks

Headline 1

Standard Input 1

Specification elements

Figure 5: Template designer – Structure and components.

Reference implementation of the configuration model This section presents an extract of a reference implementation of the configuration model in a real world application. Therefore, the sample construction notification of the individual headrest cover is used to demonstrate how to set up individual templates, as well as how to use templates to create a configuration dialogs adapted to a specific order. Figure 5 shows the four main components of a template designer tool. The central component is shown at the top of Figure 5. It is used to illustrate the structure and the design of a template. The example shows the basic template structure that can be used to describe an individual headrest covers. The template consists of one module, seven building blocks and eight simple label elements. Furthermore, one additional element ("##") is used to create a consecutive numbering in case the template is used in an order specific construction notification. Because the number of the module in a construction notification depends on the placement of the package, it is needless to display this number in the template.

440


All available elements that can be used to define the structure and the content of the template can be found in the toolbar. According to the structure of the configuration model, there are several subsets of elements. The template can be created using a drag & drop mechanism. After an element is selected, it can be placed at the desired position in the central area of the editor. Thereby, modules are visualized as tabular pages. All relevant properties of an element can be edited in the associated property editor. Each modification that changes the visual appearance of the element will be visible immediately. The last visual component of the template editor is a viewer to display the complete structure of the template in a simple tree structure. This component can be used to select particular elements in a fast and efficient manner.

##


XXXX Central piece:

*

?

in

*

?

Primary stitching:

?

Secondary stitching:

?

Side pieces:

in

?

Stitching:

?

Assemble at:

?

Property Editor Property Name Type Constraints Selection 1

?

Value VB-0000004 Building Block +XXX1

possible rela tionships m ultiple selec tion * m ultiple selec tion (starting point) restric tion standard value

Figure 6: Template designer: Modeling constraints and linked elements.

Figure 6 shows how to set up a simple selection constraint for a building block. The constraint states, that the building block is only needed, if an order contains the order code "XXX1". The configuration model also supports constraints for packages and modules. Additionally, Figure 6 as well shows how to set up relationships between elements, in order to avoid or to reduce specification errors at a later point of time. Relationships can be initialized by selecting an appropriate master element for a dependent element. If two elements are linked together, it is only possible to select those options for the dependent element, which are supported by the selected option of the master element.

441


Finally, Figure 7 shows an extract of an order specific construction notification. The order doesn't contain the order code "XXX1". Therefore, the building block "Secondary stitching" isn't needed to describe the individual headrest cover. All shown selected options for the specification elements can be selected by the use of dialogs, as shown at the bottom part of Figure 7.

1.


XXXX Central piece: Primary stitching: Side pieces: Stitching: Assemble at:

Nappa leather

*

?

in navajowhite

Thread in navajowhite Nappa leather

*

?

?

?

in navajowhite

?

Thread in navajowhite

?

Driver & Passenger seat

?

Current selection DriverProperty & PassengerEditor Seat Available elements Driver seat Passenger seat Driver & Passenger seat

Figure 7: Construction notification – adapted according to a given order.

Technical Product Modeling As mentioned in the previous section, construction notifications are documents that can be used to describe individual components in a very detailed way. By using these documents it is possible to optimize the procurement processes for individual parts within a company (Teich and Oestreich 2006). But in many cases, there is also an additional demand for standard parts needed for the final assembly of an individual component. In order to support the procurement processes of such parts too, it is necessary to transfer a construction notification directly into a bill of materials (Lieberman and Leete 2000). The main objective of this section is also illustrated in Figure 8. On the basis of the construction notification for the individual cover of a headrest, it is shown how an order specific bill of materials can be derived by merging a variant or generic bill of materials with a corresponding construction notification. All information about relevant options, needed for the transformation task is clearly determined by the construction notification. The translation of the option "driver & passenger seat" into the short form "DSPS" can be carried out easily by the use of the feature model.

442


Technical product model vs. configuration model Prerequisite for the generation of construction notifications is a powerful configuration system, as well as a suitable configuration model. Product models like bills of materials are used to describe the technical structure of a product. Both concepts are often used in product configuration literature in a different way. For example, Bieniek (2001) considers the product model as a part of the configuration model, whereas other authors use the term product model only. The product model describes the technical structure of a product or a product family. In order to be able to support the modeling of complex product families with many variants in an effective way, the use of generic structures is recommended explicitly (van Veen 1992; Jiao et al. 2000). These generic structures can be considered as a schema, used for the automatic generation of order specific product structures. It is assumed that the product model provides a suitable mechanism to perform this task. In contrast to the product model, the configuration model serves as the main knowledge base for the configuration system. For this purpose the configuration model provides a simplified marketing oriented structure of the product model. The model supports the assignment of options to features, as well as the definition of several rules and constraints (Sabin and Weigel 1998). Furthermore it contains all necessary knowledge of how to capture all required information about an individual component. This knowledge is the key for the generation and adaptation of flexible configuration dialogs. In order to ensure one of the most important functions of configuration systems — the ability to transfer a given configuration directly into a bill of materials (Lieberman and Leete 2000) — it is necessary to define an interface between both models. Using a lean interface, as illustrated in Figure 9, it is possible to link both models together and mostly preserve the independence of the product and configuration model. Prerequisite for usage of this interface is an additional feature model (Bieniek 2001) providing an integrative language. The meaning of this model is outlined in a more detailed way in (Teich and Oestreich 2006). Generic and order specific bills of materials The structure of a product describes the way in which the product is assembled from purchased parts and/or semi-finished products. The relationship between two products in this structure is a so called gozinto-relationship (van Veen 1992) that represents the fact that a product is used for the manufacturing or assembly of another product. Based on this assumption, a bill of materials (BOM) can be considered as the set of gozinto-relationships of a certain product.

443


HRC 000 001 1. XXXX - Non series headrest cover Central piece

: Nappa leather in green

Stitching

: Thread in black

1 x HRC 000 DS Constraint : (DS , DSPS) 1 x HRC 000 PS

Side pieces

: Nappa leather in yellow

Stitching

: Thread in green

Assemble at

: Driver & passenger seat

constraint : (PS , DSPS) 2 x WAS 001 003 constraint : (DS , PS) 4 x WAS 001 003 constraint : (DSPS, DS+PS)

HRC 000 001 1 x HRC 000 DS 1 x HRC 000 PS 4 x WAS 001 003

Figure 8: Construction notification vs. bill of materials.

In order to cope with complexity in production environments with a high product variety, an even more flexible product structure is required. While bills of materials are used to represent an order specific view of a product, generic bills of materials (GBOM) are used to model a complete product family with all of its variants (van Veen 1992; Jiao et al. 2000). An order specific BOM can be derived from a GBOM. The decision whether a certain part of the GBOM is used in the BOM depends on the specification of the desired product. Therefore it is necessary to add constraints to the gonzito-relationships of the GBOM. Basis for the derivation of an appropriate algorithm that supports the transformation of a construction notification into a bill of materials is the existence of a formal model for generic bills of materials, as well as order specific bills of material. These models are presented within the following both sections. The notation applied here follows those used by van der Aalst (1997). Feature Model

Configuration Model

Technical Product Model

Figure 9: Configuration model vs. product model.

444


Modeling generic bills of materials A generic bill of materials GBOM can be expressed as a tuple GBOM = (P, root, Rmandatory, Roptional, V, D, C). P is the set of all products or parts that are used within the GBOM. Based on the assumption that a GBOM is organized according to the low level code procedure (Günther and Tempelmeier 1997), P can be divided into different subsets P = {P0,…Pd-1} where d represents the number of low level codes. A subset Pi = {pi1…pia}, i = 0...d-1, a = |Pi| contains all parts p with the same low level code i. The root of a GBOM is represented by root. It is defined root ∈ P0, |P0|=1. root is the only part of the low level code of 0. The model presented here distinguishes between two types of gozintorelationships. The first type Rmandatory ⊆ P × P describes all obligatory relationships between parts. Every edge rm = (px, py) ∈ Rmandatory, px ∈ Px, py ∈ Py, x > y states that a part px has to be contained in the BOM obligatory in case that the BOM contains the product py. The requirement that px is assigned to a higher low level code as py ensures that the structure is acyclic. In contrast Roptional ⊆ P × P × C represents the set of optional edges ro = (px, py, ca) ∈ Roptional, px ∈ Px, py ∈ Py, x > y, c ∈ C. A product px is only required in a BOM if a condition c is fulfilled. The condition ro ∈ Roptional ⇒ rm(px, py) ∉ Rmandatory ensures that there is no obligatory and optional edge between two products px and py at the same time. V describes the set of features which characterize the GBOM. It is defined as V = {V1 … Vn} where n is the number of features. All possible specifications of the features are described in domains D = {D1… Dn}. A tangible specification of a feature is presented as dfg ∈ D, f = 1…n, g = 1...|Df|. C finally represents the set of constraints which are required for the modelling of the optional gozinto-relationships. The constraints are expressed in conjunctive n normal form as an "AND" relation of clauses Kz. It applies C : z =1 K z.

∧

Every clause Kz consists of a disjunction of an optional number of specifications d zr ∨ ¬d zs r, s ∈ {0…|Dz|}. If r = s = 0 then of a feature Vz. It holds: K z : the clause is always fulfilled. Therefore the constraint is independent from the feature Vz.

∨

∨


445

Modeling order specific bills of materials An order specific BOM can be expressed as a tuple BOM = (GBOM, P, root, R, OC). Basis for the creation of a BOM is the existence of a GBOM. The definition of P and root is equivalent to the modelling of GBOM. It is defined that the root of the BOM is equal to the root of the GBOM (root = rootGBOM). In contrast to the GBOM the BOM contains only such gozinto-relationships R that represents obligatory relationships between parts. Therefore the definition of R GBOM can be defined equivalently to Rmandadory . OC represents the values for each specification element dfg with respect to a given order. By default, every specification dfg is set to 0. If a specification dfg is used in the given order, then dfg is set to 1. It holds OC: ∀d fg ← 0 /1 f = 1...n, g = 1... |Df |. Furthermore there exists a function SAT (c, OC ) → 0 /1 , which checks the validity of a constraint c ∈ C against the order codes. If the constraint can be fulfilled then the function returns 1 otherwise 0. Extending the configuration model Up to this point, the basic model for the generation of construction notifications is independent from a particular product model. In order to transfer a construction notification into a BOM it is first of all necessary to extend the existing configuration model. As illustrated in Figure 9 it is necessary to set up an interface between the configuration and the product model. On the one hand, it is first of all required to support the assignment of a generic bill of materials to a template of the configuration model. Because the description of a package is always based on one certain template, the required generic bill of materials for the package can be identified in a fast and effective manner. On the other hand it necessary to denote those specification elements of a construction notification that are needed to determine a certain variant of the generic bill of materials. For this purpose a Boolean flag is suitable. By assigning such a flag to every specification element it is possible to identify all required options. Further information about the extension of the configuration model can be found in (Oestreich and Teich 2005).

446


Transformation algorithm On the basis of the given definitions it is now possible to define the algorithm for the transformation of a package of a construction notification into a bill of materials. The presented algorithm requires three parameters: a part p ∈ P, the low level code of p and finally a package related order code list that contains all characteristic features of the package. The algorithm starts with the initialization of p = rootGBOM and the corresponding low level code of 0. The algorithm follows closely to the procedure presented by Olsen et al. (1997). The presented algorithm (Figure 10) is based on the processing of the whole structure of a GBOM by using a depth first search strategy. Thereby, the algorithm adds any gozinto relationship rm ∈ Rmandatory to the order specific BOM by default. For any gozinto relationship ro ∈ Roptional it is necessary to check whether the assigned constraint can be fulfilled or not. If the constraint check succeeds the gozinto relationship is also added to the order specific BOM, otherwise it will be rejected.

Figure 10: Transformation algorithm.

Conclusions In the first part of this paper a new approach for the configuration of complex individual products was introduced. Instead of modeling product variety by extensive product models, a flexible configuration model can be used to deal with complexity. Therefore, it is possible to clearly identify an individual item by a construction notification. The generation process of those documents is well supported by the configuration system. Prerequisite are existing templates. Each


447

template describes one individual package in all of its facets. On the basis of a specific order, the configuration system is able to identify all required templates and adapt them according to the specification of the order. The resulting individual dialog must be finally filled out by an advisor of the manufacturer. The second part of this paper dealt with technical product modeling. While a construction notification describes an individual part in a detailed way, there is also often an additional demand for standard parts needed for the final assembly of the individual component. For this purpose an algorithm was presented that supports the transformation of construction notifications into bill of materials. On the basis of existing variant or generic bills of materials and the content of a completely specified construction notification it is possible to derive an order specific bill of materials. The algorithm can be considered as a bridge between existing ERP systems and the construction notification configuration system. Further work on this concept will deal with customer integration. At the moment, construction notifications can be used to handle internal product complexity only. In order to decrease the perceived complexity for customers and to improve the willingness to purchase, it is necessary to extend the configuration model with visualization possibilities. Furthermore, it is necessary to provide a reduced form of construction notifications to the customer because not all characteristics of a notification are relevant to them.

References Aldanondo, M., Rougé, W. and Véron, M. (2000). Expert configurator for concurrent engineering: Caméléon software and model. Journal of Intelligent Manufacturing. 11: 127–134. Alders, K. (2006). Komplexitäts- und Variantenmanagement bei der AUDI AG. In: Individualisierte Produkte – Komplexität beherrschen in Entwicklung und Produktion. Lindemann, U., Reichwald, R. and Zäh, M.F. (eds.). Berlin, Heidelberg: Springer, 221–238. Alford, D., Sackett, P. and Nelder, G. (2000). Mass Customization: an automotive perspective. International Journal of Production Economics. 65(1): 99–110. Anderson, D. (1997). Agile Product Development For Mass Customization. Chicago, London, Singapore: IRWIN Professional Publishing. Ardissono, F., Friedrich, G., Goy, A., Holland, M., Petrone, G., Russ, C. and Schäfer, R. (2003). User: Adaptive Configuration of Products and Services. Proceedings of the 18th International Joint Conference on Artificial Intelligence (IJCAI). Workshop on Configuration. Acapulco, Mexico. Barker V.E. and O'Connor, D.E. (1989). Expert systems for configuration at Digital: XCON and beyond. Communications of the ACM. 32(3): 298–318. Becker, H. (2007). Auf Crashkurs. Automobilindustrie im globalen Verdrängungswettbewerb. Berlin, Heidelberg, New York: Springer. Bermann, B. (2002). Should your firm adopt mass customization. Business Horizons. 45(5): 51–60.

448


Bieniek, C. (2001). Prozessorientierte Produktkonfiguration zur integrierten Auftragsabwicklung bei Variantenfertigern. Aachen: Shaker. Blattberg, R.C. and Glazer, R. (1994). Marketing in the information revolution. In: The marketing information revolution. Blattberg, R.C., Little, J.D. and Glazer, R. (eds.). Boston: Harvard Business School Press, 9–29. Blecker, T., Friedrich, G., Kaluza, B., Abdelkafi, N. and Kreutler, G. (2005). Information and Management Systems for Product Customization. New York: Springer. Bourke, R.W. (2000). Product configurators: key enablers for mass customization. MIDRANGE ERP, URL: www.bourkeconsulting.com/documents/Aug2000 IntroCfgrs.pdf (Retrieval: May 12, 2005). Broekhuizen, T.L.J. and Alsem, K.J. (2002). Success Factors for Mass Customization: A Conceptual Model. Journal of Market – Focused Management. 5(4): 309–330. Chin, R.C.C. (2005). How Mass customization Changes the Design Process: MIT Media Lab’s Concept Car Project. Proceedings of the 3rd Interdisciplinary World Congress on Mass Customization and Personalization (MCPC 2005). Tseng, M.M. and Piller, F.T. (eds.). Hong Kong. Da Silveira, G., Borenstein, D. and Fogliatto, F.S. (2001). Mass customization: Literature review and research directions. International Journal of Production Economic. 72(1): 1–13. Davis, S. (1987). Future Perfect. Reading: Addison-Wesley Publishing. Desmeules, R. (2002). The impact of Variety on Consumer Happiness: Marketing and the Tyranny of Freedom. Academy of Marketing Science Review. URL: www.amsreview.org/articles/ desmeules122002.pdf, (Retrieval: December 15, 2006). Dietrich, A.J., Hümmer, W. and Meiler, C. (2004). A Meta Model based Configuration Approach for mass-customizable Products and Services. Proceedings of the 4th Workshop on Information Systems for Mass Customization (ISMC 2004). 4th International ICSC Symposium on Engineering of Intelligent Systems (EIS 2004). Funchal, Portugal. Duray, R., Ward, P.T., Milligan, G.W. and Berry, W.L. (2000). Approaches to mass customization: configurations and empirical validation. Journal of Operations Management. 18(6): 605–625. Felfering, A., Scheer, C. and Loos, P. (2006). How to recommend configurable products? Proceedings of the 17th European Conference on Artificial Intelligence (ECAI). Workshop on Configuration. Riva del Garda, Italy, 5. Fleischhanderl, G. (2005). Configurators in innovative or standardized business processes. Proceedings of the 19th International Joint Conference on Artificial Intelligence (IJCAI).Workshop on configuration. Edinburgh, Scotland, 72. Ford, H. (1922). My Life and Work. Garden City, New York. Forza, C. and Salvador, F. (2002). Product configuration and inter-firm coordination: an innovative solution from a small manufacturing enterprise. Computers in Industry. 4: 37–46. Günther, H.-O. and Tempelmeier, H. (1997). Produktion und Logistik. 3rd edition. Berlin, Heidelberg, New York: Springer. Günthner, W.A, Wilke, M., Zäh, M.F., Aull, F. and Rudolf, H. (2006). Produktion individualisierter Produkte. In: Individualisierte Produkte – Komplexität beherrschen in Entwicklung und Produktion. Lindemann, U., Reichwald, R. and Zäh, M.F. (eds.). Berlin, Heidelberg: Springer, 63–87. Haag, A. (1998). Sales configuration in business processes. IEEE Intelligent Systems. 13(4): 42–49. Hamprecht, H. (2006). Das heftige Hochrüsten der Haus – Tuner. Automobilproduktion. No. 2. Herlyn, W.J. (1990). Zur Problematik der Abbildung variantenreicher Erzeugnisse in der Automobilindustrie. Düsseldorf: VDI.


449

Holthöfer, N. and Szilagyi, S. (2001). Marktstudie: Softwaresysteme zur Produktkonfiguration. Paderborn: Heinz Nixdorf Institut. Huffman, C. and Kahn, B.E. (1998). Variety for Sale: Mass Customization or Mass Confusion. Journal of Retailing. 74(4): 491–513. Hümmer, W., Meiler, C., Dietrich, A.J. and Müller, S. (2004). Data Model and Personalized Configuration Systems for Mass Customization — A Two Step Approach for Integrating Technical and Organizational Issues. Proceedings of the International Conference on Economic, Technical and Organizational Aspects of Product Configuration Systems (PETO). Edwards, K., Hvam, L., Moldrup, M., Moller, N., Pedersen, L. and Riis, J. (eds.). Copenhagen, Denmark. Jiao, J., Tseng, M.M., Ma, Q. and Zou, Y. (2000). Generic Bill-of-Materials-and Operations for HighVariety Production Management. Concurrent Engineering. 8(4): 297–321. Kneppelt, L.R. (1984). Product structuring considerations for Master Production Scheduling. Production and Inventory Management. 25(1): 83–99. KPMG (2005). Impulse in der Automobilindustrie. Der Automobilmarkt in Asien, URL: www.Kpmg.de/library/ brochures\_surveys/12025.htm (Retrieval: December 27, 2005). Lampel, J. and Mintzberg, H. (1996). Customizing customization. Sloan Management Review. 37(1): 21–30. Lieberman, M. and Leete, B. (2000). Getting a Grip on Configurators – Comparing Configurators Part II. URL: www.plmic.com/news/plm-article-Getting-Grip-Configurators-part2.html# Busness_Analysis_plm. (Retrieval: May 20, 2006). MacCarthy, B.L., Brabazon, P.G. and Bramham, J (2003). Fundamental modes of operation for mass customization. International Journal of Production Economics. 85(3): 289–304. Mather, H. (1982). Bills of Materials, Recipes and Formulations. Atlanta: Wright Publishing Company. o.V. (2005). Wahnsinn mit Methode. Automobilproduktion. 1: 38–42. Oestreich, E. and Teich, T. (2005). Überführung von Produktkonfigurationen auf der Basis dynamischer Dokumentstrukturen in Variantenstücklisten. Industrie Management. 31(3): 39–42. Oestreich, E. and Teich, T. (2006) Transforming construction notification into bill of Materials. In: Proceedings of the 17th DAAAM International Symposium 2006. Katalanic, B. (ed.).Wien, Austria: DAAAM International, 271–272. Olsen, K.A., Saetre, P. and Thorstenson, A. (1997). A generic bill of materials based on a programming language notation. Proceedings of the Norsk Informatikk Konferanse. Piller, F.T. (2003). Mass Customization – Ein wettbewerbsstrategisches Konzept im Informationszeitalter. 3rd Edition, Wiesbaden: Deutscher Universitäts-Verlag. Piller, F.T, Möslein, K. and Reichwald, R. (2000). Information as a Critical Success Factor for Mass Customization, Or: Why Even a Customized Shoe Not Always Fit. Proceedings of the ASAC-IFSAM 2000 Conference. Montreal, Quebec, Canada. Piller, F.T., Möslein, K. and Stotko, C.M. (2004). Does mass customization pay? An economic approach to evaluate customer integration. Production Planning & Control. 15(4): 435–444. Pine, B.J. II (1993). Mass customization: The new frontier in Business Competition. Boston: Harvard Business School Press. Roach, G.M., Cox, J.J. and Sorensen, C.D. (2005). The product design generator: a system for producing design variants. International Journal of Mass Customization. 1(1): 83–106. Rosenberg, O. (1996). Variantenfertigung. In: Handbuch der Produktionswirtschaft. 2nd Edition. Kern, W., Schröder, H.-H. and Weber, J. (eds.). Stuttgart: Schäffer-Poeschel. 2119–2129.

450


Sabin, D. and Weigel, R. (1998). Product Configuration Frameworks: A Survey. IEEE intelligent systems. 13(3): 42–49. Saeed, B. and Young, D. (1998). Managing the Hidden Costs of complexity. URL: www.bcg.com/publications/files/Managing_WH.pdf. (Retrieval: December 15, 2006). Salvador, F. and Forza, C. (2007). Principles for efficient and effective sales configuration design. International Journal of Mass Customization. 2(1/2): 114–127. Sinz, C. (2006). Comparing Different Logic-Based Representations of Automotive Parts Lists. Proceedings of the 17th European Conference on Artificial Intelligence (ECAI): Workshop on Configuration. Riva del Garda, Italy, 41–44. Teich, T. and Oestreich, E. (2006). Produktdifferenzierung durch Individualisierung von Ausstattungspaketen im Rahmen der Mass Customization im Bereich der Automobilindustrie. In: Virtuelle Organization und Neue Medien 2006 (GeNeMe06). Meißner, K. and Engelien, M. (Eds.). Dresden, Germany: 399–412. Tiihonen, J. and Soininen, T. (1998). Product configurators – Information System Support for Configurable Products. Using Information Technology During the Sales Visit. Cambridge. Tseng, M.M and Jiao, J. (2001). Mass Customization. In: Handbook of Industrial Engineering. New York: Wiley-Interscience, 684–709. van der Aalst, W.M.P. (1997). Designing workflows based on product structures. Proceedings of the 9th International Conference on Parallel and Distributed Computing and Systems. van Veen, E. (1992). Modeling product structures by generic bills-of-materials. New York: Elsevier. Vollmann, T.E., Berry, W.L. and Whybark, D.C. (1988). Manufacturing planning and control systems. Illinois: McGraw-Hill. Zipkin, P. (2001). The Limits of Mass Customization. MIT Sloan Management Review. 42(3): 81–87.

Author Biographies Dr. Erik Oestreich studied business and computer science, majoring production and industrial management as well as information technology, at the Chemnitz University of Technology. Since 2003 he has been working for a German premium car manufacturer as a system analyst in the area of product configuration systems. In 2009 he finished his PhD project. Within this project, he has developed a novel approach towards configuration of individual parts of complex technical products. His main research interests are mass customization, product configuration systems as well as variety and complexity management. Contact: [email protected] Prof. Dr. Tobias Teich leads the Institute of Management and Information at University of Applied Sciences Zwickau. Before entering his recent position in Zwickau in fall 2002, he worked at the Technical University at Chemnitz and has been an associate professor of production management (1998-2002). His research focuses on supply chain management, ERP-systems, evolutionary optimization, competence networks and business intelligence. Contact: www.fh-zwickau.de/index.php?id=1123 | [email protected].

4.3

Mass Customization of Responsive Automated Assembly Cells Ulrich Berger Chair of Automation Technology, Brandenburg University of Technology Cottbus, Germany Sarfraz Ul Haque Minhas Chair of Automation Technology, Brandenburg University of Technology Cottbus, Germany Ralf Kretzschmann Chair of Automation Technology, Brandenburg University of Technology Cottbus, Germany Veronica Vargas Chair of Automation Technology, Brandenburg University of Technology Cottbus, Germany

The automotive industry is distinguished by regionalization and mass customization of products. This necessitates increased product diversity and decreased lot sizes. Thus more product types have to be handled along the process chain and common production paradigms will fail. Hence, Rapid Manufacturing (RM) will be used for manufacturing small individual lot sizes. Nevertheless, new solutions for joining and assembling these components are needed. The state-of-the-art production control solutions at the robot cell level as well as the plant level have certain disadvantages, such as manufacturer dependent programming of industrial robots and difficulty in implementation of synchronized robot simulation and complicated robot program execution. Additionally, common human machine interfaces do not provide necessary functions to interact with the devices in the robot cell. The data trafficking between devices within a robot cell is very high because of the need to process raw data. A comprehensive and real time intelligent production control and monitoring system can overcome these limitations. It is a cluster of intelligent soft computing algorithms and smart intelligent peripheral devices. The concept proposed in this chapter is based on three interlinked main modules: a technology data catalogue (TDC), a Product Process Resource Module (PPR) and a central programmable automation controller (PAC) for real-time sensor/actor communication. The Technology Data Catalogue (TDC) retrieves, shares, processes and structures relevant engineering data. The automated scheduling processor creates an optimized and/or adaptable work plan based on feature technology. It has access to the TDC to extract information about the applicable technologies needed for manufacturing process and its operations such as material characteristics, measuring and

451

452


monitoring conditions and their attributes. The processes are controlled by a Programmable Automation Controller (PAC) and Human Machine Interface Modules (HMI) for human interaction. The PAC and its supported software have widespread functionality especially for real time applications such as data logging and high frequency measurements and optimized control. Moreover, it provides homogeneous and standard interfaces to access heterogeneous devices in the system. Its compatibility and flexibility with the conventional programming languages has enabled in developing customized palette of functions and tools for scalable level of control solutions. Besides widespread accessibility, it can also influence the execution of robot program after pre-processing the data coming from various peripheral devices and calculating new robot paths by segmentation of detected curves and paths corresponding to the process planning information via one interface. The Human Machine Interface is developed using a Personal Digital Assistant (PDA) in which different levels of user profiles with customized functionalities. The application software for the HMI portable devices will be based on graphical programming platform LabView and Microsoft.Net. The Human Machine Communication is based on the ontological based approach as a natural language interaction system for filtering and translating into machine command. This translated command is executed finally. The concept is in the process of demonstration in a laboratory set-up with distinct assembly and joining processes for experimental validation in European research and development projects.

Introduction The current situation in the automotive industry is characterized by increasing requirements from the customer side on quality and individualization of products and, at the same time, imminent pressure on product prices. Car manufacturers create new product segments and enrich existing segments with more possibilities for individualization like regionalization and customization of products. The product diversification is combined with ongoing reduction of product life cycle time and an acceleration of innovation (Kuhn et al. 2002). Furthermore, the automotive industry is characterized by enabling innovations in light-weight vehicle structure, energy efficient power-train solutions and assistance systems. New manufacturing paradigms for automotive structures and components force the automotive industry to continuously promote the development of cost-efficient and innovative vehicles, with high-added customer value, increased personalization capabilities and environmental sustainability. According to the recent Global Technology Revolution 2020 report, issued by the RAND Corporation, there is a strong trend for on-demand manufacturing of components and small products to individual personal or corporate specifications (Silberglitt 2006). According to Ruohonen et al. (2006), several researchers suggest that mass customization should be adopted along the complete value chain. Therefore the whole product


453

lifecycle has to be customized by shifting towards mass customization. In this context, manufacturers need to focus on main drivers of mass customization i.e. customer requirements, information technology and operational capabilities. Mass customization is considered as an oxymoron that combines two contradictory production approaches i.e. craft and mass production. It provides customers with individualized products at the same efficiency as mass production (Pine II 1993; Tseng and Jiao 2001). But this approach triggers internal and external complexity (Blecker et al. 2006). The external complexities are relevant to the problem that arise in manufacturer supplier relationship and business. This chapter focuses on resolving internal complexities through intelligent control strategies and therefore discussion about external complexities is beyond its scope. Internal complexity is experienced inside the company while handling extensive product variety induced in mass customization [Blecker et al. 2006]. Although the development and process planning of diversified products is enabled with help of powerful CAD/CAM and PLM systems but handling of different products in the production line is still solved unsatisfactorily. Consequently, in a mass customized scenario, the production will be the bottleneck along the life cycle. The mass customization induced internal complexities of manufacturing industry in general and the automotive industry in special can be resolved by adopting intelligent production strategies. This chapter gives a comprehensive approach for intelligent production monitoring and control of automated assembly cells in the automotive industry to cope with the mass customized induced internal complexities. In this context, the overall objective is to employ the role of knowledge in full scope for intelligent control of production by precise adaptation and fast ramp up of technologies and equipment, parameter settings and enhanced production quality through intelligent process monitoring. State-of-the-Art: Mass Customized Assembly Cells As mentioned before, the adaptation of manufacturing cells in order to handle diversified products is a key to successful automated assembly and joining operations of customized parts. At present, the production control strategies cannot resolve the complexities that are generated due to the variety induced in the manufacturing systems. The state-of-the-art for the production of mass customized products can be divided into the technologies for manufacturing diversified components and for joining and/or assembling them to complex mass customized products. Besides already existing manifold efforts in the sector of Rapid Manufacturing (RM), the introduction of such principles in automated manufacturing cells is still in the beginning (Park 2006). RM technologies are

454


based on linear thermo-mechanical and chemical processes, varying ingot material properties, recipes and process parameters, whereas automated manufacturing tasks are based on complex operational conditions, real-time monitoring and heuristic decision features. The non-availability of ubiquitous operational knowledge and the absence of dynamic and explicit knowledge recuperation procedures minimize the achievement of on-demand capabilities. In this context, the Japanese NISTEP report no. 99 from 2005 identifies with high priority the establishment of a technology for converting implicit knowledge on manufacturing and manufacturing technique into explicit knowledge (NISTEP 2005).

Figure 1: Standard approach of an assembly cell.

Due to the absence of concrete implicit knowledge extraction and interpretation mechanism, a large number of automotive manufacturers are trying to handle mass customization by adapting the standard automated manufacturing/ assembly cell through hit and trials. A general structure of this cell (Figure 1) comprises of several components. The main components are the industrial robots with the robot control (RC) and additional programming devices. These devices can be classified as on-line and off-line devices. It is a common practice to generate robot programs through an off-line work planning module based on product, process and resources (PPR) information. Furthermore, the robot is connected via the robot control and programmable logic controller (PLC) with supplementary devices like sensor controllers.


455

The typical assembly cell consists of Industrial Robot (IR) with the Robot Controller (RC), the Human Machine Interface (HMI) and the Work Planning Module (WPM) based on Product, Process and Resources (PPR) information as an off-line programming system. Several additional devices can be used depending upon the nature of the operations being performed. A Cell Control System (CCS) is a combination of Programmable Logic Controllers (PLC) that synchronizes all devices, fixtures and equipment in assembly cells. Thus, sensors and their associated controllers have to monitor diversified assembly and joining tasks in order to modify the robot program and consequently the robot paths. This standard approach can work successfully for processing customized applications but a lot of interfaces have to be kept or developed and modification of control parameters. As a result, when an assembly cell based on this architecture is subjected to mass customization, exhibits certain disadvantages:

Limited flexibility in programming of industrial robots due to the manufacturer dependent programming environment,

Offline verification of industrial robot programs which in some cases limits the adaptation and consumes lot of time,

Complicated synchronized simulation of robots and its implementation,

Large number of interfaces and non-compatibility issues,

Fixed but highly branched program (spaghetti code) execution corresponding to the sensor data inputs and

Hindrance in database connectivity with certain Human Machine Interfaces such as Personal Digital Assistants (PDAs) for fast and remote interaction.

Figure 2: Missing link between explicit and implicit knowledge.

456


Furthermore, the actual implementation has lower capability, high downtimes of the automated machinery and consequently high troubleshooting costs. Besides the above mentioned complexities, the assembly cells are not precisely adapted due to the missing link between the ubiquitous operational knowledge (e.g. process knowledge, maintenance knowledge, technology or equipment ramp up knowledge) and the explicit knowledge. In most cases, the operational knowledge is not collected at assembly cells level and therefore the precise adaptation is not possible because of the unavailability of information about the actual ongoing process and its conditions. As a result there is a strong need to develop a knowledge based guidance system as a subsystem of automated production systems. These disadvantages can be eliminated if restructuring of production control is made upon intelligent methodologies and algorithms to deal mass customization. Requirements for New Assembly Cells The main challenges for coping mass customization at an assembly cell level is to achieve high production quality in case of high product variety and the mechanism for automatic variant specific product and process configuration. Therefore an effective and efficient automation solution is required to face these challenges. It needs a holistic redesign of assembly systems exploiting the state-of-the-art concepts of intelligent and flexible automation, robotic technologies and sensory systems. As a result, mass customization will incorporate high flexibility in the production system. High flexibility in product development cycle needs for reconfiguration and adaptation of assembly setups. Highly flexible devices, i.e. fixtures and robot grippers, are required for handling multi variants. Modularization of knowledge driven platform for data handling and automatic configuration of products as well as processes is essential for fast customization of automated assembly cells. Meeting these demands requires a fast and responsive production control system which is triggered by intelligent monitoring. Such systems can quickly reconfigure themselves to allow flexibility not only in producing new products but also changing the system itself. The approach for intelligent production control is described in the following section. Approach for Intelligent Production Control From the previous description, it can be concluded that existing production systems have rigid parameter settings and equivalent production rates or throughput times. The control architecture based on Programmable Logic Controllers has limited room for modifications to make the system highly responsive to the


457

dynamic demands. In addition to that, the signal processing units have strict working points and thresholds. Therefore production process cannot be effectively controlled in accordance with the requirements of mass customization. The mass customization responsive state-of-the-art production control solutions are dependent on humans for adaptation as well as fixing correlated problems. At the same time, expectations and requirements for production control have increased, and there is a growing need for new methods and algorithms that go beyond traditional control methodologies. The intelligent production control can be the cluster of intelligent production controller based on knowledge base (semantic nets) and other intelligent soft computing algorithms such as artificial neural networks, fuzzy logic, genetic algorithms (Melin (2004), Carlsson (2001)) and smart peripheral devices. Intelligent production automation requires the replacement of deterministic control sequences by a common negotiation schemes and individual decision algorithms. These intelligent algorithms can be considered as a base for optimization, problem diagnosis, multi-variant parameters adaptation, multi-criteria decision making, distributed decision making, distributed scheduling, resource allocation, data mining and data fusion for combining monitoring output data. Such approach helps in increasing the performance and accuracy of control and in predicting failures or disturbances in processes, machines, or devices and to fix deadlocks in overall production processes. A hybrid algorithm consisting of several cognitive algorithms can be employed as a self learning base for the system in order to be more efficient in automatic and precise configuration. The production control approach for production system in general and automated assembly cells in special is devised to achieve high adaptability, extensibility and re-usability. The intelligent approach is designed on the following points:

A highly knowledge driven platform such as technology data catalogue contains the systematic production parameters and functional correlations and coherencies;

Intelligent modeling methods deal with multi-variant parameter correlations and production control parameters considering their interdependencies;

Highly flexible and applicable centralized control set-ups take all technical specifications and functionalities for fast and smooth adaptation of new materials and processes, faster ramp up of new technologies and ensure direct and transparent quality control

458


Customer requirements and analysis The general structure of the mass customized core (combination of assembly cells) is shown in Figure 3. The customer demands are the input to this core (as shown in Figure 3) and production metrics are taken as an output in order to put a check on the core’s overall performance. The customer demands are analyzed and the line sequencing and parameters settings for assembly cells are done after consulting the TDC and computed through dedicated algorithms that are implemented in the project.

Figure 3: Requirement analysis & parameter settings.

Automated scheduling and planning module Varying customer demands and resulting high product variety increases the complexity in production scheduling and planning. Production scheduling and planning must therefore evolve and adopt new tools that help to achieve robust production control in order to satisfy customers needs efficiently. This highlights the necessity for mass customization supported automated scheduling and planning. The automated planning and scheduling module is divided into two phases. At first the assembly setup layout is defined followed by the allocation of assembly cells and their reconfiguration in the product sequencing phase. Layout definition phase One objective of the approach is to handle different costumer demands (CD) on one assembly setup. Therefore, this setup must be highly flexible and reconfigur-


459

able. In order to handle varying demands in one assembly setup, the probability of the CD variant or CD set (CDS) has to be determined. But, before analyzing the CDS, the workflow for generating CDS has to be outlined. The customer will select different concrete requirements / demands for its mass customized product from a list of production possibilities for each demand. Therefore a morphological box (Udo 2004, Dennis 2000)(as shown in Figure 4) is developed. The morphological box is known from the field of facilitation technique. In this box, different alternatives are shown against different demands. The demands are ordered in a vertical form whereas alternatives for each demand are ordered horizontally.

Figure 4: Morphological box for customer demands.

Finally, the selection of optimal option for each demand is summarized in a single customer demand set (CDS). A comfortable way to analyze the CDS is by using a Pareto chart (Figure 5). The total occurrence of each different customer demand set is illustrated in the Pareto chart. Thus, it is possible to classify CDS into A (80% occurrence), B and C products. A priority case based on highest market value/demand is generated. The layout is defined in accordance with type A products so as to handle their assembly in an efficient way. Evaluation and defining the most appropriate layout is made through intelligent, virtual simulation tools like "DELMIA QUEST" or "taraVRbuilder" (Figure 6). The evaluation metrics are the overall cycle time and the allocation efficiency of the actual allocation. These metrics are evaluated to assess the accomplishment of the objectives like the reduction of value-added time by reducing the buffer and transportation time of each part.

460


Figure 5: Specimen Pareto chart.

Figure 6: "taraVRbuilder"example.

Product sequencing phase Mass customization supported assembly layout definition is followed by utilization of intelligent algorithms for determining appropriate assembly cell sequencing and resource allocation. As each mass customized product has to be assembled


461

in the given assembly layout, therefore customer demands for each product, which are summarized in CDS, have to be analyzed and the corresponding assembly cells in the assembly setup layout have to be sequenced, allocated and reconfigured at the right time. This issue is addressed by generating an optimized work plan based on feature technology (Berger et al. 2005). This challenge is comparable with scheduling of NC programs in machine tools. There are already established methods like MRP I (Material Requirements Planning), MRP II (Manufacturing Resource Planning), KANBAN, Optimized Production Technology (Eversheim 1996) for this purpose. Berger and Kretzschmann (2007) presented an approach for ordering and scheduling machine operations with help of benchmark feature-based operations. Furthermore, the possible alternative machining operation sequences are transferred into a directed graph. With the help of an algorithm based on graph theory, the processing of this directed graph is enabled. The results acquired after the employment of the Floyd-Warshall algorithm (Jungnickel 1990) will be a "cost-efficient" sequencing order of NC operations (Hamelmann 1996). Thus, the cost metric is introduced as a general metric to assess feasibility and effort for executing machine operations at alternative machines. This methodology is transferred to the assembly domain. The workflow of the product sequencing is outlined as an example in Figure 7. The Technology Data Catalogue (TDC) provides all relevant applicable technologies for the joining and assembling tasks corresponding to the given requirements which are fixed in the Customer Demands Set (CDS). The sequencing system has manifold access to the PPR (process, product, resource) hub to get information about the produced mass customized product and other relevant information such as materials, joining technique applicability and tolerances. The sequencing system classifies assembly cells with their process parameters into the following application categories:

Capability: This assembly cell provides an assembly task that is to be executed to meet the costumer demands.

Incapability: This assembly cell is not executable to meet the costumer demands.

Non-allocatable: This assembly cell is not allocatable at the right desired allocation time; therefore this alternative branch will get a high effort.

Selected: This assembly cell with the selected parameter setting is sequenced. Going into the details; at first CDS will be mapped into the processes of the assembly layout (I) with the help of the variant process planning documents which are stored in the PPR hub. At this point, the system can determine the capable assembly cells that can be used to execute joining and assembly tasks. The desired

462


tasks influence the work piece in order to meet customer demands. The branches with the incapable assembly cells are marked as non-sequenced. The capable assembly cells with different parameter settings are then assigned. In the second step (II), the allocatability for each assembly cell is checked for the desired allocation time. If the cell is not allocated, then the effort of the corresponding branch will be increased sharply provided that the other alternatives are incapable of doing the same task. In the third step, the effort of the remaining assembly cells with concrete parameter settings is designated (III) with the help of introduced metrics. Thus, each parameter setting of each remaining assembly cell is assigned with individual effort. The application of Floyd-Warshall algorithm determines the sequence of assembly cells with parameter setting which has in total the least sum of efforts. The sequenced work plan is used to process the mass customized products in the assembly setup. The calculation of this optimal work plan can be done in real-time, because of the low runtime O(n²) of the Floyd-Warshall algorithm (Turau 2004).

Figure 7: Product sequencing work flow.

Knowledge driven system: a review for conception Efficient product and process data management is very essential in mass customization scenarios. The product development process can be defined as a crossfunctional, inter-company and market oriented process which requires constant


463

interaction based on the exchange of data to respond to customer needs (Calabrese 1999). Therefore, it involves distribution of explicit knowledge within the company through formal communication channels whereas implicit knowledge requires co-operation (Calabrese 1997). According to Van der Bij et al. (2003), higher levels of knowledge dissemination and information exchange leads to a better understanding of technology capabilities and trends, market, customer and competitors- and competitive actions, which are essential information to design the manufacturing process and determine the product features and specifications. Higher levels of knowledge dissemination and information exchange leads, as well, to a considerable decrease of technical uncertainties in the mass customization knowledge driven process. Moreover, greater levels of information exchange within an organization increase the likelihood that the new product is well responsive to the mass customization. It has been suggested that information can be exchanged formally within the boundaries of defined mechanisms, such as structured methods and formal processes; or informally; and both horizontally, e.g. cross-functional, and vertically within the organization (Perks 2000, p.182; Van der Bij et al. 2003, p.164; Calabrese 1999, p.440). A systematic management can influence knowledge sharing by implementing formal procedures for guiding information flows; moreover, there are mechanisms which can originate such process (Berends et al. 2006, p.88–91):

Diffusion: members of an organization select and communicate existing information without being oriented towards a particular problem.

Information retrieval: someone who needs a particular piece of information obtains it by asking someone who has it.

Information pooling: members of an organization working together pool information; not only factual information but also questions, suggestions and instructions are transferred.

Collaborative problem solving: new information is developed with regard to a shared problem.

Pushing: someone chooses to provide someone else with the existing information. It involves thinking that the other person needs to know something, or that certain information might be useful for his research activities.

Thinking along: someone develops new ideas with regard to someone else’s problem. It may yield new ideas, hypotheses or questions.

Self-suggestion: in the same way as one can think about someone else’s problem, one can also think about one’s own problem during an interaction.

464


The need to explain one’s own problem stimulates one to come up with new explanations, solutions, arguments and conclusions. Nevertheless, sharing knowledge among members of an organization may be a complex activity. As long as the knowledge is not shared, it can not be exploited by the organization (Choo 1996). Research based on organizational memory indicates that there are sophisticated computerized tools for externalizing personal knowledge such as idea managing tools, expert systems, among others; nevertheless, there is still a danger that tacit knowledge may remain as personal knowledge due to the complexity to share information throughout distributed or big companies (Klammer and Mathias 1998). According to Basson, Bonnema and Liu (2004), there are also many tools, such as CAD systems, that are available to manage the design information -the final product definition-. In contrast, much of earlier type of design information is currently not captured in structured or systematic ways due to its level of abstraction; it is supposed that a high level of abstraction is related to less detailed information. Further, the access to information has been solved, by large-scale computer networks, but the processing and interpretation of retrieved information remain a problem due to heterogeneity of the data. There can be distinguished three main heterogeneity problem categories (Stuckenschmidt and van Harmelen 2005):

Syntactical problems, e.g. data format heterogeneity

Structural problems, which are originated because the same objects and facts can be described in different ways using homonyms (the use of the same word with different meaning), synonyms (the use of different words with the same meaning), etc.

Problems of semantics, which refer to intended meaning of terms in a particular context or application. They occur due to the inherent context dependency of information that can be only understood in the context of their original source and purpose.

The first kind of problem can be solved through standards that are used as interfaces to integrate different information sources. Structural problems and semantic conflicts can be partially solved by one to one structural mappings; if structural mappings do not apply, like in the case of large-scale information sharing, the semantics of the information has to be taken in account in order to decide how different information items relate to each other. It has been proposed the use of ontologies as technology for approaching the problem of explicating semantic knowledge about information (Stuckenschmidt and van Harmelen


465

2005). An ontology can be defined as "a set of formal terms, usually with a hierarchical organization, with associated formal definitions that specify their relationships with the other formal terms, and a set of constraints about their use in the knowledge representation of the domain studied" (Goossenaerts and Pelletier 2001); or "a term used to refer to the shared understanding of interest" (Uschold and Gruninger 1996); or an "explicit specification of a conceptualization" (Gruber 1993a and Gruber 1993b). The knowledge representation through ontological relation in the form of semantic nets will constitute a knowledge management system which is named as Technology Data Catalogue (TDC). The TDC contains the intelligent relationship of assembly processes and product features as well as other relevant process dependent information. Technology data catalogue As mentioned before, the structured storage of application specific knowledge is an essential objective of the presented approach and the aim of gathering, retrieving, structuring, processing and sharing relevant engineering data/information in an intelligent way is addressed by the development of a Technology Data Catalogue (TDC). The TDC provides structured information about the "best-practice" settings of the corresponding assembly setup devices. Data sources are determined and a suitable knowledge representation structure is defined in order to store the relevant implicit and explicit knowledge. Afterwards, the semantically designed contents of the TDC are identified. At the end, the processing and exchanging methods of the TDC are described. Following the ideas of Stuckenschmidt and van Harmelen (2005) to overcome data format heterogeneity of the explicit knowledge sources, it is suggested to use product data standards STEP. STEP provides unambiguous, computer interpretable representation of product data (Fowler 1995). The product data is taken from different sources like CAD/CAE/CAM, Product Data Management (PDM) and Enterprise Resource Planning (ERP) systems. Furthermore, it is supported by the use of the EXPRESS language and EXPRESSI. EXPRESS (see Trippner (2000)) is a data specification language that is used to represent the structure of data and any constraints that may apply to it. It is used to define the data on which programs operate. The following Table 1 highlights the "generic resources" that are the information resources of explicit knowledge for the Technology Data Catalogue. Berger and Thiebus (2006) identify the shop floor operators as source for implicit knowledge. Furthermore, Nonaka and Takeuchi (1995) introduced "Cycle of Organizational Learning" in order to model the transfer mechanism between implicit and explicit knowledge. As a pull

466


mechanism of filling up the TDC, Berger and Thiebus (2006) used "Delphi surveys" to interview the shop floor operators in order to gather the implicit knowledge and to externalize them into explicit knowledge for the TDC.

Figure 8: Semantic net linkage. Table 1: Generic resources of STEP. Part no.

Information resources

41

Fundamentals of product description and support

42

Geometric and topological representation

43

Representation structures

44

Product structure configuration

45

Materials

46

Visual presentation

47

Shape Variation, Tolerances

49

Process Structure and Properties


467

In addition to that, Lepratti and Berger (2004) introduced shop floor oriented human-machine-interface (HMI) models for the easy robot path correction in assembly tasks in the automotive industry. With the help of this easy to use HMI (Figure 9), the suitable parameter settings for robot tasks can be enabled.

Figure 9: HMI for easy robot path correction (Berger and Lepratti 2004).

Thus, the Lepratti and Berger gathered the optimal "best-practice" parameter settings for these tasks. Furthermore, Berger and Kretzschmann (2007) outlined a new shop-floor oriented information system which enabled the operator to propose modification in planning process. This externalization proposed knowledge can be taken as one solution for filling up the TDC. Moreover, in order to workout problems of semantics in knowledge representation, an ontological system is defined (Figure 10). Specifically, this model contains: (i) A database, (ii) Translators and (iii) Filters. (i) The database with shared vocabulary (concepts, instances, attributes, etc.) allows the use of terminology most appropriate to the particular context. It solves those problems that are generated due to ambiguities among terms. Here the ontologies are built according to the following design criteria (Gruber 1993b):

Clarity: Ontology should be completely defined and should effectively communicate the intended meaning of defined terms.

468


Coherence: Ontology should be coherent: that is, it should allow inferences that are consistent with the definitions.

Extendibility: Ontology should be designed to anticipate the uses of the shared vocabulary.

Minimal encoding bias: The conceptualization should be specified at the knowledge level without depending on a particular symbol-level encoding. An encoding bias results when representation choices are made purely for the convenience of notation or implementation.

Minimal ontological commitment: Ontology should require the minimal ontological commitment sufficient to support the intended knowledge sharing activities.

Figure 10: Elements of ontological system.

(ii) Translators determine the relation among concepts coming from different sources ensuring the coherence of the exchange of data between the systems and the TDC. In the literature, translators have already been proposed and successfully tested, e.g. in Goossenaerts and Pelletier (2001). (iii) Filters enable sorting out information that better match the requested technology (Lepratti 2005; Basson, Bonnema and Liu 2004). Apart from Delphi surveys as discussed before, the second way of filling up the TDC is in the form of formal statements. In this context, the Ontological System needs to be updated consistently to reflect changes in the production environment. It must be maintained regularly by a knowledge manager that takes the constant feedback from the other departments (i.e. design and manufacturing). It is also the responsibility of the knowledge manager to verify any effect on the system before achieving any change. The TDC also fulfills other requirements, highlighted in the literature as general requirements for information systems (Basson, Bonnema and Liu 2004). some of them are for example:


469

Facilitate the extraction of information selectively, avoiding redundancies,

Improvement and increment information by recording informal statements, generated for example at the shop floor: avoiding that they remain as personal hidden knowledge,

Ensuring consistency of the information.

The last component of the TDC, a semantic net, defined in the literature as "a graphic notation for representing knowledge in patterns of interconnected nodes and arcs" (Sowa 1992), facilitates the dynamic navigation through the information, the information visualization and the sharing of information through formal procedures. The advantages of semantic nets to represent knowledge (Kendal and Creen 2007) supporting mass customization are as follows:

Adaptable method of representing knowledge because many different types of objects can be included in the network.

Relatively easy to understand because of the graphical environment.

Can be used as a common communication tool between the knowledge engineer and the human expert during the knowledge acquisition.

Figure 11 shows a semantic net developed in K-Infinity software. In the semantic net, the TDC is mainly divided in two nodes: "workpiece" and "work plan". These nodes provide accurate process parameters that can be used by the Analysis Module and Programmable Automation Controller (PAC) modules. These modules are described in the following sections. Mechanisms which facilitate the knowledge sharing (Berends et al. 2006) are also accomplished to assure that operational knowledge will not remain personal knowledge. After defining the structure of TDC, its contents are determined. As stated earlier, the approach deals with controlling of assembly setups that experience mass customization. Therefore the TDC includes information about:

Assembly operations

Applicability of parameter settings for assembly setups

Relevant measuring parameters

Parameter settings correlated with multi-variant criteria

Assembly setup sequencing

Products

Correlation between independent customer criteria Thus, the feature-technology is utilized for handling the information about assembly tasks. A feature is a technical information item which is summarized as

470


an aggregation of characteristics (geometry, relations and constrains) of a product (VDI 2003). The application scope of feature is in the product life cycle including the manufacturing. Cai (2006) introduced a feature-based machining process planning data reference model in the scope of machining. This methodology is mapped and adapted into the assembly scope. Therefore the introduction of assembly features for handling the assembly information will be enabled. The external data access of the TDC is realized with the graphical representation of the semantic network. To enhance the applicability and dissemination of gathered information, the exchange of the TDC via web-portal is useful. Hence, Cai (2006) introduced an web-enabled exchange reference model via customized web-portal solution. As already mentioned, translators are used to integrate different data source and to interlink different TDC from different domains of discussion. The advantage is to get a comfortable exchange of information among different heterogeneous data sources. Thus, the fast heterogeneous information access for handling mass customized products requirements is enabled.

Figure 11: Semantic net developed in K-Infinity.

The contents of the TDC contain the rapid change of the assembly environment and application scenarios. Therefore, the TDC must be enriched regularly with manually acquired data on one side and on the other side; the implementation of data mining and automatic algorithms is focused in order to link new application or issue in the TDC. In this regard, Berger et al. (2006) developed an extended


471

knowledge management model to enable the process and storage of shop floor operator experience. This model (Figure 12) consists of four different layers. The enhancement is the consideration of different user groups with their relevant user profiles in the scope of mass customization. The first layer named application layer realizes the interface between the operator and the system. It provides the interface and the profile-related functions to the users. The second layer called perspectives layer offers a user profile based view on the whole system. These perspectives refer to the user profiles mentioned before. The knowledge management layer consists of several parts and is of high importance. This layer is the core of the whole model, because it links different data sources (DS) from the data layer to the domain model (TDC in this case).

Figure 12: Extended knowledge management model (Berger et al. 2006).

According to this TDC, the user can save and access different experiences by linking corresponding information objects. An information object is a kind of class, to which different objects with the same structure and properties can be assigned. All information objects in the data layer (for example claims) can be assigned with relations to other information objects. The advantage of the varying views implemented in the perspectives layer is that the relations of a core domain model can be changed according to the requirements of the corresponding user profile (see additional relation in Figure 12). Finally, the implemented TDC stores and provides all necessary information about the assembly domain. In short, the

472


TDC is the consolidated container, which interlinks the implicit and explicit knowledge of assembly domain by using a semantic net for the graphical human machine interface. PAC based intelligent production control In modern production and assembly processes, conventional PLC-based architectures (Noe 2007) are state-of-the-art for controlling activities. These types of architectures exhibit certain disadvantages. The main disadvantage is that they are highly non flexible to mass customization. Modifications are not only uneconomical but cause lot of interface problems also. Moreover, highly data driven and knowledge intensive system to cope mass customization in an intelligent way cannot be created by incorporating traditional control systems like PLCs. Therefore, the Programmable Automation Controller (PAC) and use of intelligent components and modular design promise to bring essential benefits for reconfiguration and adaptation of automated production system in general and automated assembly cells in special. This adaptable and reconfigurable control architecture is fast responsive to make quick product changes (adaptability), easy programmable and reprogrammable (flexible) and resilient to manufacturing tolerances (compliant).

Figure 13: Architecture and interfaces of the new assembly cell.


473

This new intelligent approach and the resulting system comprises of the Programmable Automation Controller (PAC), the Technology Data Catalog (TDC) and Product Process Resource (PPR) modules. The PAC is also the central device to gather all data for monitoring systems of the cell. The PAC is connected to robot controller (RC) in order to influence the execution of the robot program. The central scheme is to pre-process all sensor data in a powerful CPU provided by the PAC. Thus, the PAC can calculate new robot paths by segmentation of detected curves and paths (Berger et al. 2006) in order to control the robot program corresponding to the process planning information via one interface. Furthermore, the technician can influence the intelligent process control via the new HumanMachine Interface (HMI) communication device called personal digital assistant.

Figure 14: Work flow for intelligent control.

The powerful PAC architecture offers interfaces to further databases at the PAC to store best practice decision of technicians and concerned people in order to re-use these experiences later on. The PAC represents a central process control in order to pre-process sensor data using intelligent algorithms. Therefore, the complexity of the robot programs can be reduced. One of the most vital components of the intelligent control unit is the automated monitoring and processing of assembly and joining parameters in real-time. The PAC is further connected to standard automation devices like the robots controller or PLC based manufacturing cells (as shown in Figure 13). Intelligent production control is achieved by integration

474


of production simulations; intelligent monitoring able to recognize its environment; and novel feedback loops, which link the simulation and the practical validation. To enable mass customization and prevent complication in terms of adaptation response time, the workflow for intelligent control of automated cell is systematically divided into four phases and illustrated corresponding to three components (Figure 14). The first component is the human interaction for controlling; the second component is the automatic execution of the assembly task. The third one is the selection of external interfaces and data that are used to realize the automatic execution. These interfaces are determined and are in implementation phase in EU FP6 FUTURA-project. Three different user levels (Table 2) are introduced for human interaction during automatic execution of assembly tasks. Table 2: User levels for the design concept. Profile

Acknowledge level

Operator level

Expert level

User background

Non-skilled worker

Cell operator

Process planner

Presented information

Suggested action alternative

Different alternatives

All process information

Selection of alternative

Acknowledge the presented alternative

Selection among different alternatives

Free choice and configuration

Level of guidance / wizard

High

Medium

Low

Programmable Automation Controller (PAC) The Programmable Automation Controller (PAC) concept combines programmable logic controller (PLC) ruggedness with PC functionality under open, flexible software architecture. The PAC scope is summarized in Figure 15. The PXI module (Figure 16) from National Instruments (NI) offers a lot of functionality with high rate data logging capability and high frequency measurements data handling. Employment of this module has enabled highly flexible control responsive to mass customization as it carries out automatic functions to the process control and a computer based train planning and optimization in a system. A lot of standardized hardware interfaces can be integrated in order to measure specified signals of different measuring devices. The incorporated network interfaces like Ethernet, CAN, PROFIBUS and VXI permit the communication with different hardware platforms. The integration of databases (e.g. MS Office, Oracle) and CAD systems can be made without


475

complications. The real time embedded controller and PC based graphical programming environment (LabView) enables a fast adjustment of control and monitoring parameters as well adjusting thresholds to respond mass customization in the assembly cell. Moreover, new monitoring systems and interfaces can be quickly incorporated and programd due to highly friendly programming environment.

Figure 15: PAC scope.

Figure 16: National Instruments PXI/compactPCI [NI, 2007].

Monitoring and measurement systems Real time monitoring systems play a crucial role in maintaining high quality products at high production rates and low cost. The traditional ways of offline checking of assembled components are inefficient in terms of time, material, and

476


productivity. Real-time non contact monitoring is as accurate as off-line assembly checks; yet faster, cheaper, and increase productivity, when perfected for highly volatile and dynamic mass customized applications. In the field of real-time monitoring of assembly operations, various sensors and monitoring have shown promising advantages specific to applications. The state-of-the-art assembly/ joining operations for Body in White are as under:

Welding especially remote laser welding

Adhesive Bonding

Clinching

Screwing and Riveting

The first two assembly operations i.e. welding and adhesive bonding are considered to be the most successful and well responsive to mass customization. An intelligent assembly process monitoring and fault diagnosis platform is being developed by interfacing a data acquisition and control setup with a knowledge base system. LabView based data acquisition and control software for effective monitoring of abrupt changes of critical assembly process variables and the real time fault identification system. A multi-sensor monitoring system is in process of development to monitor and enable assembly of multi-variants. The multisensory system includes combination of vision and acoustic sensors to ensure reliable and accurate measurements of a huge variety of features, shapes and surfaces and to detect materials. Application specific sensor activation approach has been devised to reduce redundancies in monitoring systems. The monitoring system is guided by the TDC according to the application scenarion. As a result, the monitoring system acts as a semi autonomous unit and will react fast to mass customization. Peripherals for exercising intelligent control Besides the control unit and monitoring system, there are some peripherals for human interaction at scalable level as well as variant flexible handling devices that can handle a huge variety of multi-variant segments. These devices are as follows:

Handheld devices:

Variant flexible fixtures and devices

Handheld Devices (PDA and / or Tablet-PC) Handheld devices (Figure 17) are becoming more important and applicable in industrial applications. The reasons are portability, technical enhancements in case of powerful CPUs, standardized communication and graphical interfaces as well


477

as specifying standard (programming) software interfaces. Thus, a normal handheld device has access via WLAN or Bluetooth to the local area network of the shop-floor. Consequently, a handheld mobile device has access to all relevant information. Furthermore, the powerful CPU and graphical interfaces enable the processing (Figure 17b) and visualization (Figure 17a) of complex data and information gathering in the assembly and production.

Figure 17: (a) Handheld device; (b) Processing of process parameters (Teamster, 2007).

The implementation of application software for handheld devices is enabled by common programming language (e.g. LabVIEW and Microsoft .NET). Furthermore, the availability of PDA HTML-Browser offers the possibility to create and use HTML-pages to access all relevant information in an easy way. The advancement in the human-machine communication is based on ontological approach as a natural language interaction system. As shown in Figure 18, natural language user command is filtered with the help of ontology. Then, the filtered and modified command is translated / mapped with translation rules into an intermediate command. After checking the syntax and semantic of the command, it is transferred into machine command and finally executed. The use of different models and rules in the knowledge base enable the step-by-step translation of the natural language into a specific machine command. This approach enables the technicians to interact the system rapidly in highly mass customized adaptation. Variant flexible fixtures and devices The conventional handling devices such as machine vises, jigs and ordinary fixtures have limited scope in handling various components of varying dimensions and shapes. The assembly cells equipped with such devices may exhibit highly

478


non flexibility when subjected to mass customization. This problem can be addressed by the use of variant flexible fixtures and devices in robot cells. Relevant resources for an assembly cell, which have to be adapted to flexible issues, are listed in the following:

Gripper

Material provision, deposit and transport

Singling units

Special equipments and tools

Figure 18: Structure of the natural language interaction system (Berger and Lepratti, 2004).

An example for the designing of a gripper concept is described here. For the forces transmitted by gripper, different active principles are known. Firstly, the force-locked connections are based on friction fit. The die friction force affects parallel to the effective areas. In addition to the pressure forces, further shear forces fix the effective areas. The transmitted forces depend on the composition and areas of the effective area. Secondly, form-locked connections fix the position and orientation of the work piece by using reaction force on the effective areas to bear the weight and inertial force of the work piece. Finally, a positive substance joining is being realized by using adhesion and cohesive power between connecting effective areas to ensure strong forces. A bonding agent supports this effect.


479

The disadvantage of employment of specified and individual bonding agents is the difficulty in separation of holding and held components. Comparing the discussed force transmitting concept, the positive substance joining concept is not realizable for grippers. The other introduced concepts for force-locked and form-locked connections are practicable. Nevertheless, the use of form-locked connection results in variant independent application. Achievements and features: Mass customized automated assembly cells The main characteristics of the mass customization responsive automated approach for automated assembly cells are:

Interactive man-machine interface acting as assembly assistance system

Use of a universal data catalogue and standard interfaces

Production knowledge management by dynamic data processing

Work planning module (WPM) comprising work planning features by using algorithms from the Graph Theory

Modular extensible and scalable (joining and sensing) devices by generic interface solutions

Enabling real time control applications through the PAC technology

Monitoring of risk and capability features by error detection and maintenance

Easy implementation into production lines due to generic solutions

Use of variant flexible fixtures and devices

Standardization of data transfer Interfaces (e.g. ODBC, dll)

Multi-dimensional sensor data transfer

Highly frequent (real-time) measurement and state monitoring capabilities

Self-adaptation of parameters

Automated determination of process parameters borders as well as monitoring strategies

Automatic correlation of multi-variant parameters

Direct incorporation of operational feedback

Technical Realization At present, the approach for mass customized responsive automated assembly cells through intelligent production control is in realization phase. For dealing mass customization in a systematic way, the modularity and scalability of variants

480


are considered so as to handle huge variety of products. Thus, the assembly of Body in White (BIW) in the shop floor will be developed with the help of various scalable modules. The fast sequencing and reconfiguration of assembly cells is under practical realization to enable assembly of modular and scalable Body in White segments. The reconfiguration in terms of handling is enhanced by means of variant flexible fixtures. The bench mark part i.e. side frame component, is shown in Figure 19. The complete approach is being implemented in a laboratory set-up (shown in Figure 20) considering cold joining process. However, the work will be extended to other joining processes in near future. The current experimental setup consists of the following components. Hardware setup:

Central processing unit PAC

Handheld mobile device MOBIC

KUKA Industrial Robot KR 15

Sensor technology for measuring tasks

Software setup:

Process layout with taraVRbuilder

TDC for assembly task with K-Infinity

Monitoring and processing with LabView

Process scheduling with in-house developed planning tools

Figure 19: Specimen side-frame.

The robot-based experimental setup is focusing on cold joining methods like adhesive bonding with one or two component bonding material. Consequently, the


481

TDC includes technology parameter settings for adhesive joining and the modular and scalable segments of benchmark part.

Figure 20: Laboratory set-up for joining and assembly tasks.

The high operability of the selected PAC as central processing is outlined in Figure 21. In this case, the PPR hub (CATIA/DELMIA V5) is interlinked with PAC by a dll-interface which is completely implemented. The database (TDC) uses an ODBC interlink to exchange the corresponding data. Furthermore, the PAC is connected through standard multi-channel I/O to the assembly cell components and monitoring devices. The process scheduling tool is also in implementation and testing phase. The experimental validation is an ongoing process; therefore the approach is undergoing experimental validation in research and development projects. One of the projects is the EU Sixth Framework Program "FUTURA: MultiFunctional Materials and related Production Technologies integrated into the Automotive Industry of the Future" (FP6-2004-NMP-NI-4-026621), which is pursuing those innovative concepts that facilitate the integration of multifunctional materials in the automotive industry according to mass customization demands. The objective is the designing and assembly of modular and scalable as well as hybrid body and chassis structures. The scalability enables accommodation of sizes and design requirements for different models, while the modularity

482


concept facilitates an easy derivation of variants and a flexible assembly and/or joining process.

Figure 21: Interface schematics for the MC Intelligent Production Control.

Conclusions In this chapter, the intelligent approach for controlling of automated assembly cells has been outlined. This approach is developed in order to cope mass customization in automotive industry and to enhance its competency. The incorporation of intensive knowledge driven system in automated assembly cells will enable rapid adaptations to handle highly dynamic volumes and changing variants. Mass customized responsive intelligent production control is meant for automatic scheduling of highly dynamic production volumes in terms of type and quantity whereas conventional production control has general scheduling solutions based on fixed production volumes. The incorporation of knowledge base into the intelligent production control approach will identify cross correlations of multivariant parameters including automatic setting of thresholds. This is far better than the conventional approach where parameter settings are made after experimental validation and statistical determination. Systematic acquisition and implementation of functional requirements and identification of specifications of peripheral devices at shop floor level through intelligent semantic net methodologies will enhance knowledge level at assembly cell level. There will be easy to incorporate


483

new materials, features and hybrid assembly/ joining processes due to technology data catalogue. The standardized data acquisition, filtering and processing of CAD and simulation modeling data based on dll and ODBC interfaces will enable a generic solution for handling multi-variants in the same setup, thereby reducing redundancies and rapid adjustments. Employment of multi-sensor based monitoring systems, graphical programming platform and intelligent computing algorithms will yield wide spread functionality and flexibility conducive for mass customization.

References Basson, A.- H.; Bonnema, G. M.; Liu, Y. (2007): A flexible Electro-Mechanical Design Information System. In: Tools and Methods of Competitive Engineering, Imre Horváth and Paul Xirouc hakis (ed.), v.2: 879- 889, Millpress Rotterdam Netherlands, 2004. Berends, H.; Van der Bij, H.; Debackere, K.; Weggeman, M. (2006): Knowledge sharing mechanisms in industrial research. R&D Management. 36(1): 85–95. Berger, U.; Kretzschmann, R. (2007): Development of a holistic guidance system for the NC process chain for benchmarking machining operations. In: Proceedings of the 12th IEEE Conference on Emerging Technologies and Factory Automation, Greece, September 25–28, 2007. Berger, U.; Kretzschmann, R. (2007): Aufbau einer werkstattgerechten Informationsversorgung. IndustrieManagement, 23(4). Berger, U.; Kretzschmann, R. (2007): Aufbau einer werkstattgerechten Informationsversorgung. IndustrieManagement: Zeitschrift für industrielle Geschäftsprozesse, 23(4). Berger, U.; Kretzschmann, R.; Vargas V. (2007): Intelligent production monitoring and control for mass customization of automated manufacturing cells in the automotive industry. In: Proceedings of the 4th MCPC 2007 World Conference on Mass Customization and Personalization, USA, October 07–10, 2007. Berger, U.; Kretzschmann, R.; Cai, J.; Weyrich, M. (2006): Toward the Knowlegde-based Enterprise. IFIP International Federation for Information Processing. 183(Jan): 351–361. Berger, U.; Noack, J.; Kretzschmann, R. (2006): Automatic Generation of Robot Paths from CAD-Data Based on Linear and Circular Approximation. Proceedings of the 4th IFAC-Symposium on Mechatronic Systems. Berger, U.; Thiebus, S. (2006): Wissensmanagement in der Planungsphase. In: Industrie-Management: Zeitschrift für industrielle Geschäftsprozesse, 22(6). Berger, U.; Thiebus, S.; Kretzschmann, R. (2006): Knowledge Management for Ramp-up: Approach for Knowledge Management for Ramp-up in the Automotive Industry. Proceedings of the 9th IFAC Symposium on Automated Systems Based on Human Skill and Knowledge, Nancy, France, 23–25 May, 2006. Berger, U.; Cai, J.; Weyrich, M. (2005): Ontological Machining Process Data Modeling for Powertrain Production in Extended Enterprise. Journal of Advanced Manufacturing System (JAMS). 4(1): 69–82. Blecker, T.; Abdelkafi, N.; (2006): Mass Customization: State of the Art and Challenges, in: Mass Customization: Challenges and Solutions, edited by T. Blecker et al. New York: Springer, 1–25.

484


Blecker, T.; Abdelkafi, N.; Kaluza, B.; Friedrich, G. (2006): Controlling variety-induced complexity in mass customization: a key metrics-based approach, International Journal of Mass Customization. 1(2–3): 272–298. Buede, D.M. (2000): The Engineering Design of Systems: Models and Methods, Wiley Series in Systems Engineering and Management by Wiley. Cai, J. (2006): Development of a Feature-based Machining Process Planning Data Reference Model for Web-enabled Exchange in Extended Enterprise. Ph.D. Thesis. Calabrese, G. (1999): Managing information in product development. Logistics Information Management, 12(6) 439-450. Calabrese, G. (1997): Communication and co-operation in product development: a case study of a European car producer. R&D Management. 27(3): 239–252. Carlsson, C.; Walden, P. (2001): "Intelligent Systems and soft computing" from Proceedings of the 34th Hawaii International Conference on System Sciences. Choo, C.W. (1996): The knowing organization: How organizations use information to construct meaning, create knowledge and make decisions. International Journal of Information Management. 16(5): 329–340. Dransfeld, S. (2007): Measurement and Supervision in Automated Production. Ph.D. Thesis. Eversheim, W. (1996): Organization in der Produktionstechnik- Band 1 Grundlagen, 3. Auflage, VDIVerlag Düsseldorf. Fowler, J. (1995): STEP for Data Management, Exchange and Sharing. Technology Appraisals Ltd., Great Britain. Goossenaerts, J.B.M.; Pelletier, C. (2001): Enterprise Ontologies and Knowledge Management. In: K.-D. Thoben, F. Weber and K.S. Pawar (ed.) Proceedings of the 7th International Conference on Concurrent Enterprising: "Engineering the Knowledge Economy through Co-operation" Bremen, Germany: 281–289. Gruber, T.R. (1993a): A Translation Approach to Portable Ontology Specifications. Knowledge Acquisition. 5(2): 199–220. Gruber, T.-R. (1993b): Toward Principles for the Design of Ontologies Used for Knowledge Sharing. International Journal Human-Computer Studies. 43: 907–928. Hamelmann, S. (1996): Systementwicklung zur Automatisierung der Arbeitsplanung, VDI Verlag Düsseldorf. Jungnickel, D. (1990): Graphen, Netzwerke und Algorithmen, Wissenschaftsverlag, Mannheim. Kendal, S.; Creen M. (2007): An Introduction to Knowledge Engineering. Springer, USA. Klamma R.; Mathias J. (1998): Driving the Organizational Learning Cycle: The Case of Computer-Aided Failure Management. In: Baets, Walter R. J. (ed.): Proceedings of the 6th European Conference on Information Systems (ECIS'98), Aix-En-Provence, France, Vol. 1. Granada, Euro-Arab Management School. Kuhn, W., Eversheim, W. (2002): Schneller Produktionsanlauf von Serienprodukten, Ergebnisbericht der Untersuchung `fast ramp-up`, Dortmund: Verlag Praxiswissen. Lepratti, R. (2005): Ein Beitrag zur fortschrittlichen Mensch-Maschine-Interaktion auf Basis ontologischer Filterung. Logos Verlag Berlin. Lepratti, R.; Berger, U. (2004): An Interaction System for Easy Robot Path Correction. In: Proceedings of the 7th IFAC Symposium on Cost Oriented Automation, Ottawa, Canada, June 7–9, 2004, CD-ROM. Lepratti, R.; Cai, J.; Berger, U.; Weyrich, M. (2004): Towards the knowledge-based Enterprise. In: Proceedings of ICEIMT 2004, Toronto, Canada, 2004: 351–361.


485

Melin, P.; Castillo, O. (2004): "Soft Computing for Intelligent Controlof Nonlinear Dynamical Systems", International Journal of Computational Cognition. 2(1): 45–78. National Instruments, 2007, www.ni.com/swf/presentation/us/pac/, reference;17- 12-2007. NISTEP (2005): Nistep Report No. 99, Science and Technology Foresight Center National Institute of Science and Technology Policy (NISTEP) Ministry of Education, Culture, Sports, Science and Technology. Nonaka, I.; Takeuchi, H., (1995): The Knowledge-Creating Company, Oxford Univ. Press. New York. Park (2006): Rapid Manufacturing Today, www.rm-platform.com, Rev. 2007-08-31 Perks, He. (2000): Marketing Information Exchange Mechanisms in Collaborative New Product Development, the Influence of Resource Balance and Competitiveness. Industrial Marketing Management. 29(2): 179–189. Pine 11, B.J. (1993): Mass Customization: The New Frontier in Business Competition, Boston, Massachusetts: Harvard Business School Press 1993. Ruohonen, M., Riihimaa, J.; Mäkipää, M. (2006): "Knowledge based mass customization strategies:cases from Finnish metal and electronics industries", Int. J. Mass Customization. 1(2/3): 340–359. Silberglitt (2006): Global Technology Revolution 2020. Rand Corp. Sowa, J.F. (1992): Semantic Networks. In: Shapiro Stuart C., Wiley (ed.) Encyclopedia of Artificial Intelligence. Stuckenschmidt, H.; van Harmelen, F. (2005): Information Sharing on the Semantic Web. SpringerVerlag Berlin Heidelberg Germany. Teamster (2007): www.teamster.se, reference; 17-12-2007. Trippner D.; Anderl, R. (2000): STEP: Standart for the Exchange of Product Model Data, 1. Auflage, Teubner Stuttgart Leipzig. Tseng, M.; Jianxin, M. J. (2001): Mass Customization, in: Gavriel Salvendy (Ed.): Handbook of Industrial Engineering: Technology and Operations Management, 31d Edition, New York et al.: John Wiley & Sons, INC. 2001: 684–709. Turau, V. (2004): Algorithmische Graphentheorie, 2. Auflage, Oldenbourg München. Lindemann, U. (2004): Methodische Entwicklung technischer Produkte: Methoden flexibel und situationsrecht anwenden, Springer-Verlag. Uschold, M.; Gruninger, M. (1996): ONTOLOGIES: Principles, Methods and Applications. Knowledge Engineering Review. 11(2). Van der Bij, H.; Song, M.X.; Weggeman, M. (2003): An Empirical Investigation into the Antecedents of Knowledge Dissemination at the Strategic Business Unit Level. Journal of Product Innovation Management. 20: 163–179. VDI: Information Technology in Product Development, VDI Guideline 2218, VDI, 2003.

Author Biographies Prof. Dr.-Ing. Ulrich Berger is a full Professor in Centre of Automation in Brandenburg University of Technology Cottbus Germany. He holds a Diploma degree of Mechanical Engineering from the University of Stuttgart and a Doctorate degree in Engineering Sciences from the University of Bremen. His lecture topics cover innovative manufactur-

486


ing technologies, automation and robotic systems. Prof. Berger pursued various research activities at national and international level. His professional career embraces several responsible positions in the automotive sector. He holds appr. 90 technical and scientific papers in the field of manufacturing technologies, advanced machinery, factory automation and product development. He performed several contributions to renowned conferences as chairman, speaker or author. Prof. Berger is a full member of the German Association of Engineers (VDI) and a guest/corresponding member of the IEEE, ASME and SPIE. His scientific interests focus on Digital Manufacturing Systems, New Machinery and Control Solutions. Contact: www.aut.tu-cottbus.de | [email protected]

M.Sc. Sarfraz Ul Haque Minhas is a PhD candidate and research fellow at the Brandenburg University of Technology, Centre of Automation, Cottbus Germany. He is involved in European Commission Research Projects in the area of intelligent production monitoring and control. He holds B.Sc. Mechanical Engineering Degree from the University of Engineering and Technology Lahore and M.Sc. Mechatronics degree from Hamburg University of Technology. He served as Lecturer in faculty of mechanical engineering in the University of Engineering and Technology Lahore and research assistant in University of Armed Forces Hamburg Germany. He has participated in several projects related to topics of real time monitoring systems, intelligent production control, non linear dynamics, management information systems and mass customization. Contact: www.aut.tu-cottbus.de | [email protected] M.Sc. Ralf Kretzschmann is a PhD candidate and research fellow at the Brandenburg University of Technology, Chair of Automation Technology, Cottbus Germany. He is working for European Commission projects in the area of production planning and manufacturing simulation. He holds a Master of Science in Information and Media Technology from Brandenburg University of Technology Cottbus. His scientific interests focus on new robot application systems and process planning principally for the automotive industry. He has also participated and presented articles in various conferences and workshops related to these areas. Contact: www.aut.tu-cottbus.de | [email protected] M.Sc. Veronica Vargas is since June 2007 a PhD candidate and scholarship holder from the International Graduate School C at the Brandenburg University of Technology Cottbus, Germany. Her main topic is the development of a process chain for the production of compressor parts with functional surfaces. Her research interests are manufacturing systems for compressor parts, axiomatic design, knowledge management, semantic nets and ontologies. She has participated in conferences on Intelligent Systems and Automation, and Cost Effective Automation in Networked Product Development and Manufacturing. Prior to her current activities, she obtained B.Sc. degree Industrial and Systems Engineering degree at the Instituto Tecnológico y de Estudios Superiores de Monterrey (ITESM), Mexico City; and, as a DAAD scholarship holder, her M. Sc. Technology and Innovation Management degree at Brandenburg Technical University Cottbus and University of Applied Sciences Brandenburg, Germany. Contact: www.aut.tu-cottbus.de | [email protected]

4.4

A Prioritization Algorithm for Configuration Scheduling in a Mass Customization Environment Ashok Kumar Department of Management, Grand Valley State University, USA Frank T. Piller Technology and Innovation Management Group, RWTH Aachen University, Germany

Mass customization (MC) as a business strategy seeks to deliver highly customized product to customers at affordable prices that are consistent with mass production efficiencies. Despite a significant volume of research in mass customization, a severe paucity of work in quantitative areas related to operations management, such as scheduling, inventory control, distribution systems, statistical process control, etc. has been well documented. In this chapter, we begin to fill this gap by proposing a methodology for scheduling the production of an arbitrary number of configurations of a product when the production budget and time are limited. A specific contribution of this chapter relates to the development of three measures of the value associated with each configuration of the product. These measures are more general than the profit motive usually employed in scheduling configurations. These measures depend on cost proportion, profit proportion, and a hybrid of these two measures. Using these value measures, we formulate a mixed integer linear programming model that would yield an optimal sequence of configurations to maximize the total value of the production over a given period. Dual constraints on budget and production time make the problem NP Hard in strong sense. These are reducible to 2-dimensional Bin packing or Knapsack problem after elimination of certain constraints. Given the dynamic nature of configuration demands and constantly evolving system state, an efficient heuristic solution with tight bounds that can be developed quickly is considered preferable over an optimal solution that takes long time and substantial computer resources to develop. Accordingly, two heuristic solutions are constructed that are quick as well as efficient.

Introduction Mass customization (MC) as a business strategy offers the promise of a unique and powerful generator of sustained strategic advantage. According to the existing literature, MC clearly provides significant strategic advantage in two mutually conflicting strategic priorities – price and customization. Furthermore, Piller and 487

488


Kumar (2006) have argued that by virtue of the cellular/modular manufacturing structure associated with the implementation of the MC strategy, firms can gain additional competitive value in quality and responsiveness. Finally, with customer co-design as an integral aspect of mass customization, customer service and satisfaction improve significantly under this strategy. Mass customization strategy, therefore, provides significant competitive advantage in all five competitive dimensions simultaneously – a unique accomplishment. According to Kumar et al. (2007), the research volume as well as the application volume of mass customization strategy grew exponentially during 1993 through 2004 after the incubation period was over in 1993. Since 2004, the trend has flattened somewhat although still remains impressive. However, the research that addresses optimization of operational modules such as scheduling, inventory management, stochastic and deterministic distribution systems, remains in acute shortage. Once again, Kumar et al. (2007) observe in the context of MC that, "there is a clear paucity of quantitative modeling and optimization routines that provide decision support to manufacturing and service systems as well as supply chains." This chapter is an initial attempt to fill this void. In this chapter, we propose a model, formulation, and two heuristic solutions to the basic configuration scheduling problem in a mass customization environment. Considering the fluid and dynamic nature of such environment, it is necessary that such solutions be developed quickly and should be in close proximity of the optimal solution. Considering the horizon of a single period of appropriate length, we develop a heuristic solution to sequence the entire demand of product configurations that maximizes the total value of production in that period. The term value here is used as a generalized measure of a desired outcome such as profit. In addition, a formulation is proposed along traditional lines that considers maximizing profit per se as the objective. The value of a configuration is determined by the total value of its variants. The value of a configuration is represented as a sum total of the values contributed by its module-variants and the standard portion of the product. Three different metrics are defined that reflect the value contribution of each variant and consequently for each configuration. The first metric uses the proportion of total cost contributed by a variant; the second measure considers proportion of profit contribution by a variant; and the last measure is constructed by integrating the two value measures just described. In all cases, the measure is so defined so that larger-the better holds true. This allows transformation of the objective function to maximization of profit, should that be the required objective, as is traditionally the case.


489

In order to make the model and solution as flexible as possible, we relax the condition that the entire demand of the product received in a certain period must be produced in a single period. Some purists may argue that this is inconsistent with the mass customization environment and may violate the lead-time promises. There is a simple response to this argument in the way the model is formulated. Realistically, no one can guarantee that given the production capacity, no matter how large, all demand in every period can be produced within the limits of available capacity. Furthermore if that is true, i.e., the entire demand can be produced within a given period, there is no issue; the problem solution is trivial and no sequencing is necessary. Of course, we also stipulate a budgetary control and that may need sequencing of configurations. When that is the case this model will come handy. Finally, we do deal with the interdependency between periods created by unfulfilled demand. Not only we reduce the value contribution of such demand in the current period but we increase their priority to produce by increasing the weight of such demand in the next period. Finally, this formulation also allows compulsory production of unfilled demand in any given period by simply adjusting the budgeted dollars and time so that an appropriate amount is first allocated to such demand that must be produced. Considering that three value measures that can be used to customize the priority of scheduling and with provision of allowing units from previous period to be produced on higher priority and allowing some units to be carried over to the following period we believe that we present a model here that is extraordinarily flexible and practical. In the next section, we provide a brief background on mass customization and the solution methodology. Background and Seminal Business Applications Mass customization strategy: Selected successful applications Coined in 1987, MC picked up significant momentum and currency in recent years. Companies such as Dell, Nike, Adidas, Land’s End, TC2, McGraw Hill, Motorola, Hertz, and numerous others attribute significant gains in their financial and strategic positions to the deployment of MC as their strategy. Owing to its unique potential in creating an alignment between two rival competitive priorities – price and customization– MC is being employed as a competitive strategy by a number of innovative start-ups with great success. Ping Fu, the CEO of GeoMagic, for example, was just named "Entrepreneur of the Year-2005" by the business magazine Inc. A recent article in U.S. News & World Report (November 21) underscores the tremendous movement registered by the MC strategy and its impact on customer expectations over the last few years: "It used to be [that]

490


consumers wanted something state-of-the-art," says Wharton professor of marketing Barbara Kahn. "Now they want something tailored to them." Indeed, Brand Keys, a research firm that studies customer loyalty, found that customization is 30 percent of what draws a person to a brand today as opposed to 6 percent in 1997. Among numerous successful mass customization stories described in this article, Scion’s (Toyota) story places the power of mass customization in perspective. Literature review As indicated previously, there is very little work that addresses production scheduling or supply chain distribution policies suitable for mass customization environment. We discuss briefly the salient literature in mass customization below. MC was initially spawned in industry. The academia had to catch up with practice and its early literature reflects that. To-date the MC literature is still essentially descriptive and mostly qualitative with little quantitative treatment limited to certain modules of MC knowledge. Example of such modules will include delayed differentiation (Tayur, Ganeshan, and Magazine, 1999, Lee and Tang 1997, Garg and Tang 1997, and risk pooling ((Eppen and Scgrage (1981), Schwarz (1989), Kumar et al. (1995)), have received significant quantitative exposure in other contexts. The decision rules that are provided are mostly empirically-based rules of thumb. MC was initially conceived as a production strategy. In the archival literature Davis (1987) represents the inception of the MC idea and the origination of its terminology. The book by Pine (1992) pioneered in explaining to businesses how MC is useful for competitive and financial advantage and indeed provided a practical footing from idea to shape the adopting of an MC strategy. It also, laid down the markers when it is useful to transition from a cost-based MP strategy to a cost + customization based MC strategy. The papers by Pine (1993) and Pine et al. (1993) are seminal papers in MC that constituted the basis for considering mass customization as a viable strategy and also as a bonafide area of research. Kotler, (1989) was the first to give MC a marketing dimension by considering it as an alternative to traditional mass marketing strategy. Kotha (1995) explored the dynamics of pursuing both MP and MC strategies in the context of the National Bicycle manufacturing company. The competitive impact of both strategies was analyzed for a given company and an analytical framework was developed to position a company on the MC to MP continuum so as to maximize its strategic financial performance. It also laid down (rather broadly) the capabilities needed for successful implementations of MC. An interesting early bridge between MC and Supply Chain Management was provided by Salvador et al. (2004). An interesting bridging between MC and


491

postponement as an operational method to move toward mass customization strategy was provided by Van Hoek et al., (1999). Alptekinoglu (2004) used a game-theoretic approach to provide theoretical underpinnings in analyzing MC vs. MP. Piller (2005) is a trend-analyzer and a trend-setter paper. Through 12 propositions, this paper takes stock of what has been done by industry (very little) and what remains to be done (a lot) in making MC an effective business strategy. The first paper to survey the MC literature appears to be by Da Silveira et al. (2001). Among other things it discussed the notion of MC Enablers and their impact on the development of production systems. Various approaches to implementing MC are compiled and classified and future research directions are outlined. Kumar et al., (2006) provide an epistemological review of the literature and Lee and Tang (1997) provide the taxonomy of delayed differentiation which is an integral part of MC strategy. They capture the costs and benefits associated with" a strategy to "redesign products and processes so as to delay the point of product differentiation." They "apply this model to analyze some special cases that are motivated by real world examples." The first practical attempt at developing MC performance metrics appears to be due to Kumar (2004). The given metrics compute the degree of customization and the degree of mass production propensity. When combined the results provide a practical effectiveness measure for a strategy of MC implementation. A practical application based on PC configurations is provided. Zipkin (2001) represents one of the few papers outlining MC’s limitations. Combining the customers' perspective on customization and the capabilities needed for implementing mass customization, he develops the case that mass customization is not always the best strategy, nor can it be adopted by all businesses As has been shown in Kumar et al. (2006) research on MC has grown steadily and at an exponential rate since the concept was introduced. It has progressed independently and piecemeal within different disciplines without a comprehensive framework or an exhaustive model that brings these research perspectives together in a coherent way. To date, we have not seen a concerted effort to bring together different business perspectives to enhance our understanding of MC as a competitive strategy. The knapsack problem The problem addressed in this chapter, namely, the determination of the sequence and quantity of each configuration to be produced in a given production period can be shown to be equivalent to a class of problems called the Knapsack problems. The goal of the exercise is to maximize the total value of the configura-

492


tions produced, given the measure of the "value" of each configuration. The problem as formulated requires two resource constraints to be complied with, one with the budgetary cap and the other with the total time available. This formulation is consistent with what are known in operations research/management science literature as 2-dimensional knapsack problems. The variants of the knapsack problem with significant body of research available include 0-1 knapsack problems (KP), single-dimensional (SD), and multidimensional problems (MD). The configuration sequencing problem addressed here is equivalent to the two-dimensional (2D) knapsack problem, which is subset of MDKPs. The KP is among the oldest in operations research/management science area and has applications virtually in every field of business. Specific applications available in literature include capital budgeting (Pendharkar and Roger 2006, Lu et al. 1999, Weingartner and Ness 1967), project selection (Kleywegt and Papastavrou 2001), auctions (de Vries and Vohra 2003), distribution (Akcay and Xu 2004), and mass customization (Kumar et al. 2007). For excellent surveys of the state of art in Knapsack problems and heuristic solutions, we recommend reading Lin (1998), Freville (2004), and Kellerer et al. (2004). The multi-dimensional knapsack problem (MDKP) can be formulated as follows: n

Max

∑V

Yj

j

j=1 n

s.t.

∑R

ij

Yj ≤ L i i = 1, 2, ..., m; j = 1, 2, ..., n

j=1

Yj ≤ U j j =1, 2, ..., n and Yj ≥ 0. The goal of the generalized or multi-dimensional knapsack problem (MDKP) is to find the quantity Yj of item j (j =1, 2, …, n) out of a total available quantity Uj that should be filled in an m-dimensional knapsack such that total value of items in knapsack is maximized. Each item j occupies Rij units of dimension i, i =1, 2, …, m and has a value Vj. When Yj is restricted to a value of 1 or 0, this is called a multi-dimensional knapsack problem. For i =1 and 2 the problem is called, singleand two- dimensional knapsack problem respectively. This problem, including the 0-1 knapsack problem are proven to be NP-Hard in strong sense; even the singledimensional problem is shown to be NP-hard (Garey and Johnson 1979, Lin


493

1998). Due to the NP-hard nature of knapsack problems, most solution approaches have been heuristical in nature. There have been but a few optimal solution approaches focusing on a Dynamic Programming or Branch and Bound solution approach. These may need total enumeration in the worst case (Kolesar 1967, Shapiro 1968, Greenberg and Hegerich 1970, Cabbot 1970). There is considerable amount of research in literature dedicated to 0-1 knapsack problems (KP) while relatively little attention is paid to MDKPs. To the extent that all MDKP’s can be transformed into 0-1 problems, the solution methodology for 0-1 KPs applies to MDKPs as well. Unfortunately, if the number of variables or constraints is large, the conversion adds an exponential number of variables in the size of the dimensions, taxing the computational efficiency considerably, thus defeating the very purpose of why heuristic solutions are considered or developed in the first place. Therefore, the heuristics that perform well for 0-1 knapsack problems, do not necessarily do so for MDKPs in terms of solution time. Hence, dedicated heuristics are needed for MDKP’s. To our knowledge, there are few efficient heuristics suggested in the literature that solve MDKPs: Kochenberger, McCarl, and Wyman (1974), Pirkul and Narasimhan (1986), and Akcay, Haijun and Xu (2007), and Dahl and Foldnes (2006). Kocheberger et al. in fact extended the work of Toyoda (1975). The primal gradient algorithm of Toyoda (1975) was an efficient variant of Senju and Toyoda (1968) which employed a dual gradient method. Weingartner, and Ness (1967) were among the first to suggest an efficient heuristic solution ("Look Ahead") for a SDKP and demonstrated the application of their heuristics to capital budgeting problems. Akcay, Haijun and Xu (2007) propose a greedy heuristic solution, which is primarily intended for the general multi-dimensional knapsack problem. Their heuristic differs from the existing greedy heuristics in two aspects: First, it uses the "effective capacity," defined as the maximum number of copies of an item that can be accepted if the entire knapsack were used for that item alone, as a prioritizing variable. Second, their heuristic adds decision variables to the solution in batches (as opposed to one by one) and consequently improves computational efficiency significantly, especially for large-scale problems. Their heuristic has a complexity of polynomial time or pseudo-polynomial time depending on whether α, a flexible factor in their algorithm, equals 1 or is a fraction. Dahl and Foldnes (2006) solve a wireless telecommunication problem that has the structure of MDKP with additional restrictions. These restrictions relate to the set of specific knapsacks to which items can be assigned. They use both randomized and deterministic heuristics that relax the assignment variables from binary to linear, while maintaining the bounds on these variables at 0 and 1.

494


Forrest et al. (2006) solve a generalized MDKP which imposes additional color constraints on MDKP. They use Dantzig-Wolfe decomposition to develop a tight upper bound. Their solution methodology, though complex, is quite efficient for large size problems; it takes less time than it takes to solve the LP relaxation of the original formulation while yielding close to optimal solution. Chekuri and Khanna (2006) present a heuristic solution for a multiple knapsack problem which is polynomial time. It should be pointed out that multiple knapsack problem (MKP) is less general than MDKP. MKP is a simple generalization of single knapsack problem in that it has more than one bins; each with a known but different capacity. Development of Value Metrics for Product Variants In this section, we develop three measures that are in direct correspondence with the profitability potential of a specific product variant. In order to maintain consistency of terminology, let us first define several terms. Note that these terms apply for a manufacturing company that operates in a mass customization environment. We use the example of a product module. Product Module: In an idealized mass customization setting, products are designed in a modular fashion such that each module is dedicated exclusively to serve one function (or embody one aspect) of the product. For instance, RAM, CPU, Video Card, Hard Disk are all modules of a computer that serve one function each. Module Variant: Each module of a product has several variants; each variant provides a different level of performance of the function to which the module is dedicated. Thus a RAM may come in many memory sizes: 128 MB, 256 MB, 512 MB, 1 GB, etc. The variants of a module have essentially the same external shape and size generally and are assembled in the same fashion, i.e., are designed for interchangeable assemblies. For instance all RAMs, CPUs, Video cards fit the same spaces or slots in the mother board of all computers in an identical fashion. Sometimes the variants may be differentiated by a characteristic or aspect. Laptops, PCs, and cars, for instance, come in different colors. Product configuration A product configuration is a unique combination of module variants assembled on to a standardized product. Thus, a certain configuration of a product would differ from another in at least one module variant, and consequently in the performance


495

level of at least one function. A given configuration of a PC may differ from other in one or more of the following functionalities: hard disk size, random access memory size, video card size, monitor dimensions, etc. We now define notations that will be used in developing the metrics that serve as a surrogate measure for the composite value of a product and eventually an algorithm for sequencing the production of these configurations given production parameters and constraints. Notation m

=

T t Im

= = =

Index for modules used in production, m = 0, 1, …, M. Here m = 0 represents the standard part of the product used in every configuration. Total number of distinct configurations Index for configurations, t = 1, …, T. Number of variants of the module m; where each variant of module m is indexed as Im, I = 1, 2, … I.

K0

=

V0

=

Kim

=

Vim Ntim

= =

Itim

=

Xim

=

Yim

=

αt D

= =

Setup cost for the standard portion of the product which is common to all configurations. Variable cost for the standard portion of the product which is integral to all configurations. Setup cost for producing the ith variant of mth module, regardless of quantity > 0. Variable cost for producing one unit of the ith variant of module m. Number of units of the ith variant of module m used in configuration t. if the variant is not used, N tim = 0. Also, assumed that only one standard part is used in each configuration. Indicator variable = 1, if the variant i of module m is used in configuration t, 0 otherwise. Set up time for producing any number of units of the ith variant of module m. im = 0 for the standard product. Variable time for producing one unit of the ith variant of module m. im = 0 for the standard product. Fraction of total demand for configuration t, t = 1, 2, …, T. Total demand of all configurations

Then: m=M

T

=

∏ (I m =1

m

+ 1) = Total Number of configurations of a product.

496


The addition of 1 implies that there could be a configuration that does not use module m. For instance, you may buy a PC without a wireless card. Mass customization metrics In this section we develop three measures of "value" of a given product configuration. In computing such a value of a configuration, we use the common principles: larger the better for profit and smaller the better for cost. We would first compute the costs and profits associated with a single batch of production of each configuration. The cost of production of the ith variant of the mth module = T   K im + ∑ α t D N tim Vim if α t > 0 Cim =  t =1 0 Otherwise

(1)

Therefore, the total cost of production = Sum of the costs of each variant that is produced and the cost of standard product.

C = K 0 + D V0 +

T

Im

M

∑∑∑ I

tim

C im since D > 0

(2)

t =1 m =1 i =1

If Rt is the revenue per unit of configuration t, then the total profit is given by (since the orders for αtD units of configuration t have been already received): T

Π=

∑α

t

D Rt − C

(3)

t =1

We now propose three metrics that capture and reflect the "value" of all variants of modules. For reasons of consistency in all three metrics, and to use them identically in the formulations later, we define them in such a way that they display a monotone, smaller-the-better characteristics.

497


The cost-proportion based metric for the ith variant of mth module (CPim):

CPim =

T   I tim  K im + ∑ α t D N tim Vim  t =1   T

M

Im

(4)

K 0 + D V0 + ∑∑∑ I tim Cim t =1 m =1 i =1

The profit-proportion based metric for the ith variant of mth module (PPim): The metric developed in this section represents the proportion of the total profit contributed by a variant as a fraction of the total profit. In order to develop this metrics, we will first need to compute the proportion of profit that we attribute to the ith variant of mth module in a configuration t. To do so, we need to first find a way to distribute the total profit from a configuration between the variants use in that configuration. We assume that the profit earned by a variant is in proportion of the variable cost of the variant. If this assumption is not suitable for certain specific cases, a different assumption can be made and incorporated appropriately in the ensuing analysis. Suppose βtim represents the fraction of profit attributed to the ith variant of mth module from configuration t. Then,

β tim =

N V

(5)

tim im M i = Im

V0 + ∑ ∑ N tim Vim m =1 i =1

We can now spread this ratio across the totality of the configurations to compute the proposed mteric: t =T

∑β PPim = 1 -

tim

Rt

t =1

Π

(6)

Notice that the point of subtracting the ratio term from 1 is to make this metric smaller the better as stipulated earlier on. The hybrid cost-profit proportion metric (HCPPim) A logical way to include the effect of cost as well as profit is to create a hybrid metric that captures the effect of both the CPim and PPim. We suggest the product of the two metrics proposed above:

498


HCPPim = CPim * PPim After substituting for appropriate terms from equation (3) – (6), this can be written as follows: T     I K +  tim  im ∑ α t D Ntim Vim    t =1    * 1 − HCPPim =  T M Im     K 0 + DV0 + ∑∑∑ I timCim   t =1 m =1 i =1   

  t=1  T α t D R t - C  ∑  t=1 t=T

∑β

tim

Rt

(7)

where Cim and C are computed from equations (1) and (2) respectively. The Configuration Sequencing Problem in Mass Customization Setting In this section, we formulate the problem for optimal sequencing of the configurations for which demand is already known. Our formulation of the configuration sequencing problem (CPS) is characterized by the following observations:

Production planning occurs at the beginning of each period of a fixed time interval, e.g., a week.

The amount of each configuration to be produced is determined by (1) demand placed by customers in this period or a prior period, (2) backlog of demand carried forward from previous period(s), (3) Production capacity, and (4) production budget available

The backlogged demand has priority over the current demand while sequencing production.

Due to the dynamic nature of demand in mass customization environment and short shipment lead-times, a renewal of planning activity occurs at the beginning of each period. This erases any effects of previous decisions that might be suboptimal in light of the new demands and other developments during the period. Hence, it is acceptable to develop the sequencing schedule that is optimal over a single period.

In compliance with the mass customization practices, the firm employs delayed differentiation to reduce lead-times and production costs of uncertainty. Therefore, the demand for each configuration is available with precision. An exact demand αt D for configuration t is expected to be produced in the current planning period, in addition to .

499


The goal of the configuration sequencing is to maximize value of production within a single period of planning where value is measured by the metrics defined before.

We can now define the configuration sequencing problem (CSP) as follows: Given the demand for each configuration t, we need to determine the production quantity Qt of configuration t such that we (1) comply with the production budget ($B) constraint, (2) do not exceed the available production time/capacity (Y), and (3) production quantity of any configuration does not exceed its respective demand, (4) If, for any configuration, the demanded quantity is not completely produced in the relevant period, a penalty is applied that reduces the value of the contribution. The exact correlation of the delay in delivery and the corresponding loss of value is not generally available. Our formulation reduces the value of the portion of demand that the program was unable to a fraction γ of the profit that would have accrued from that configuration. Our goal is to maximize a generalized measure of profit that we have defined as "value" in the previous section. Normalizing the metric Since the three measures of value developed above are each less than unity, we first normalize these measures on a scale of 10 to ensure adequate contributions from the value terms in the objective function. Let’s annotate the value of manufacturing the ith variant of the mth module as Zim. Zim = {CPim, or PPim, or HCPPim) The normalized value of Zim is then given by:

Z 'lm =

Z im * 10 Max{Z im , ∀i, ∀m)

The value of configuration t, Ωt, is then given by:

Ωt =

M

Im

∑∑ m = 0 i =1

' I tim Z im

(8)

500


Formulation of the configuration scheduling problem (CSP) As we formulate the CSP, it is important to define and understand the structure of the objective function. We discuss this in the following section: Objective function of CSP In its most exhaustive form, the objective function of the CSP would comprise three separate components:

A "reward" component that maximizes the value of the current production

A "penalty" component that levies a penalty for the demand that remains unfilled due to lack of capacity, scheduling, or resource (budget or time) issues

A "must-produce" quantity of the configurations that is backlogged from previous periods and the delivery lead-time considerations require that such quantities must be produced in the current period, capacity and resource constraints permitting.

Additional notations We know that αtD is the demand for configuration t that should be met in the current period. Let us define the additional notation as follows: Qt

=

γt

=

δt

=

B

=

R Ct(u)

= =

Rt(u)

=

Quantity of configuration t to be produced in the current period inclusive of backlogged demand, if any. Loss in value per unit of configuration t as it would not be produced in the current period. Backlogged demand of configuration t that must be produced in the current period to meet deadline obligations stemming from promised lead-times. Maximum allowed cost of production in the current period (= Budget). Total time for production in the current period. Cost of producing u units of configuration t in a single batch (to maximize economies of scale) Time needed to produce u units of the tth configuration in a single batch (to maximize economies of scale)

CPS is formulated as follows: T

Max

∑Q t =1

T

t

Ω t - γt

∑ (α D + ∂ t

t =1

t

− Qt ) Ω t

501

VOLUME 1: STRATEGIES AND CONCEPTS T

= Or, equivalently, Max

∑Q

t

(1 − γ t ) Ω t

t =1

s.t.

Qt ≤ αt D + δt Qt ≥ δt

for ∀ t

(9)

for ∀ t

(10)

M Im T   T  K 0 + V0 ∑ Q t + ∑∑ K im I tim +  ∑ Q t Vim N tim   ≤ B m =1 i =1  t =1  t =1 

(11)

M Im T   T  X 0 + Y0 ∑ Q t + ∑∑  X im I tim +  ∑ Q t Yim N tim   ≤ R m =1 i =1  t =1  t =1 

(12)

All variables ≥ 0. First we need to deal with equation (10) which is inconsistent with MDKP. This is not difficult. There is certain cost (B0, say) and time (R0, say) associated with production of δt units, given below: B0 = K 0 + V0 R0 = αt X 0 + Y0

M Im   T  + + δ K I ∑ ∑∑  im tim  ∑ δ t Vim N tim   t m =1 i =1  t =1  t =1  T

T

M

Im



∑ δ + ∑∑ X t

t =1

m =1 i =1



I

im tim

 T  +  ∑ δ t Yim N tim    t =1 

(11')

(12')

If B > B0 and R > R0, a feasible solution exists. We can now net out these resources from the RHS of (11) and (12) and delete inequality (10) and proceed to solve the resulting MDKP. Thus, After deletion of (10), B and R will be replaced by: T M Im    T    B B − V0 ∑ δ t + ∑∑  K im I tim +  ∑ δ t Vim N tim     t =1 m =1 i =1   t =1   

(11'')

M Im T    T    R R − Y0 ∑ δ t + ∑∑ X im I tim +  ∑ δ t Yim N tim     t =1 m =1 i =1   t =1   

(12'')

The second term in the objective function represents the penalty for configurations that were not completed in full quantity demanded. Inequality (9) requires that production of any configuration does not exceed its demand; inequality (10) requires that production of any configuration must equal or exceed the backlogged demand. Inequalities (11) and (12) are respectively the budgetary and total time

502


available constraints, which are self explanatory. The configurations whose demand is not completed in any given period that must be produced as soon as possible, preferably in the following period. It is a matter of managerial policy and lead-time trade offs. However, these considerations can be easily incorporated in the above formulation. Finally, it should be pointed out that a more traditional formulation where maximizing profit, rather than a surrogate measure such as value, is easily obtained by maximizing the following objective function: M Im  T    ∑ α t D R t  - α t K 0 + V0 Q t + ∑∑ m =1 i =1  t =1  

  T    K I +  im tim  ∑ Q t Vim N tim     t =1    

Prioritization at modular level The mass customization environment requires not only the product to be modular but also the manufacturing process to be cellular. Ideally, for each module m, there will be a flexible cell that produces the variants of a single module. We assume that there would be enough capacity to produce all the variants needed to produce Qt units of configuration t; t = 1, 2, …, N. Should there be need to sequence the variants of a given module, we can simply rank these in the order of their decreasing value Zim and produce in that order. The integrated problem where we simultaneously decide the order of variants and of configurations is an enormously complex problem. The sequencing proposed here serves the goal of prioritization and sequencing in a pragmatic way. The order generated here is consistent with the way the configuration sequence will be developed heuristically (next section), it may, however, not be globally optimal. Given the dynamic and devolving nature of demands and small lead-times, an efficient heuristics is much more appropriate and economic than a global optimal solution arrived at the expense of significant cost and time. The Optimal and Efficient Heuristics Solutions for CSP Due to the value associated with each configuration and dual resource constraints, the CSP is equivalent to a two-dimensional Knapsack problem or an equivalent 2-D Bin packing problem. Note that CSP is an NP-Hard problem in strong sense since with any one constraint alone (budgetary or time), the problem reduces to Bin Packing/Knapsack problem which is NP-Hard. With both constraints simultaneously present, the problem is harder. It can be solved optimally by using any Mixed IP solver using Branch and Bound or dynamic programming methods, provided the number of configurations is not too large for available computer

503


storage. However, since these solutions may require complete enumerations in the worst case, a heuristic solution is desirable. In order to develop a heuristic solution, we can draw from a wide variety of heuristic solutions available in Knapsack or Bin Packing literature. Other problems that have close kinship with the configuration scheduling are: the Partition problem, Multiprocessor scheduling problem, and the Subset sum problem. Each of these has several heuristics associated with them, and one can also draw upon that knowledge base. Development of a heuristic solution to CSP Before presenting the heuristic, we will first compute the cost and time of producing U units of configuration t. The cost Ct(U) of producing U units of configuration t is given by: M

Ct(U) = K 0 + UV0 +

Im

∑∑ (K

im

+ U N timVim ) I itm for U > 0.

(13)

m =1 i =1

The time Rt(U) of producing u units of configuration t is given by: M

Rt(U) = X 0 + UY0 +

Im

∑∑ ( X

im

+ U N timYim ) I itm for U > 0.

(14)

m =1 i =1

Let’s also compute the break-even quantity for configuration t, U Bt below which it is unprofitable to produce configuration t.

{U Bt = U : Rt – Ct(U) ≥ 0}, Simple algebra yields: M

B

Im

Rt − K 0 − ∑∑ K im m =1 i =1

Ut ≥

M

V0 +

Im

∑∑ I

tim

(15)

N tim Vim

m =1 i =1

We are now ready to suggest a heuristics solution to CSP. An important observation about this heuristic is that if the CSP as formulated above is infeasible because of the cost- or time- overrun, the heuristic still develops a feasible solution which should be close to optimal. Step 1:

Define parameter, the utility value of configuration t, Ut =

βt for configuration t, t = 1, 2, …T. Ct τt

504


Rank configurations 1 …, T in accordance with decreasing values of Ut. Let’s call this sequence of configurations {U[1], U[2], …, U[T]: U[i] > U[i+1], i = 1, 2, …, T-1}. Since there is no confusion after this ranking, let’s re-index the corresponding sequence of configurations as 1, 2, …, T. Step 2:

Set B = Available budget, R = Total available time,

Step 3:

Initialize: B Β R R, t = 1, Sequence S (Q1, Q2, …, QT).

Step 4:

If δt < U Bt , Qt = 0. Go to Step 10. Else go to Step 5.

Step 5:

If δt + αt D > U Bt Check if the remainder of the budget permits production of δt + αt D units: If B > Ct(δt + αt D) Then Go to Step 6. Otherwise Go to Step 8.

Step 6:

Check if the remainder of the time permits production of δt + αt D units: If R > Rt(δt + αt D), then go to Step 7. Otherwise go to step 9. Otherwise go to step 8.

Step 7:

Qt = δt + αt D B Β − Ct(δt + αt D) R R − Rt(δt + αt D) tt+1 Go to Step 4.

Step 8:

M Im    B − K 0 − ∑∑ I tim K im  m =1 i =1  Qt =  M Im   V0 + ∑∑ I tim N tim Vim  m =1 i =1  

Where Z indicates the nearest lower integer to Z. Go to step 10


Step 9:

505

M Im   − − R X I tim X im  ∑∑ 0  m =1 i =1  Qt =  M Im    Y0 + ∑∑ I tim N tim Yim  m =1 i =1  

Where Z indicates the nearest lower integer to Z. Go to step 10 Step 10:

If t < T, Qk = 0, k = t+1, … , T. S = (Qt, t = 1, 2, …., T) S' = (δt + αt D - Qt, t = 1, 2, …., T) STOP.

The quantities in set S are the actual production quantities for configurations 1 through T. The quantities in set S' are those that will be backlogged for future production. These will serve as δt’s for the next period, for instance. A further practical refinement of formulation is possible where the quantities δt could be further partitioned in two parts for each configuration t. The portion εt could be treated as must-produce (as has been demonstrated here) and the portion δt - εt could be inserted in the objective function with higher contribution value than Ωt. Further details are skipped. A second efficient heuristic This heuristic is developed along the lines of Ackay et al. (2007) and is specifically suited for mass customization setting associated with configuration scheduling. Ackay et al. report after exhaustive testing that their heuristic ranks first in terms of the speed of finding feasible solutions while maintaining a decent proximity with optimal solution. These attributes make the heuristic presented below well suited to mass customization environment. The heuristic is presented below. Step 1:

Initialize: Qt 0, all t = 1, …, T. Set E {t | Qt = 0, All t} Set S { St = 0, All t}

506


Unused budget and Unused Time: B0 B, R0 R Max

quantities

allowed

for

each

configuration

t:

u t

Q ← α t D + δ t for ∀ t ∈E. Step 2:

Compute Effective Capacity for each configuration t:

M Im M Im   R Y X im  − − − − B K K ∑∑ ∑∑ 0 im 0 0  0   m =1 i =1 m =1 i =1 , Z t = Min  , t I tim N tim Yim I tim N tim Vim     for ∀ t ∈E.

Set Zt = Closest integer less than Zt. If Zt = 0 for ∀ t ∈E then go to Step 6. Otherwise go to Step 3. Step 3:

Find the configuration that contributes maximum value at effective capacity: Compute: Wt = Z t (1 − γ t ) Ω t for ∀t ∈ E. The configuration that is most beneficial to produce is t*, where: t* = arg max {Wt } . t ∈E

Step 4:

Determine the quantity of configuration t* to produce within the bounds:

{

Z t * = min Z t * , Q ut*

}

To prevent multiple accounting of set up cost: S t * ← S t * + 1 Step 5:

Update the values. (i)

Q t* ← Q t* + Z t* Q

507


(ii)

If S t * ≤ 1:

  M Im  B0 B0 −  K 0 + Zt* V0 + Itim  ∑∑ K im + Zt* NtimVim    m=1 i =1     M Im  and, R0 R0 −  X 0 + Z t* Y0 + I tim  ∑∑ X im + Z t* N tim Vim    m =1 i =1   Elseif St* > 1:



B0 B0 − 

M

Im

∑∑ Z

 m =1 i =1 

and R0 R0 − 

M

t

*

 I tim N tim Vim  

Im

∑∑ Z

 m =1 i =1

(iii)

t*

 I tim N tim Vim  

Compute new upper bounds;

Q ut* ← Q ut* − Z t * (iv)

Set E {E – t*}

If E is a null set, go to Step 6. Otherwise, go to step 2. Step 6:

Stop. Q t * are the quantities that should be made of each configu-

ration t*.

Note that this heuristic is slightly more complex because of the set up costs that apply to both the modules as well as to the standard product. However, it can be seen that we have addressed that issue by introducing the variable S t * which is raised by a value of one every time this item is produced and the set up cost adjustment occurs only the first time. Clearly, this does not add much to the computational burden of the Ackay et al.’s heuristics.

508


Optimal solution to CSP An optimal solution can be developed for CSP using two well known methods: (1) Dynamic Programming, and (2) Branch and Bound. Readers are referred to Some purists of mass customization would argue that one needs to produce the entire demand in one period. This may not always be possible since that may need very large capacity and could be significantly uneconomical for high-volatility demands. In the above formulation, we have assumed that the promised lead-times are large enough so that some variants may spill over and be produced in the following period, allowing a cost-optimal solution to be developed. Performance of the heuristic solution Without providing an optimal solution, which would be time consuming and perhaps beyond the available computer capacity for most real-world problems, we would like to estimate the performance of this heuristic through the bounds that other heuristic solutions have obtained in the context of Bin packing and Knapsack problems. Notice that since the two problems are reducible to each other, the bounds described below apply to both of them. The bin packing problem has been analyzed intensively in the combinatorial programming literature. In a single-dimension bin packing problem, i.e., the problem with a single constraint, the best fit decreasing and first fit decreasing strategies are among the simplest heuristics solutions. They have been shown to use no more than 11/9 OPT + 1 bins (where OPT is the number of bins given by the optimal solution) (Yue 1991). The simpler of these, the First Fit Decreasing strategy, operates by first sorting the items to be inserted in decreasing order by volume (value in our case), and then inserting each item into the first bin in the list with sufficient remaining space. The sorting step is relatively expensive, but without it we only achieve the looser bound of 17/10 OPT + 2. A more efficient version of FFD uses no more than 71/60 OPT + 1 bins. (Dósa 2007, Vazirani 2006). It has been proved in leterature that the bound 11/9 OPT + 6/9 for FFD is tight (Garey and Johnson 1985, Johnson et al. 1974). Although these simple strategies are often good enough, efficient approximation algorithms have been demonstrated (Johnson et al. 1974) that can solve the bin packing problem within any fixed percentage of the optimal solution for sufficiently large inputs (this is called an asymptotic polynomial-time approximation scheme). This is an advantage the problem has over many other common NP-hard problems, some of which cannot be approximated within any constant factor at all. In more recent analyses of the Knapsack problem, 96% OPT heuristics are available for multiple dimension Knapsack problem.


509

Conclusions Mass customization as a strategic tool is gaining currency and momentum; yet it remains largely ill-addressed in the operations and supply chain management areas where rigotous optimization routines are needed to make informed decisions. There is a dire need to address operational issues such as production scheduling, inventory management, distribution, and routing issues that impact the supply chain of firms practicing mass customization. In this chapter, we take a step towards filling this void. Specifically, we developed a mixed integer linear program that yields optimal production policy, i.e., quantities and sequence of module-variants and configurations to be produced when the production capacity and budget are limited. In a mass customization environment, where the demand landscape is constantly evolving, it is necessary to develop solutions quickly and in close proximity to optimality. The exact solutions can be developed using such approaches as Branch and Bound and Dynamic Programming. However, given the NP-Hard characteristic of the problem studied in this chapter, these approaches could be very slow; requiring a complete enumeration of all possibilities in the worst-case scenario. Two heuristic solutions are, therefore, proposed that have been modified from the available literature related to multi-dimensional knapsack problem but tailored to suit the mass customization setting. In addition, the strategic nature of mass customization efforts requires that the objective of the managerial scenarios be extended beyond simple profit motive. Accordingly, three measures of the value of each product configuration are constructed. We hope this work will be further developed on a microscopic level where capacities are narrowed down to the machine/capability levels. In a larger research thrust, we hope that this work will encourage mass customization researchers to look into other areas of operational and supply chain optimization, such as distribution, lead time acceleration and minimization, and quality control techniques specifically tailored to mass customization environment.

References Akçay, Y., Haijun L. and Xu, S. H. (2007). Greedy Algorithm for the General Multidimensional Knapsack Problem, Annals of Operations Research. 150(1): 17–29. Alptekinoglu, A. (2004). Mass Customization vs. Mass Production: Variety and Price Competition. Manufacturing & Service Operations Management. 6(1): 98–103. Cabot, A. V. (1970). An Enumeration Algorithm for Knapsack Problems, Operations Research. 18(2): 306–311.

510


Chekuri, C. and Khanna, S. (2006). A Polynomial Time Approximation Scheme for the Multiple Knapsack Problem, SIAM Journal on Computing. 35(3): 713–728. Da Silveira, G., Borenstein, D. and Fogliatto, F. S. (2001). Mass customization: Literature review and research directions, International Journal of Production Economics. 72(1): 1–13. Dahl, G., Foldnes, N. (2006). LP Based Heuristics for The Multiple Knapsack Problem with Assignment Restrictions, Annals of Operations Research. 146(1): 91–104. Davis, S. M. (1987). Future Perfect. Adison-Wesley, Reading, MA. (Updated Edition: Perseus Group Books, (1997). de Vries, S. and R. V. Vohra (2003). Combinatorial Auctions: A Survey, INFORMS Journal on Computing. 15: 284–310. Dellaert, B. G. C.; Stremersch, S., (2005). Marketing Mass-Customized Products: Striking a Balance Between Utility and Complexity, Journal of Marketing Research. 42(2): 219–227. Dósa, G. (2007). The Tight Bound of First Fit Decreasing Bin-Packing Algorithm is FFD(I) ≤ (11/9)OPT(I) + 6/9, ESCAPE, to appear in Springer LNCS. Duray, R., Ward, P. T., Milligan, G. W., Berry. W. L. (2000). Approaches to Mass Customization: Configurations and Empirical Validation, Journal of Operations Management. 18: 605–625. Eppen, G. (1979). Effects of Centralization on Expected Costs in A Multi-Location Newsboy Problem, Management Science. 25(5): 498–501. Forrest, J. H., Kalagnanam, J., and Ladanyi, L. (2006). A Column-Generation Approach to the Multiple Knapsack Problem with Color Constraints, INFORMS Journal on Computing. 18(1): 129–134. Freville, A. (2004). The Multi-dimensional 0-1 Knapsack Problems: An Overview, European Journal of Operations Research, 155, p.1-21. Garey, M. R., and Johnson, D. S. (1985). A 71/60 Theorem for Bin Packing, Journal of Complexity. 1, 65–106. Garey, M. R. and Johnson, D. S. (1979). Computers and Intractability: A Guide to the Theory of NPCompleteness. W.H. Freeman. Garg A. and Tang C. S. (1997). On Postponement Strategies for Product Families with Multiple Points of Differentiation, IIE Transaction. 29(8): 641–650. Greenberg, H. and Hegerich, R. L. (1970). A Branch Search Algorithm for The Knapsack Problem, Management Science. 16(5): 327–332. Kellerer, H., Pferschy, U. and Pisinger, D. (2004). Knapsack Problems. Springer. Kleywegt, A. J. and Papastavrou, J. D. (2001). The Dynamic and Stochastic Knapsack Problem with Random Sized Items, Operations Research. 49(1): 26–41. Kochenberger, G. A., McCarl, B. A. and Wyman. F. P. (1974). A Heuristic for General Integer Programming. Decision Sciences. 5: 36–44. Kolesar, P. J. (1967). A Branch and Bound Algorithm for the Knapsack Problem, Management Science. 13(9): 723–735. Kotha, S. (1995). Mass Customization: Implementing the Emerging Paradigm for Competitive Advantage, Strategic Management Journal (Special Issue, Technological transformation and the new competitive landscape). 16: 21–42. Kotler, P. (1989). From Mass Marketing To Mass Customization, Planning Review. 17(5): 10–19. Kumar, A. (2004). Mass Customization: Metrics and Modularity, International Journal of Flexible Manufacturing Systems. 16(4): 287–311.


511

Kumar, A. (1995). Risk-Pooling Along a Fixed Delivery Route Using a Dynamic Inventory Allocation Policy, Management Science. 41(2): 344–360. Kumar, A., Gattoufi, S. and Reisman, A. (2007). Mass Customization Research: Trends, Directions, Diffusion Intensity, and Taxonomic Frameworks, Int. J. of Manufacturing Systems. 19(4). Kumar, A., Piller, F. T., Reisman, A. and Stecke, K. E. (2007). A Prioritization Algorithm for Configuration Scheduling in Mass Customization Environment, Proceedings of 2007 MCPC Conference, MIT Media Lab, Boston, MA. Lee, H. and Tang C. (1997). Modeling the Costs and Benefits of Delayed Product Differentiation, Management Science. 43(1): 40–53. Lin, E. Y. (1998). A Bibliographical Survey on Some Well-Known Non-Standard Knapsack Problems, INFOR. 36(4): 274–317. Lu, L. L., Chiu S. Y. and Cox, L. A. (1999). Optimal Project Selection: Stochastic Knapsack with Finite Time Horizon. Operations Research. 50: 645–650. Pendharkar, P. C. and Rodger, J. A. (2006). Information Technology Capital Budgeting Using a Knapsack Problem, International Transactions in Operational Research. 13(4): 333–351. Piller, F. T. (2004). Mass Customization: Reflections on the State of the Concept, International Journal of Flexible Manufacturing Systems Focused Issue. 16(4): 313–334. Piller, F. T. and Kumar, A. (2006). For Each, Their Own, Industrial Engineer. 38(9): 40–45. Pine, B. J. (1993). Making Mass Customization Happen: Strategies for the New Competitive Realities. Planning Review. 21(5): 23. Pine, B. J., Bart, V. and Boynton, A. C. (1993). Making Mass Customization Work, Harvard Business Review. 71(5): 108. Pine, B. J. (1992) Mass Customization: The New Frontier in Business Competition. Harvard Business School Press. Pirkul, H. (1987). A Heuristic Solution Procedure for the Multiconstraint Zero-One Knapsack Problem. Naval Research Logistics. 34: 161–172. Pirkul, H. and Narasimhan, S. (1986). Efficient Algorithms for the Multiconstraint General Knapsack Problem. IIE Transactions. 195–203. Salvador, F., Rungtusanatham, M. J., Forza, C. (2004). Supply-chain Configurations for Mass Customization. Production Planning & Control. 15(4): 381–397. Schwarz, L.B. (1989). Model for Assessing the Value of Warehouse Risk-pooling Over Outside Supplier Leadtime, Management Science. 35(8): 828–842. Senju, S. and Toyoda, Y. (1968). An Approach to Linear Programming with 0-1 Variables, Management Science. 15(4): B196–B207. Shapiro, J. F. (1968). Dynamic Programming Algorithms for the Integer Programming Problem-I: The Integer Programming Problem Viewed as a Knapsack Type Problem, Operations Research. 16(1): 103–21. Tayur S., Ganeshan R. and Magazine, M. (1999). Quantitative Models for Supply Chain Management, Kluwer International Series, Kluwer Academic Publishers, MA. Toyoda, Y. (1975). A Simplified Algorithm for Obtaining Approximate Solutions to Zero-One Programming Problems. Management Science. 21(12): 1417–1427.

512


Van Hoek R. I., Peelen E. and Commandeur H. R. (1999). Achieving Mass Customization through Postponement: A Study of International Changes, Journal of Market-Focused Management. 3(3-4): 353–368. Vazirani, V. V. (2006). Approximation Algorithm. Springer. Weingartner, H. M. and Ness, D. N. (1967). Methods for the Solution Of The Multi-Dimensional 0/1 Knapsack Problem, Operations Research. 15(1): 83–103. Yue, H. (1991). A simple proof of the inequality FFD(L) ≤ (11/9)OPT(L) + 1, for all L, for the FFD binpacking algorithm, Acta Mathematicae Applicatae Sinica. 7: 321–331. Yue, M. and Zhang L. (1995). A simple proof of the inequality MFFD(L) ≤ 71/60 OPT(L) + 1, for all L for the MFFD bin-packing algorithm, Acta Mathematicae Applicatae Sinica. 11: 318–330. Zipkin, P. (2001). The Limits of Mass Customization, Sloan Management Review. 42(3): 81–87.

Author Biographies Dr. Ashok Kumar is currently a professor of management at Grand Valley State University. He received his Ph. D. degree in Operations Management from Purdue University and taught a variety of operations related courses at Purdue, Ball State, Case Western Reserve, and Grand Valley State University. His research encompasses Mass Customization, Six Sigma, Quality Assurance, Supply Chain Systems, Advanced Manufacturing Systems, Operations Strategy, and Meta Research in Operations Management. His publications have appeared in Management Science, Operations Research, European Journal of Operations Research, IEEE Transactions in Technology and Management, International Journal of Flexible Manufacturing Systems, and other highly regarded journals. He has won numerous awards for distinguished research, teaching, professional service, and industrial expertise; including the Best Dissertation Award (Runner up) from Production and Operations Management Society, Outstanding Faculty Award from Seidman College of Business, Distinguished Contributions to a Discipline award from Grand Valley State University, Distinguished Service to a Discipline Award from Association of Indian Management Scholars, and numerous Best or Highest Quality Paper awards and Who’s Who in America’s Teachers awards. He has also received Outstanding Contributions to Quality Discipline award from National Institute of Quality Assurance, India. Contact: www.ashok-kumar.org Dr. Frank Piller leads the Technology & Innovation Management Group at RWTH Aachen University. He also is a co-founder of the MIT Smart Customization Group at the Massachusetts Institute of Technology, USA. Before entering his recent position in Aachen in spring 2007, he worked at the MIT Sloan School of Management and has been an associate professor of management at TUM Business School, Technische Universitaet Muenchen (1999-2004). His research focuses on mass customization, open/user innovation, and methods to increase the efficiency and effectiveness of the innovation process. As a founding partner of Think Consult, a management consultancy, he helps his clients to serve their customers better by using truly customer-centric strategies. Contact: tim.rwth-aachen.de | [email protected].

4.5

Procurement Mechanisms for Customized Products Songlin Chen Nanyang Technological University, Singapore Mitchell Tseng Hong Kong University of Science and Technology, Hong Kong, China

Customization is essentially a "pull" system and customers' demand for customized products is the ultimate force that drives a customization business. This chapter views customization from the demand side and looks into customers' procurement decisions for customized products. A conceptual framework based on contract theory and axiomatic design theory is constructed to characterize the essential decisions involved in procuring customized products. Based on the framework, this chapter identifies the key barriers that prevent customers from effectively tapping into the value of customization and explores alternative procurement mechanisms to overcome these barriers.

Introduction Customization has been recognized as a frontier for manufacturers to gain competitive advantage in an increasingly diversified and dynamic marketplace. Recent years have witnessed rapid increase in output of customized products, spanning from capital goods like airplanes and machine tools to consumer goods like cars, computers, printers, etc (Selladurai 2004; Moser and Piller 2006). With the proliferation of customized products and spread of customization technology, there emerges a new competitive landscape where multiple manufacturers compete on customization for customers' patronage. For example, both Boeing and Airbus customize airplane interiors for airliners; Dell, HP, and Lenovo allow customers to configure their own computers; both Adidas and Nike offer custom made sneakers, etc. The increasing availability of customized products gives customers more choices that could potentially best fulfill their individual specific needs. In the meanwhile, the escalating competition on customization among manufacturers shifts bargaining power towards customers' favor. However, the reality is that procurement of customized products is often a lengthy and costly process. In the context of industrial procurement for custom products like machine tools and equipment, purchasing personnel need to go through 513

514


painstaking preparation, evaluation, and back-and-forth negotiations (Cavinato et al. 2000). The administrative cost of procuring customized products is often significantly higher than that of standard products. Not surprisingly, customized products are often avoided by purchasing professionals whenever possible. In the context of consumer goods, customers often get "confused" by the large number of options offered in customization (Huffman and Kahn 1998). The difficulties of procuring customized products translate into high cost and burden to customers, which could offset the additional value of customization over a standard solution. Furthermore, as customers' procurement decisions precede the actual production of a customized product, difficulties in procurement discourage customers from customizing. This study provides a conceptual framework to understand customers' procurement decisions, to identify the key barriers that prevent customers from effectively tapping into the value of customization, and to explore alternative procurement mechanisms to overcome these barriers. The context of discussion is mainly based on procurement of custom made industrial goods, which are more complex hence more representative of the intricate challenges involved in procuring customized products. The rest of the chapter is organized as the following. Section 2 reviews relevant research in economics and engineering literature. Section 3 presents a conceptual framework to characterize the essential decisions, information and incentive structure in procuring customized products. Section 4 identifies the key barriers faced with customers in procuring customized products, and Section 5 discusses alternative procurement mechanisms and compares their relative performance in overcoming these barriers. This chapter concludes with a summary and discussion of future research in Section 6. Relevant Literature A defining feature of procuring customized products (relative to standard products) is that product specifications become decision variables, which are coupled with other decisions like price, delivery schedules, service terms etc. Determining product specification is essentially an engineering design problem, while procurement in general is an economic contracting problem. In the broad sense, procurement encompasses activities from need identification, writing specifications, supplier selection, product reception and return, etc (see Cavinato et al. 2000). This study uses procurement in a narrow sense by focusing on the stage of supplier selection and contracting, which contain the key decisions that determine the overall performance of procurement.


515

Procurement: Customers as principals From an economic perspective, procurement is basically transaction viewed from buyers' perspective. Through procurement, a buyer selects a supplier (or a number of suppliers) and comes to agreement with the supplier(s) upon terms of exchange concerning the product or service in terms of price, quality, delivery schedule, etc. Procurement in general can be approached as a contracting problem and the relationship between the buyer and suppliers in procurement can be characterized as principals versus agents (Laffont and Tirole 1993). There are two critical issues confronting the buyer (or principal) in procurement. One is "adverse selection", which describes a contracting situation where there is hidden information (when selecting a supplier, the buyer does not have full access to suppliers' local information in terms of cost, quality etc.). The other is "moral hazard", which describes a contracting situation where there is hidden action (after the contract is signed, the buyer is not able to accurately monitor the actual fulfillment of the contract) (Bolton and Dewatripont 2005). Adverse selection and moral hazard render possibility of strategic behavior by suppliers at the expense of the buyer’s interest. The use of economic theory (game theory in particular) to address market transaction problems with asymmetric information and incentives generally falls into the field of mechanism design (Royal Swedish Academy of Sciences 2007). Hurwicz (1960) defines a mechanism as "a communication system in which participants send messages to each other and/or to a "message center", and where a pre-specified rule assigns an outcome (such as an allocation of goods and services) for every collection of received messages". In the context of procurement, reverse auction is one of the most studied and commonly used mechanisms. When cost is the only dimension of suppliers' private information, a reverse auction can overcome "adverse selection" by engaging suppliers in competitive bidding on price. Reverse auctions have been implemented within advanced information systems and have henceforth transformed the procurement function of large corporations like GE and Motorola, who have reported billions of dollars in cost savings (Sawhney 2003; Metty et al. 2005). However, the majority of success has been concentrated on standard or "commodity" type products, the specifications of which have been predetermined with little ambiguity. In other words, the quality of these types of product is verifiable and the terms of contract can be made contingent on the quality of the product (Che 1993). When the product is customized, it becomes very hard to specify and verify contractually about the features or functionalities of the product. A particular problem concerning product customization is that the customer is unable to

516


accurately articulate needs in terms of concrete and clear requirements (Zipkin 2001). Asymmetry of information is not only about suppliers' cost but also on customization capabilities. As a result, both "adverse selection" and "moral hazard" become severe in procuring customized products, and a reverse auction based solely on price becomes ineffective or counter-productive (Kwak 2002). There is a growing stream of research devoted to address information asymmetry over multiple dimensions in procurement mechanism design, which will be discussed in details below. Product customization: Customers as co-designers In contrast to the principal-agent relationship assumed by economists, researchers in engineering and operations management community take a dramatically different view upon the relationship between customers and manufacturers in customization. Customized products, by definition, are created according to customer-specific requirements, and customers are integrated in the process of product creation by providing key design inputs. Customers' involvement in product design has been recognized as a critical identifier of customization (Duray 2002). Berger and Piller (2003) recognize the importance of customer interaction and propose to treat customers as collaborative designers (co-designers) in customization. Piller et al. (2004) further captures the concept of customer interaction in customization as "economies of customer integration", arguing that the economic value of customization lies in the process of product co-creation. Hippel (2005) takes product customization as a type of innovation and argues that successful customization requires integration of two sources of information: need information and solution information, which, however, are usually distributed asymmetrically between customers (or users) and manufacturers. Users have better need information because they have better understanding of local environment and intended use of the product; manufacturers have better solution information because of their expertise in product design and production etc. However, both need information and solution information could be "sticky" in the sense that they are costly to acquire, transfer, and use in a new location. Hippel advocates the use of design toolkits to transfer solution information to users so as to enable customization by users. With similar functionalities to design toolkits, product configurators have been widely adopted as a tool to integrate customers into product customization, most famously by Dell on personal computers. Product configurators are essentially software systems that apply artificial intelligence (AI) techniques to codify existing product knowledge in the form of models, cases, rules, and/or constraints


517

etc (Sabin and Weigel 1998). With product configurators, customization can be simplified to a sequence of selection of predefined attribute options, which can be performed by customers or sales personnel with relative ease and high level of accuracy (Forza and Salvador 2002). Despite the proven success of design toolkits and product configurators, these systems are more sales tools rather than procurement tools. This is evidenced by their lack of support, sometimes intentional prevention, for customers to compare competitive offerings from different manufacturers. In general, the economics and engineering literature views procurement of customized products from different perspectives and advocates different approaches to engage customers in customization. The economics literature is primarily concerned about information asymmetry, supplier incentives, and the associated contract implications. Economic mechanism design is applied as the main methodologies to address "adverse selection" and "moral hazard". The engineering literature, on the other hand, is mainly concerned about asymmetry and stickiness of (customer) need information and (manufacturer) solution information, and collaborative engineering design is assumed as the main methodology. By synthesizing these two streams of literature, the following section presents a unified framework that captures the essential decisions in procuring customized products. A Decision Framework for Procuring Customized Products On a conceptual level, procurement of customized products can be taken as a special case of procurement where the product specification needs to be collaboratively defined by customers and manufacturers. It can be conceptualized as a contracting problem with an embedded co-design problem. Contracting and codesign correspond respectively to the economic and engineering aspect of procuring customized products. These two aspects imply different, but coupled, sets of decisions, which are situated in different information and incentive structure. Decisions The decisions involved in procuring customized product can be generally summarized according to the following questions: 1) which supplier, 2) what product (specification), 3) when to deliver (lead time), 4) how many (quantity), 5) how much (price), and 6) what if (warranty), etc. The decisions on supplier, quantity, price, and warranty generally fall into the economic domain concerning contracting, while specification and lead time are mainly within engineering

518


domain concerning co-design. Without loss of generality, price ( p) and specification ( s) are selected as the representative decisions in these two domains and an agreed tuple of ( s, p) henceforth represents a procurement contract. Although product specification has been generally defined as technical descriptions of a product, it actually has dramatically different interpretations by customers and manufacturers. To a customer, product specifications describe, with legal authority, what product she is entitled to receive in terms of product features and functionalities, which in turn will determine to a large extent of the value ( v ) of customization, i.e. the customer’s maximum willingness to pay. To a manufacturer, product specification represents the legal commitment to deliver a product as specified. Product specification guides, as well as binds, a manufacturer’s operations including design, production, delivery, etc., thus has a large impact on the cost ( c ) to deliver a customized product. A critical issue is that these two interpretations are often not congruent. As quoted in a judge’s comments on a contract dispute, "an unsophisticated owner reading a performance specification thinks of a Mercedes-Benz, while the contractor sees a Volkswagen" (Hartman 1997). The actual meaning of price is also worth some clarification. Although price has been generally assumed as the monetary amount that a buyer pays a supplier in exchange for a product or service, it has dramatically different implications depending on whether the price is paid upfront as a lump sum, or after receiving the final product, or in multiple installments. Besides the issue of timing, the amount of price can also be either fixed or contingent on some other factors. For example, both fixed-price contracts (the buyer offers the seller a pre-specified price for delivering the final product) and cost-plus contracts (the buyer reimburses the manufacturer for costs plus a stipulated fee) are commonly used in procuring construction services (Bajari et al. 2001). Between fixed-price and costplus, there is a range of contracting arrangement under the name of risk-sharing contracts. The timing and amount of payment can be used as a strategic tool for the customer to prevent opportunistic behavior ("moral hazard" in particular) from suppliers (Interested readers can refer to Laffont and Tirole (1993) for more comprehensive and in-depth discussion concerning contract design and the associated pricing issues in procurement.). The focus of this study is focused on exploring mechanisms to overcome decision barriers in procuring customized products. To simplify discussion without loss of generality, upfront, fixed-price contracts are assumed.


519

Information Defining specification for a customized product requires input of (customer) need information and (manufacturer) solution information. This study adopts the domain concept of axiomatic design (Suh 1990) to represent the information structure in customization, which can be taken as a special form of design. According to Suh, design in general can be viewed as a series of what-to-how mappings from customer needs {CN } to functional requirements {FR} , to design parameters {DP} , and finally to process variables {PV } . {CN } represents a customer’s real, but often hidden, needs; {FR} is the articulated customer needs in terms of product functionality and features; {DP} represents a technical solution that satisfies {FR} ; and {PV } describes how the design solution can be produced. Collectively, {FR, DP, PV } represents a complete set of product specification.

s = { FR, DP, PV }

(1)

In the context of product customization, it can be generally assumed that customer’s need information is reflected in {FR} while a manufacturer’s solution information is reflected in {DP} and {PV } . The mapping relationships between

{FR} and {DP} , {DP} and {PV } are characterized by design matrix [ A] and [ B ] , respectively. It is worth noting that design matrix may or may not be in numeric form but generally indicate the inter-relationships between different functional domains in design (Suh 1990).

{FR} = [ A]{DP}

(2)

{DP} = [ B ]{PV }

(3)

The price of a customized product is subject to market forces in terms of demand and supply. More specifically, the value of a customized product to the customer ( v ) and the cost for a manufacturer to deliver the product ( c ) establish the upper and lower bound of a price window between the customer and the manufacturer. v and c can be generally assumed as functions of {FR} and {DP, PV } , respectively.

v = V ({ FR} )

(4)

c = C ({ DP, PV } )

(5)

520


(6)

c≤ p≤v

Figure 1 depicts the interrelationship among different decision variables in procuring customized products. With multiple manufacturers, the customer needs to interact with each manufacturer to define product specification. The final price depends on factors including the customer’s preference tradeoff between functionality and price, her bargaining power, the level of competition among manufacturers, as well as the mechanisms that the customer selects supplier (e.g. negotiation or auction, which will be discussed in more detail below). Customer Economics (Contracting)

Value

Manufacturer Price

Cost

Specification Customer Needs

Engineering (Co-Design)

Functional Requirements

Design Parameters

Process Variables

Solution Information

Figure 1: A decision framework of procuring customized products.

Incentives Depending on the specific identity and context, customers could have different agendas in procurement. For example, government procurement agencies care more about social welfare while purchasing departments of industrial firms are more concerned with cost reduction. Without loss of generality, this study assumes a generic utility function (u ) and profit function (π ) to represent respectively customers' and manufacturers' objectives in customization. Both functions assume a quasi-linear structure (Che 1993).

u = V ({ FR} ) − p − Tc

(7)

π = p − C ({DP, PV } ) − Tm

(8)


521

Tc and Tm represent transaction cost incurred during procurement to the customer and manufacturer, respectively. To maximize utility and profit, respectively, customers and manufacturers are not fully aligned in incentives. As price ( p) is concerned, customers and manufacturers, as in any buyer-seller relationship, are mutually opposed in incentives. The competition between customers and manufacturers as well as the competition among manufacturers entice strategic behavior by withholding or misrepresenting private information in terms of cost, quality, etc. As product specification ( s) is concerned, truthful exchange of need information and solution information and collaborative decision making are needed to generate high-quality customized solutions. Given the dependencies between p and s , the relationship between customers and manufacturers in product customization can be best characterized as coopetition, which describes a situation where incentives for collaboration and competition coexist (Brandenburger and Nalebuff 1996). Barriers in Procuring Customized Products The decision framework presented above provides a compact framework to look at the essential decisions and their associated information/incentive structure involved in procuring customized products. By tracing the decision variables captured in the decision framework, this section discusses the critical barriers that are hampering customers from effectively tapping into the potential value of customization. Articulating requirements The advantage of customization over standardization lies in its ability to fulfill individual customers' specific needs. However, customers often have difficulty to accurately articulate needs in terms of concrete and clear requirements (i.e. { FR} ), particularly when the product is complex and the customer does not have sufficient technical knowledge (Zipkin 2001). Since { FR} is the design input in customization, inaccuracy in { FR} will distort the actual need information and mislead manufacturers in design problem solving. The resulted solution may not be what the customer expected. Customers and manufacturers need to resolve the discrepancy, which, however, is often a laborious and costly process that drives up transaction costs ( Tc , Tm ).

522


Although inaccuracy in customer requirements can be generally reduced through iterative interactions between customers and manufacturers, it can seldom be eliminated. As a result, contract for a customized product is inherently incomplete. Inaccuracy in articulation of requirements exposes customers to the risk of committing to inferior solutions. Customers may need to either initiate a design change and renegotiate the contract, or stick to an inferior solution and settle with compromise. However, any customer initiated changes could be subject to upcharges, particularly considering the specificity and non-substitutability of customized products (Bloomer et al. 2004). Evaluating solutions A second challenge in procuring customized products is the difficulty for customers to accurately evaluate solutions (i.e. { DP, PV } ). In many situations, customers are often not technically savvy and are unclear of how well their requirements are actually fulfilled. Customized solutions are often presented to customers in the form of technical configurations, prototypes, engineering drawings, and sometimes design sketches. These media of representation create "virtual images" of the final product with varying degrees of accuracy. Given the near impossibility to fully describe a product, there is inherent inaccuracy for customers to evaluate a customized solution. Consequently, the customer will be exposed to the risk of "moral hazards" by manufacturers. Although the accuracy of evaluation can be improved by requesting high-fidelity samples or prototypes, it is often achieved at the expense of higher transaction costs ( Tc , Tm ). Obtaining competitive price Customized products, by definition, are designed or produced specifically for a particular customer. Depending on the level of customization, a customized product is identified with the customer with a certain level of exclusivity. A product’s identification with a customer implies a potential best fit with the customer’s specific needs, but it also means there’s no direct substitution with other offerings in the market. In economic terms, the specificity of a customized product subjects the customer to a niche monopoly by the customization provider. Furthermore, a major, if not the only, motivation for manufacturers to pursue customization is to differentiate from competition. As a result, manufacturers engaged in customization often have distinct capabilities and offer heterogeneous solutions, which resist direct comparisons. In general, the market price of a customized product is often obscure due to lack of efficient competition.


523

Reducing transaction cost The process of procurement itself could be costly. Activities like identifying requirements, approaching manufacturers, and evaluating solutions all consume time and resources. When significant, transaction cost could wipe out the additional value of customization over a standard solution. Tc reduces customer utility (Eq. 7). Its presence limits the scope of search that the customer is able to afford in customization. Similarly, Tm reduces profit (Eq. 8) for a winning manufacturer, and represents a net loss for the losing ones. Anticipating potential losses, manufacturers need to gauge their chance of winning and the potential gains when deciding whether to participate in the competition or not. In other words, Tm creates an entry barrier for manufacturers. Within a given technological framework, transaction costs cannot be easily reduced because they are necessary byproducts of improving accuracy in articulating customer requirements and evaluating manufacturers' solutions. Procurement Mechanisms for Customized Products There is a large variety of procurement mechanisms being practiced in industry and studied in literature. This section categorizes these mechanisms into three general types based on the trading institution used, namely search-based (or fixed price), negotiation-based, and auction-based (Milgrom 2004). The main properties of each type of mechanism are presented and the performance of these mechanisms is discussed in terms of their effectiveness in addressing the barriers in procuring customized products. Search-based procurement mechanisms In search-based procurement, a customer searches for a product that best satisfies her needs from a predetermined and fixed solution space. For each product alternative, the customer is faced with a take-it-or-leave-it decision without haggling on price, product attributes, or whatsoever. This is the mechanism we employ when buying from shopping malls, online product catalogs, etc. In the context of procuring customized products, search-based mechanisms are often embedded in a configuration process, in which customers locate a product alternative by selecting a sequence of predefined product attributes, for example, buying a Dell notebook. The actual price of a configured product is derived based on base prices, price markups, discount rules, etc., which have been predetermined (Bichler and Kalagnanam 2005).

524


Search-based procurement mechanisms can be characterized with fixed solution space ( { DP, PV } ) and price ( p) . A main advantage of search-based mechanisms is that customers can search for (or configure) the product by themselves, which implies low transaction costs to manufacturers ( Tm ). In situations where sales or engineering support is needed, the level of technical skills required is generally low and can be achieved with basic training. When customers are able to articulate needs with high level of accuracy, the process of procurement becomes straightforward, which implies Tc is also low. However, if customers are unable to articulate their needs and lack of product knowledge, the process of search could be lengthy and frustrating. The main disadvantage of search-based mechanisms lies in its rigidity. A prerequisite condition is that manufacturers are able to capture and organize their customization knowledge in a structured way. However, it’s often difficult to describe a product, especially a complex one, in sufficient details without confusing customers. The rigidity of search-based procurement also prevents effective communication and, as a result, is unable to effectively handle ambiguities in { FR} . As price is fixed by manufacturers beforehand, it is difficult for customers to leverage the competition among manufacturers and obtain competitive price ( p ) . Negotiation-based procurement mechanisms Negotiation is arguably the most widely used institution in industrial procurement. Figure 2 illustrates a typical process of negotiation-based mechanism for procuring customized products. The process starts with the customer identifying a manufacturer that she expects to be the most competent (or appropriate) for the customization task and then engages in a bi-lateral dialogue with the manufacturer upon product attributes, price, or anything that is pertinent. Both parties make offers/counteroffers and collectively search for a mutually acceptable solution. If an agreement is reached, the process concludes with a procurement contract; otherwise, the customer contacts another manufacturer and repeats the process. This process is defined as "sequential search" by Wolinsky (2005). Negotiation-based mechanisms are characterized with high flexibility and richness of communication, which allow customers to specify { FR} with ambiguities and manufacturers to propose approximate solutions {DP, PV } . Through an iterative process, a joint solution ( {FR, DP, PV } ) could be gradually improved to best

525


match customers' specific needs with manufacturers' distinct capabilities. However, negotiation is inherently inefficient. Myerson and Satterthwaite (1983) have proved "…the general impossibility of ex post efficiency of bargaining without outside subsidies". In other words, there is always a possibility that negotiators may fail to reach an agreement because of lack of incentives for truth telling, even though mutually beneficial solutions are available. The outcome of a negotiation is indeterminate and uncertain. As negotiation is a bi-lateral mechanism, it is also difficult for customers to explicitly leverage the competition among manufacturers to achieve competitive price ( p ). Furthermore, since the process of negotiation is usually iterative, negotiation cost ( Tc , Tm ) are usually high.

Customer

Manufacturer

Supplier identification Requirements

Negotiation

Solution alternatives

Solution Price No

Agree? Yes

Contract

Figure 2: Procurement mechanisms based on bilateral negotiation.

Auction-based procurement mechanisms An auction is a market institution with an explicit set of rules determining resource allocation and prices on the basis of bids from the market participants (McAfee and McMillan 1987). Auctions used for procurement are also called reverse auctions, in which bidders bid to supply a product or service. Among the vast literature on auction design, the number of biddable attributes is most relevant to procurement of customized products. In single-attribute (usually price) auctions, suppliers bid solely on price; while multi-attribute auctions take into account of additional factors like quantity, quality, lead time, etc. As attributes

526


upon multiple dimensions are often converted into a single score to reflect the overall value of a bid, multi-attribute auctions are also called score auctions. Price-only auctions Price-only auctions are often implemented with a so-called Request for Quote (RFQ) process, which is illustrated in Figure 3. The process starts with the customer sending out requests to potential suppliers with documents specifying information that include product specifications, purchase quantity, a tentative delivery schedule, and a target price etc. Interested manufacturers then respond with usually, but not limited to, a product design solution (in terms of drawings, technical specifications, etc.), delivery arrangement, and a price quote. The customer will evaluate the quotes and invite qualified manufacturers to engage in a bidding process in which manufacturers compete solely on price, and lowest price bidder wins the contract.

Customer Request

Manufacturer Requirements

Quote Qualify

Solution Price

Bid Price

Lowest price

Contract

Figure 3: Procurement mechanisms based on price-only auction.

The main advantage of price-only auction lies in its efficiency in competitive price discovery ( p ) . Price-only auctions are most successfully used for procuring standard or commodity-type products. For customized products, price-only auctions require customers to be able to articulate requirements and evaluate solutions with high level of accuracy. Ambiguity in { FR} and inaccuracy in

527


evaluating { DP, PV } , when coupled with throat-cutting competition on price and arms-length relationship, exacerbates the effect of "adverse selection" and "moral hazard". Another major drawback of price-only auctions is lack of incentives for manufacturers to participate, particularly those with differentiated and superior solutions who may perceive the competition as unfair and be concerned of being commoditized (Bloomer et al. 2004). Score auctions In a typical score auction, the customer specifies the procurement request not only with functional requirements but also a score rule of bid evaluation, which indicates the customer’s preferences over different attributes (e.g. price and quality) by means of weighting factors (Figure 4). The multiplicity of bidding attributes in score auctions exempts customers from dictating a uniform set of requirements for different manufacturers hence can mitigate the challenge of articulating needs and can also give manufacturers better incentives. The resulted contract price ( p) depends on the weighting of price in the score function and the level of competition among manufacturers. In general, score auctions are more efficient in terms of contract allocation, i.e. identifying the most competent manufacturer. Customer Request

Manufacturer Requirements Scoring rule Bid

Score

Solution Price

Contract

Highest score

Figure 4: Procurement mechanisms based on score auction.

528


A critical challenge faced with score auctions is the difficulty to accurately determine the score function itself, which requires customers to make preferential tradeoffs among different attributes and to come up with clearly defined criteria for evaluating solutions. Meeting such stringent requirements is difficult, if not impossible, often at the cost of extensive bid preparation and evaluation, which drive up transaction costs ( Tc , Tm ). Mechanism comparison Table 1 summarizes the properties of the three general types of mechanisms and their relative performance in procuring customized products. Generally speaking, search-based procurement mechanism incurs the least transaction cost but requires predefinition of the solution space as well as pricing scheme. The rigidity inherent in the mechanism makes it difficult to handle customers' truly individualistic requirements. Negotiation-based procurement mechanisms are highly flexible with rich communication and have large tolerance of ambiguities. Such property exempts customers the difficult task of accurately specifying and committing to a fixed set of requirements or evaluation criteria, and exempts manufacturers from committing to a predetermined solution space or pricing scheme. With extensive information exchange, innovative win-win solutions can be constructed during a joint decision process. This can potentially best match customers' individual specific needs with manufacturers' differentiated capabilities. Negotiation is generally a good mechanism for co-design, but it suffers from inherent inefficiencies due to lack of incentives for truth telling in contracting. Reverse auctions are generally competitive mechanisms with high efficiency in contract allocation. Price-only auctions are effective in achieving competitive price but require customers to be able to articulate and commit to a uniform set of requirements. It discourages manufacturers with differentiated solutions to participate. As a result, price-only auctions are preferably used for procuring customized products that customers have sufficient technical knowledge and can articulate requirements with high degree of accuracy. Score-based auctions promise higher utility and provide suppliers better incentives. However, defining the score function is often a very difficult task. It is worth noting that these general types of mechanisms are often used in combination in practice. Holding a reverse auction, whether price-only or scorebased, often involves multiple rounds of negotiations beforehand concerning the rules of contract awarding. The implementation of a contract awarded by auction may also involve multiple rounds of negotiations concerning the actual fulfillment of the contract.

Table 1: Procurement mechanisms for customized products.


529

530


Summary and Future Research Research in mass customization has been growing tremendously in the last two decades, and we have seen rapid accumulation of both technological and managerial knowhow in customization. However, most research related to product customization has been focused on the supply side, i.e. improving manufacturers' customization efficiency. Less attention has been paid to customers' procurement decisions. Although there has been growing awareness of the importance of integrating customers in to product co-creation, research efforts have been primarily targeted at addressing the technical challenges in terms of overcoming information asymmetry and stickiness (von Hippel 2005, Piller 2004). However, the economic incentives that underlying customers' procurement decisions, particularly in face of multiple competing manufacturers, are largely overlooked. By synthesizing relevant literature in both economic and engineering, this study conceptualizes procurement of customized products as a contracting problem with an embedded co-design problem. A framework is constructed to characterize and represent the essential decisions as well as the associated information and incentive structure in procuring customized products. Based on the decision framework, the critical barriers of procuring customized products are discussed systematically by tracing the decision variables. This study also explores alternative procurement mechanisms (namely search-based, negotiation-based, and auction-based) and compares their relative performance in overcoming these barriers. In general, procurement of customized products is an important problem that deserves more research attention than it currently enjoys and it is a far more complicated problem than it seems. It involves asymmetric information, conflicting incentives, and joint decision making across engineering and economic domains. This study mainly serves the purpose of characterizing the problem and providing a conceptual framework. The argument presented in this study is qualitative in nature. Future research is needed to substantiate the conceptual framework with specific procurement scenarios and practical problem solving methodologies. Given the interdisciplinary nature of the problem, a more economics-oriented area of future research is to develop hybrid mechanisms for procuring customized products, taking into consideration of the need for supporting collaborative design; while a more engineering-oriented area of future research is to develop procurement systems and design decision support systems, taking into consideration of the underlying economic incentives. These two streams of research could potentially converge to a coherent set of mechanisms


531

and systems that can greatly enhance the efficiency of procuring customized products, and thus advancing mass customization from the demand side. Acknowledgments This research is supported by National Natural Science Foundation of China (NSFC: No.70418013) and Hong Kong Research Grants Council (RGC: N_HKUST625/04).

References Bajari, P., S. Tadelis, et al. (2001). Incentives versus transaction costs: A theory of procurement contracts. The Rand Journal of Economics. 32(3): 387–407. Berger, C. and F. Piller (2003). Customers as Co-Designers. IEE manufacturing engineer. 82(4): 42–45. Bichler, M. and J. Kalagnanam (2005). Configurable offers and winner determination in multi-attribute auctions. European Journal of Operations Research. 160: 380–394. Bloomer, J., J. Zale, et al. (2004). Battling Powerful Procurement Groups: How to Profitably Participate in Reverse Auctions. SPG Insights. Bolton, P. and M. Dewatripont (2005). Contract Theory. Cambridge, Mass., MIT Press. Brandenburger, A. and B. Nalebuff (1996). Co-opetition. New York, Doubleday. Cavinato, J. L., R. G. Kauffman, et al. (2000). The purchasing handbook: a guide for the purchasing and supply professional. New York, NY, McGraw-Hill. Che, Y.-K. (1993). Design Competition through Multidimensional auctions. RAND Journal of Economics. 24(4): 668. Duray, R. (2002). Mass customization origins: mass or custom manufacturing? international journal of Operations & Production Management. 22(3): 314–328. Forza, C. and F. Salvador (2002). Managing for variety in the order acquisition and fulfillment process: the contribution of product configuration systems. International journal of production economics. 76: 87–98. Hartman, L. (1997). Tips from Attorneys on Performance Specs, the Construction Specifier, Journal of the Construction Specification Institute. Hippel, E. (2005). Democratizing innovation. Cambridge, Mass., MIT Press. Huffman, C. and B. E. Kahn (1998). Variety for Sale: Mass Customization or Mass Confusion? Journal of Retailing. 74(4): 491–513. Hurwicz, L. (1960). Optimality and informational efficiency in resource allocation processes, Mathematical Methods in the Social Sciences. Stanford University Press. Kwak, M. (2002). Potential Pitfalls of E-Auctions, Sloan Management Review. 43(2): 18. Laffont, J.-J. and J. Tirole (1993). A theory of incentives in procurement and regulation. Cambridge, Mass., MIT Press. McAfee, R. P. and J. McMillan (1987). Auctions and Bidding. Journal of Economic Literature. 25(2): 699–738.

532


Metty, T., R. Harlan, et al. (2005). Reinventing the Supplier Negotiation Process at Motorola. Interfaces. 35(1): 7–23. Milgrom, P. R. (2004). Putting auction theory to work. New York, Cambridge University Press. Moser, K. and F. Piller (2006). The international mass customization case collection: an opportunity for learning from previous experiences. International Journal of Mass Customization. 1(4): 403–409. Myerson, Roger and M. Satterthwaite (1983), Efficient Mechanisms for Bilateral Trade, Journal of Economic Theory. 29: 265–281. Piller, F. T., K. Moeslein, et al. (2004). Does mass customization pay? An economic approach to evaluate customer integration. Production Planning & Control. 15(4): 435–444. Piller, F. T. (2004). Mass customization: Reflections on the state of the concept. International Journal of Flexible Manufacturing Systems. 16(4): 313–334. Royal Swedish Academy of Sciences (2007). Mechanism Design Theory: Scientific background on the Sveriges Riksbank Prize in Economic Sciences in Memory of Alfred Nobel. Sabin D. and Weigel R. (1998). Product Configuration Frameworks: A Survey, IEEE Intelligent Systems. 42–49. Sawhney, M. (2003). Reverse Auctions Cutting Costs. CIO.com, News Article, www.cio.com/ article/29717/Reverse_Auctions_Cutting_Costs?page=1. Selladurai, R. S. (2004). Mass customization in operations management: oxymoron or reality? OMEGA, the International Journal of Management Science. 32: 295–300. Suh, N. P. (1990). The principles of design. New York, Oxford University Press. Tseng, M. M., Jiao, J., Merchant, M. E. (1996). Design for Mass Customization, Annals of the CIRP. 45/1: 153–156. Wolinsky A. (2005). Procurement via Sequential Search. The Journal of Political Economy. 113(4): 785– 810. Zipkin, P. (2001). The Limits of Mass Customization. MIT Sloan Management Review. 42(3): 81.

Author Biographies Dr. Songlin Chen is currently an assistant professor at the System and Engineering Management Division of Nanyang Technological University, Singapore. His research is focused on the design and operations of advanced manufacturing/service systems, with special interest in mass customization, collaborative engineering, supply chain coordination and contracting, etc. Dr. Chen obtained his Ph.D. degree in Industrial Engineering and Engineering Management from the Hong Kong University of Science & Technology. Previous to that, he had a Bachelor degree in Aerospace Engineering from the National University of Defence Technology in China and a Master degree in Aeronautics and Astronautics from Stanford University in U.S. Contact: www3.ntu.edu.sg/home/Songlin/ | [email protected] Prof. Mitchell M. Tseng is Chair Professor and Director, Advanced Manufacturing Institute, Hong Kong University of Science and Technology. He is also an Adjunct Professor of MIT-Zaragoza Logistic Program. Prof. Tseng started his industrial engineer-


533

ing career in developing key enabling manufacturing technologies for IT industry. Some of them, such as configuration systems for computers, diamond machining for polygons in laser printers, are still widely used in industry. After serving in industry for two decades, he joined HKUST in 1993 as the founding department head of Industrial Engineering. He is an elected fellow of the International Academy of Production Engineers (CIRP), and American Society of Mechanical Engineers (ASME). Professor Tseng is internationally known for his research in Mass Customization and Global Manufacturing. Sponsors of his research include AT & T, Astec-Emerson, Esquel, Honeywell, Lucent Technologies, Intel, SAP, Rockwell International, Liz Claiborne, Motorola, Nokia, GAP, Ford Motor, Norvullus, Tecton, Synocus, Yuesan, OOCL, Novellus, Ove ARUP, HK Air Cargo Container Limited, and various government agencies in Hong Kong, Mainland China and EU. Contact: ami.ust.hk | [email protected]


5

Rapid Manufacturing for Mass Customization

535


5.1

Extreme Customization: Rapid Manufacturing Products that Enhance the Consumer Christopher Tuck Rapid Manufacturing Research Group, Wolfson School of Mechanical and Manufacturing Engineering, Loughborough University Min-Huey Ong Rapid Manufacturing Research Group, Wolfson School of Mechanical and Manufacturing Engineering, Loughborough University Helen Wagner Rapid Manufacturing Research Group, Wolfson School of Mechanical and Manufacturing Engineering, Loughborough University Richard Hague Rapid Manufacturing Research Group, Wolfson School of Mechanical and Manufacturing Engineering, Loughborough University

Body-fitting customized products are becoming an increasingly important area of research in terms of increasing a product’s marketability and performance. Through the capture of simple scan data using Reverse Engineering techniques and the application of Rapid Manufacturing the production of bespoke components and products is possible, both technically and economically, due to the removal of labor and tooling from the manufacturing process. This paper provides a holistic view of the concept of personalized manufacturing, incorporating results from a global survey on the propensity for bodyfitting customization, specifically, on the customization of motorcycle seating. It outlines that the important geometry capture stage must capture the deformed geometry rather than simple body scan data for the customization to be effective. In addition the methods and issues associated with manufacturing personalized seating are explored and the service requirements for motorcycle seat consumers are identified, as well as providing a route to manufacture using wholly digital techniques.

Introduction Customization and, in particular, mass customization (Pine et al. 2000), have received a great deal of attention in recent years as a method for creating increased value for manufacturers and retailers alike. Many examples of mass customization use innovative supply chain concepts to produce customized products from a 537

538


range of existing "modules" (Salvador et al. 2002). These modules are often conventionally manufactured and as such incur costs for tooling during the manufacturing process. This tooling cost (for example; from injection moulding) often predetermines the necessary volume of components that have to be produced in order to manufacture parts cost-effectively and can, due to the costs of tooling, prohibit new product development and therefore stifle innovative products and particularly bespoke or tailored products. In addition, the ability to produce components that fit the customer exactly intrinsically means that the customer is intimately involved in the component design process. This paper provides an exemplar of the marriage between personalization and a method of production that does not require tooling investment. The paper not only discusses issues of a technical nature but includes discussion on the needs and wants of the customer, with respect to an individualized seat for motorcycle use and a possible method for production utilizing an additive technique commonly known as Rapid Manufacturing (Tuck and Hague 2006). Rapid Manufacturing RM encompasses a number of technologies, examples of which include: stereolithography (SLA), laser sintering (LS), fused deposition modeling (FDM) and three dimensional printing (3DP). Although all of these processes are different, they encompass the same manufacturing philosophy. Rapid Manufacturing produces components in an additive fashion, contradicting traditional subtractive (machining) and formative (moulding) techniques. RM has been defined as the production of parts or final products directly from digital data, eliminating all tools (Dekker et al. 2003; Tuck and Hague 2006). Components are fabricated by adding successive layers of material together, based on Computer Aided Design (CAD) data. From a manufacturing and marketing perspective, there are several advantages in adopting RM. Firstly, design freedom is a great advantage to RM (Hague et al. 2003). Designers are free to design complex geometries that RM machines are able to fabricate. The direct fabrication of these parts from CAD data also means that the tooling step is eliminated, hence, designers do not have to worry about whether a mould can be made for a particular design or that the number of parts that are required to make up a component both lead to an increased cost to the consumer, especially for low volume and custom components. Removing tooling means that changes to the design can be made quickly without significant effect on cost. At the same time, the long lead time for delivery of tooling can be avoided, shortening the time-to-market of a product (Hopkinson and Dickens 2003). Without tooling, it is possible to fabricate parts and products


539

in small quantities which would not otherwise be economically viable. RM enables low volume production at a more economical cost (Ruffo et al. 2005) without the cost of tooling, the cost of low volume production by RM decreases significantly when compared to traditional manufacturing such as injection moulding. RM has already been adopted in several industries, including hearing aid (Dickens et al. 2005; Wohlers 2003), automotive (Tromans 2006; Kochan 2003) and aeronautical industries (Amato 2003), for the production of some parts. Major hearing aid companies have adopted RM as their mainstream production technique. Siemens Hearing Instruments has been producing customized hearing aids using RM techniques at a production rate of 2000 pieces per week. In the aeronautical industry, environmental systems inside fighter jets are printed out by rapid prototyping machines and have led to savings and reductions in cost and production schedules by about 50%. In other industries, the US military has set up a Mobile Parts Hospital at sites in Kuwait and Iraq, printing parts for their equipment. The army is able to replace broken parts within hours instead of waiting days or weeks for the new replacement (Aston 2005). Besides industrial usage, RM has also been adopted in consumer products. MGX, a division of Materialise of Belgium, has been using the same technology for the fabrication of customized and limited edition lamps with complex designs (MGX 2007). Mass Customization The concept of mass customization is to have high volume production of individually defined goods in a cost effective way (Hague et al. 2003; Piller and Müller 2004; Pine et al. 2000). Goods and services are individualized to satisfy a very specific customer need, at an affordable price (Hague et al. 2003). In fact, mass customization has gained much popularity in recent years. A wide range of products have the customization options for the consumers; shoes, cars, watches, clothing, bags etc, are among some of the products that have caught on to the customization trend. But, are consumers ready for customization? Bardakci and Whitelock (2005) have discussed the readiness of consumers for customization based on three factors:

Willingness to pay a premium for the product

Willingness to wait to receive the product

Willing to invest time in "designing" the product. They surveyed consumers in Turkey and UK for their readiness for masscustomized cars and found that in both countries, consumers are agreed to all three

540


criteria and concluded that there is a market for mass customized cars in both Turkey and UK. Piller and Müller (Piller and Müller 2004) conducted similar survey for customized shoes and found that potentially, there is a large market for customized shoes within Europe. They defined three levels of customization:

Style – aesthetic design

Fit and comfort – tailored according to body geometry

Functionality – the functions performed by the product Their survey found that both men and women considered fit and comfort as the most important factor for customization of shoes. The survey also found that the majority of the consumers accepted that they have to pay a premium for a customized product. The above survey results and other examples of mass customization (e.g. Mi-adidas, bicycles (Moser et al. 2005), clothing (Weyel 2000; Mueller and Wohlers 2006)) do indicate a general trend towards customization and consumers' acceptances for customization. However, each product has its own unique criteria and thus different factors of success, i.e. "customization has to be customized" (Piller and Müller 2004). It is necessary for businesses to understand what their customers want before launching into customization. RM and Mass Customization The availability of customization has been possible with advances in manufacturing technology, enabling low volume production to be achieved efficiently. RM is envisaged to be the enabler for customization (Tuck and Hague 2006; Dickens et al. 2005). As discussed earlier, the development of tool-less production in RM make it economically viable for small volume production. RM would be suited to cater for niche markets requiring unique end products. This fits well with the requirements of customization, which manufactures a product or delivers a service in response to a particular customer’s needs (Pine et al. 2000). This in turn means producing a one-off item. With a greater degree of design freedom, RM potentially, will be able to cater to any geometric requirements. This paper presents Rapid Manufacturing as a possible answer to those products where tooling restricts the economic production of personalized products, specifically, the paper presents a survey to determine the propensity for customization in the motorcycle sector along with a possible manufacturing route utilizing the application of Rapid Manufacturing coupled with Reverse Engineering to offer the ability to produce body-fitting, conformal products. The survey aims to explore if there is a market for customized motorcycle seats and the willingness of consumers to pay and wait longer for a customized seat.


541

Identifying Consumers' Attitudes Towards Personalization Motorcycling is a popular sport and motorcycles are also a common mode of transport around the world. The number of motorcycles in use in Europe was about 14 million in 2003 (Association des Constructeurs Europeens de Motocycles 2004) and in the US, there were about 8.8 million motorcycles in use in 2004 (Motorcycle Industry Council 2004). The comfort of riding motorcycles for long distances and how to improve the comfort of riding have been investigated (Koyano et al. 2003; Cossalter et al. 2006; Lai et al. 2003). Previous research (Tuck et al. 2005) has shown that seat comfort can be influenced through customizing the seat according to body geometry, thus changing the pressure distribution of the occupant on the seat. To find out the consumers' attitude towards the concept of a customized seat, a global web-based survey was carried out. The survey was hosted by a motorcycle manufacturer’s website for a period of seven weeks and received around 3200 responses. From the survey, it was found that majority of the respondents (92.2%) do not share their motorcycles with another person, only 7.8% do share their motorcycles. 51.5% of the respondents said they had experience of discomfort from their motorcycle seats while the other 48.5% said they do not have such experiences. Conversely, 92% said they felt discomfort when doing long distance travelling. Do consumers support customized seats? 81.1% of the respondents supported the idea of customized seats according to their body geometry. Because the survey was hosted on a motorcycle manufacturer’s website, it was thus not surprising that the majority of the respondents owned this manufacturer’s motorcycles, which is identified as Brand A in this paper. In order to assess if this has in anyway skewed the results, the data from the Brand A motorcyclists was separated from the other results. In total, there were about 2500 Brand A motorcyclists (78% of all survey respondents) and about 700 motorcyclists owning other brands of motorcycle (20%, excluding those who did not provide an answer). Among Brand A motorcyclists, 51.1% said they felt their motorcycle seats were uncomfortable but 48.9% do not share the same view (Figure 1). The proportion is similar among the other motorcyclists, with 53% saying they felt the discomfort from their motorcycle seats and 46.9% who do not agree to that. However, the idea of customized seats does seems to appeal regardless of the brand of motorcycle owned; among Brand A motorcyclists, 80% would like to have them and 83.7% of all other motorcyclists support the idea. This is very similar to the overall rate of 81.1%.

542


Figure 1: Differences among Brand A motorcyclists and those who own other brands of motorcycles.

The survey also found that motorcyclists are already using customized seats, which are known as aftermarket seats, either bought from seat manufacturers or self-modified. The standard seats that come with the motorcycle are known as "stock seats" and about 2% of the respondents have given up their stock seat for an aftermarket seat. The main reason given for changing the seat is comfort; some personalized their seats by adding gel padding while there are others who did it for aesthetic reasons, such as changing to a leather cover or changing colors.

Figure 2: The amount of money respondents are willing to pay for a customized seat.


543

Willingness to pay and wait Having established a need for customization of seats, it is not surprising to find that nearly 79% were willing to pay extra for a customized seat on their motorcycles. Most of the respondents were willing to pay up to 250 Euros for a customized seat (66%), while about 30% were willing to pay between 250 Euros to 500 Euros and only a minority were willing to pay higher than 500 Euros (nearly 3%) (Figure 2). Likewise, the majority of motorcyclists were willing to wait longer for customized products. The majority of the respondents (64%) in the survey were willing to wait for up to one month for their customized products and 22% were willing to wait for more than a month (Figure 3).

Figure 3: The length of waiting period the respondents were willing to put up with.

Important factors to motorcyclists The respondents were also asked to rank a list of criteria which are most important to them when they acquire a customized seat: "effective" and "quality/reliability" are the highest ranking factors followed by wellbeing and injury prevention (Figure 4). This result tallies with the comments left by the respondents. 4.6 % of those who left their comments stressed on the functionality of customized seats; they wanted to know the real benefits of a customized seat, whether it will reduce the discomfort that comes with long distance travelling or it will help to reduce the stress to their body. Having identified the need and want for a customized seat for motorcycles a possible production method is proposed. RM has been identified sue to the fact that it is an agile and flexible technique for producing components without the need or investment in tooling and thus it could enable personalized manufacturing for just such an application. RM works directly from 3D CAD data

544


and as such digital techniques need to be employed to capture the geometry of the consumer.

Figure 4: Factors that are considered to be important to the motorcyclists when acquiring a customized seat.

Producing a Customized Seat RM requires the production of suitable digital data in the form of a .STL file; this file can be produced in most 3D CAD packages and then be transferred to the RM system. The STL file is then sliced by software into discrete layers that are reconstructed by the system hardware to form the physical part or component. The data can be produced in the form of a totally new CAD design or can be a reproduction of an existing shape. In the latter case, Reverse Engineering (RE) techniques are often used as a precursor to the CAD modeling task. The study described here was based around the concept of body fitting customization for a seating application for pilots, however, the principles and methodology are transferable to a number of seating applications, importantly, for where it is necessary to capture a "deformed" shape rather than the exact "standing" shape of the body. As such the subjects' interaction with the seat in question had to be carefully considered. The steps for the customization process were as follows: Geometry capture, Scanning, Data manipulation, and Manufacture-


545

Geometry capture The first step was to capture the shape data from the subject under investigation. In order to obtain the correct geometry for the seated profile, a method was devised that could reliably capture the reaction between the seat and the buttocks. In order to do this a method of "moulding" the buttock shapes whilst in contact with the seat was devised. Through studying methods of production for other types of customized seating (most notably Formula 1 Racing) a method was identified to capture the necessary geometry. An adaptation of these techniques was sought, which enabled geometry to be captured in-situ and was re-usable. The capture mechanism was based on that of a Burnett bath seat produced by RBF Products Ltd (2005). The system consisted of a polymeric bag filled with polystyrene beads with a valve placed where air can be sucked out when required. This resulted in a stiffening of the bag, thus capturing an imprinted shape (Figure 5).

Figure 5: Photograph of Burnett Seat in position, shown with a custom profile imprint.

Scanning In order to produce the necessary 3D CAD model, data for both the seat shape and the customized profile was needed. This data was captured using a 3D non-contact scanner, Model Maker X70, produced by 3D Scanners (2005). The scanner was connected to a co-ordinate measuring arm; in this case a FARO Technologies Inc (2005) Gold Arm. All scanning was done on-site, a photograph of the system in use is shown in Figure 6.

546


Figure 6: Non-contact 3D scanner in use.

The scanner system produced a series of points based on the position of the arm and data collected by the laser scanner. Such a series of points is often referred to as a point cloud. One example of the point clouds generated from the 3D scanner with a custom profile is shown in Figure 7.

Figure 7: Point cloud of custom seat profile.


547

The point cloud represents a number of individual data points collected by the scanner; each point is allocated an x, y and z co-ordinate in free space. The point cloud was only the first stage of the data manipulation process. During subsequent modification of this data, RE software was necessary to convert the point cloud data into a 3D CAD model. Data manipulation Following the data collection process, specialist RE software, Raindrop GeoMagics Studio (2005) was used to process the raw point cloud data and produce a final CAD file for the customized impressions. Point clouds were imported from the scan software and the data was subsequently processed in order to reduce the file size. The larger the file the more time consuming operations on those data sets become. Reducing the point cloud data allowed more efficient manipulation of the data. A polygonal point surface was then "wrapped" onto these points and any holes bridged or filled in order to blend the customized surface with the seat shape. The surface was then smoothed sufficiently to remove unwanted creases from the scanning operation but leaving the custom profile intact. When a satisfactory polygonal surface had been produced, a Non-Uniform Rational BSplines (NURBS) surface was created to import the file into a CAD package for further editing or manipulation. Three sets of seating data were manipulated to form the CAD models shown in Figure 8.

Figure 8: Seat CAD files.

Manufacturing Once the NURBS surfaces had been created, the part was then converted to the STL format for production on additive manufacturing equipment. The customized

548


seats were produced using a 3D Systems SLA 7000 stereolithography (SL) (3D Systems 2005) machine, the build orientation was with the seating surface facing upwards in order to minimize build time and produce a good quality upper surface, shown in Figure 9.

Figure 9: Finished custom seat.

Discussion Extreme customization takes the viewpoint that customization should take a holistic approach rather than solely be concerned with product or system design. By their very nature customized products have a symbiosis with the customer as without them we cannot have customization! In addition to this linkage products and components must also display certain attributes in order to be desirable to consumers. The survey has identified effectiveness and quality/reliability as the two most important attributes for the customized seat. Importantly, these two characteristics rank well above (over 50% c.f. 12%) features such as status and low cost. The example given in this paper utilizes body-fitting customization which includes the customer within the design process and thus engages the consumer from the outset. Extreme customization must lead to an experience which the consumer finds satisfying in order for the concept to form a successful business model. The survey carried out to assess the needs and desires of consumers has


549

lead to some exciting initial findings that are encompassed by the three following drivers:

Independence – over 90% of the respondents to the survey stated that they did not share their motorcycle with another person. This is significant as it shows that the motorcycle is a personal item and that other rider’s comfort will not be affected through the change of the motorcycle seat to a specific body geometry.

Discomfort – over 50% of those surveyed suggested that in general they did find their current seat platforms uncomfortable. This figure significantly increases, to over 90%, when long distances are taken into account.

Support – with the previous two drivers in mind over 80% of respondents to the survey supported the idea of using body geometry to change the form of the seat and therefore influence the comfort. Cost of seat and lead times are obviously important factors for consumers considering the purchase of any product, but for customized products this is amplified due to the personal nature of the product in question. Respondents to the survey have indicated that on average a price of €250 and a waiting time of 1 month would be acceptable. The price ties in with those consumers who have bought aftermarket seats (2% of respondents) who have paid around €230 for a basic seat.

Referring to previous work by Bardakci and Whitelock (2005) the survey has shown, with a limited sample, that the three characteristics for successful customization are apparent, there is a want and a willingness to be involved in the process and a willingness to pay and wait for a customized product. In addition, the concept fits Piller and Müller’s (2004) levels of customization in that it enables customization for a fit and comfort scenario. Considering the flexibility of the reverse engineering and rapid manufacturing system shown in this paper the lead time of one month from scan to delivery should be easily achievable. A similar study by Tuck et al. (2005) has shown the production of customized automotive seating to be possible within a 48hr time period. The use of RM has meant that the labor requirement usually seen in the production of customized goods was moved from the manufacturing process into upstream processes such as design. Figure 10 shows the typical timeline for each of the processes involved in the manufacture of a single seat. In total, from capturing the geometry to completing the seat the process took around 48hrs to complete. The vast majority of this time (35 hrs) was taken building the parts using the SLA 7000. However, this time required little to no operator involve-

550


ment. In fact, the major labor content was involved in the data manipulation or design phase of the operation.

Figure 10: Timeline for RM custom seating (Tuck et al. 2005).

Conventional methods of manufacture lie in a spectrum that ranges from using labor intensive processes at one end to the use of expensive general machinery at the other. Figure 11 shows a typical manufacturing paradigm for low to high variety products and respective production volumes. The dotted line is the line of least cost for the manufacturer. Traditional customization occurs in the top left of the Figure whereas mass produced items are suited to the bottom right. The labor intensive processes of professional customization and service shop manufacturing result in a very high unit cost (at least in developed economies). Mass production can result in low unit costs but only at the expense of high product standardisation. This conflict between product volume and variety has been spanned by concepts such as flexible manufacturing systems and mass customization. RM can be considered as a new type of agile manufacturing system that is capable of high product variety but without the requirement for high labor input. This is due to the fact that no tooling is required and a high degree of automation. This leads to a possible alternative manufacturing supply paradigm as shown in Figure 12. In addition to these possible benefits to the customization organization the concept of RE coupled with RM, has shown that it is possible to capture the deformed shape of the consumer. This is important, as it has been previously shown that the geometry of the seat can influence its comfort. Simply scanning the body without taking into account the forces and weight distribution acting upon the seat will not necessarily produce something that is comfortable and meets the requirements of the consumer. It must be stressed that the biggest impact factor for consumers was to produce a seat that enhanced effectiveness and thus comfort. Currently comfort


551

is a subjective process, one persons comfortable can be another’s uncomfortable, circumventing this subjectivity has been the subject of much research (Ebe and Griffin 2000). For successful customization leading to a comfort improvement further research into the mechanisms of "manufacturing comfort" will be necessary.

Figure 11: Manufacturing supply characteristics for differing product variety and volume.

Figure 12: Manufacturing supply characteristics including RM.

Caveats do exist with the customization process outlined here, particularly with the scanning data manipulation aspects of the process. These aspects of the customization process still require skilled labor in their operation and thus the process is liable to errors during scanning and operator interpretation during manipulation, this can lead to errors in the final CAD and thus the part. However, Siemens (2007) have shown that when a significant market exists for these types of customized process then autonomous processes and reliable processes can be developed. Conclusions This paper has shown that there is a both a want and need for body-fitting customized products and has provided an example of Extreme Customization principles, acknowledging and showing that the consumer must be engaged both in the concept and design of the customized product. The survey of motorcycle

552


consumers has presented a compelling case for the application of customization for motorcycle seating products, a clear need (particularly for long distance riding) has been uncovered for seats that can be supplied around the €250 price point with a 1 month delivery time. Rapid Manufacturing is an ideal candidate for the production of these customized products. Coupling RM with RE enables the production of truly conformal products that can influence the comfort and performance for the customer. This method of manufacture opposes the traditional view (Figure ) that customized manufacturing requires a labor intensive operation in order to service the consumer. Indeed, the flexibility of this labor has been taken up by the autonomous manufacturing process, in this case stereolithography. RM is truly an agile manufacturing process that can even allow the production of multiple and varied customized or standard components simultaneously. Acknowledgments The contributions of the EU Framework 6 Program project "Custom-Fit" and UK DTI Foresight Vehicle funding for the "Management, Organization and Implementation of Rapid Manufacturing" project are duly acknowledged. Thanks also go to Dr Massilimiliano Ruffo for the digital manipulation of the customised seats.

References Association Des Constructeurs Europeens De Motocycles (2004). Motorcycles vehicles in use in EU, Available from: www.acembike.org/html/start.htm. Amato, I. (2003). Instant Manufacturing. Technology Review. Aston, A. (2005). If you can draw it, they can make it. BusinessWeek. Bardakci, A. and Whitelock, J. (2005). A comparison of customers' readiness for mass-customization. European Business Review. 17: 397–410. Cossalter, V., Doria, A., Garbin, S. and lot, R. (2006). Frequency-domain method for evaluating ride comfort of a motorcycle. Vehicle System Dynamics. 44: 339–355. Dekker, C., Dickens, P., Grimm, T., Hague, R., Hopkinson, N., Soar, R., Tromas, G. and Wohlers, T. (2003). Part 7: Rapid Manufacturing. In Wohlers, T. (ed.) Wohlers Report 2003. Wohlers Associates. Dickens, P., Hague, R., Harris, R., Hopkinson, N., Tuck, C. and Wohlers, T. (2005). Part 6: Rapid Manufacturing. In Wohlers, T. (Ed.) Wohlers Report 2005. Wohlers Associates. Ebe K. and Griffin M. J. (2000). Quantitative prediction of overall seat discomfort, Ergonomics. 43(6): 791–806.


553

Hague, R., Mansour, S. and Saleh, N. (2003). Design Opportunities with Rapid Manufacturing. Assembly Automation. 23: 346–356. Hopkinson, N. and Dickens, P. (2003). Analysis of rapid manufacturing using layer manufacturing processes for production. Proceedings of the Institution of Mechanical Engineers Part C: Journal of Mechanical Engineering Science. 217: 31–39. Kochan, A. (2003). Rapid prototyping helps Renault F1 Team UK improve championship prospects. Assembly Automation. 23: 336–339. Koyano, M., Kimishima, T. and Nakayama, K. (2003). Quantification of static seating comfort of motorcycle seats. JSAE Review. 24: 99–104. Lai, H.-C., Liu, J.-S., Lee, D. T. and Wang, L.-S. (2003). Design parameters study on the stability and perception of riding comfort of the electrical motorcycles under rider leaning. Mechatronics. 13: 49–76. Moser, K., Liertz, R. and Piller, F. (2005). Integrating suppliers into the mass customization value chain: the case of Steppenwolf. 3rd World Congress on Mass Customization and Personalization. Hong Kong. Motorcycle Industry Council (2004). MIC releases new motorcycle/ATV owner survey, Press Release, Available from: www.mic.org/mic.cfm. Mueller, T. and Wohlers, T. (2006). Part 2: Service providers. In Wohlers, T. (Ed.) Wohlers Report 2006. Wohlers Associates. Piller, F. T. and Müller, M. (2004). A new marketing approach to mass customization. International Journal of Computer Integrated Manufacturing. 17: 583–593. Pine, B. J., Peppers, D. and Rogers, M. (2000). Do you want to keep your customers forever? In Gilmore, J. H. & Pine, B. J. (Eds.) Markets for One: Creating customer-unique value through mass customization. Boston: Harvard Business School Press. Ruffo, M., Tuck, C. and Hague, R. (2005). Cost estimation for Rapid Manufacturing: Laser sintering production for low-medium volumes. Proceedings of Imech E Part B: Journal of Engineering Manufacture. Salvador, F., Forza, C. and Rungtusanatham, J. (2002). Modularity, product variety, production volume and component sourcing: Theorizing beyond generic prescriptions. Journal of Operations Management. 20: 549–575. Tromans, G. (2006). Automotive Applications. in Hopkinson, N., Hague, R. & Dickens, P. (Eds.) Rapid Manufacturing: an industrial revolution for the digital age. John Wiley & Sons. Tuck, C., Campbell, R. I., Hague, R. and Ruffo, M. (2005). Customized aircrew seating utilizing rapid manufacturing. Rapdasa 6th Annual International Conference on Rapid Product Development. Pretoria, South Africa. Tuck, C. and Hague, R. (2006). The pivotal role of rapid manufacturing in the production of cost-effective customized products. International Journal of Mass Customization. 1: 360–373. Weyel, I. (2000). Digital made-to-fit suit. Elsevier. Wohlers, T. (2003). Words of wisdom: Rapid Manufacturing on the horizon. Plastics Machinery and Auxiliaries.

Author Biographies Dr Christopher Tuck is a Lecturer in Innovative Design and Manufacturing in the Rapid Manufacturing Research Group at Loughborough University. Chris has been involved in

554


the additive manufacturing field since 2003. After completing a BEng in Materials Science and Engineering from Brunel University in 1998, he gained an Engineering Doctorate from Cranfield University in 2003 for his work on novel Fabry-Perot sensor production methods, this work was sponsored by a number of industrial companies and was successfully patented.Since joining Loughborough in April 2003, Chris has gained a great deal of experience and knowledge in the field of Rapid Manufacturing (RM). He has successfully performed research projects in the areas of management, organisation and implementation of RM for the DTI Foresight Vehicle programme (ManRM) and EU Framework 6 (Custom Fit). Chris has continued this work and has a reputation in the areas of economics, customisation and implementation of RM. Contact: [email protected]

Dr Min-Huey Ong is a member of Rapid Manufacturing Research Group at Wolfson School of Mechanical and Manufacturing Engineering in Loughborough University. Helen Wagner is a Research Associate, and member of both the Rapid Manufacturing Research Group and the Manufacturing Organization Research Group at Loughborough University. She has recently worked on the EU Framework 6 Custom-Fit project looking at the implications of customization on business and the consumer, and is currently working on the Innovation and Productivity Grand Challenge project, and is also in the process of completing a PhD in the field of Rapid Manufacturing management. Helen’s major academic interests lie in change management and the implementation of modern manufacturing concepts. Contact: [email protected] Dr Richard Hague is Professor of Innovative Manufacturing and Head of the worldleading Rapid Manufacturing Research Group at Loughborough University. He has worked within Rapid Prototyping (RP) and Rapid Manufacturing (RM) research since 1993, when he joined the RP Group at Nottingham University to undertake PhD research into the use of stereolithography patterns in investment casting; this work was successfully patented and is now licensed to 3D Systems as QuickCast 2.0.Richard has significant experience in leading large and complex research projects with multiple partners and is Principal Investigator on over £4M of EPSRC (UK), DTI (UK) and EU funded research projects. His research is focused on future manufacturing technologies and he is internationally recognised as instigating and leading work within the design, implementation, materials, textiles and customisation aspects of RM. Richard is an EPSRC College member, has multiple academic journal and international conference publications in the area of Rapid Manufacturing and is referee to several key international academic journals and conferences; he also organises and Chairs the annual International Conference on RM. Richard sees Industrial collaboration as key to academic research and was instrumental in co-founding and managing the highly successful in-house Knowledge Transfer Network that is part of the RMRG – the Rapid Manufacturing Consortium (www.lboro.ac.uk/ departments/mm/research/rapid-manufacturing). Contact: [email protected]

5.2

e-Manufacturing – Making Extreme Mass Customization Real by Laser-Sintering Christof Stotko EOS GmbH – Electro Optical Systems, Germany Andy Snow EOS GmbH – Electro Optical Systems, USA

e-Manufacturing means the fast, flexible, and cost-effective production directly from electronic data. Laser-sintering is a key technology for e-Manufacturing. With these systems, a complex design idea can be turned into reality directly by solidifying plastic or metal powders or foundry sand. This allows for manufacturing almost any shape, at any stage of the product life cycle in any industry. With e-Manufacturing shall provide competitive advantage in a business environment that is dominated by ever-decreasing product life-cycles and increasing numbers of product variants. By adding freedom of design and flexibility, laser-sintering helps making products convincingly attractive, both from the manufacturers' and the customers' points of view. The technology allows for designs that are sheer impossible with other technologies – including living hinges. This value-add stems from increased functionality, vivid design, and reduced delivery times. At the same time, laser-sintering allows for manufacturing customized products at fewer costs than with conventional manufacturing methods. Saving on molds reduces time and costs. Economies of scale are fading thus liberating manufacturing decisions from lot size optimization, forecast accuracy and break even points. The article provides an introduction into the technologies behind laser-sintering and shares results from a number of use cases in various industries.

Integrating Customers Into Manufacturing Mass customization aims at "[...] that the same large number of customers can be reached as in mass markets of the industrial economy, and simultaneously they can be treated individually as in the customized markets of pre-industrial economies." (Davis 1987: 169). This calls for leaving the "cow-path" and looking for new ways of doing business. This includes business models, marketing and sales activities, product development, logistics, etc. but also manufacturing operations – the focus of this paper. To our opinion manufacturing operations have a great share on making mass customization business model profitable. However, drastic re-thinking the ways products are manufactured, is required. 555

556


"Just as mass production was the hallmark of yesterday’s Industrial Age, mass customization promises to dominate the modern stage of […] the information Age." (Cox/Alm 1998, p.8). This calls for looking into manufacturing that is enabled by electronic media. We term this as e-Manufacturing and define it as the fast, flexible and cost-effective production directly from electronic data. Using manufacturing methods that can directly be fed by customers' input to our opinion is a vital prerequisite integrating the customer into manufacturing operations. Sufficient customer interaction is one key to success in Mass customization: "As customer interaction and integration is only possible due to low transaction costs provided by electronic commerce technologies, economies of integration can be seen as a premier cost saving potential within electronic business." (Piller/Möslein 2002, p.219). Manufacturing methods differ in its feasibility of directly taking customer input. Tool based manufacturing such as injection molding prohibits customer interaction in that sense because of cost and time intensive tool design and manufacturing. CNC machining in that sense can directly take customer input, however, requires knowledge by the customer on how to properly machine with CNC algorithms. To our opinion layer manufacturing technologies are best suited to directly taking customer input to manufacturing. On the one hand side, layer manufacturing allows for great freedom of design, thus liberating the customer from required expertise on follow-up manufacturing limitations. Purely designing for the end product is sufficient for feeding layer manufacturing technologies. We regard this paradigm shift of how products are designed as "design driven manufacturing".

Manufacturingdriven design

Design-driven manufacturing

Figure 1: Paradigm shift in design and manufacturing.

Helander and Jiao point out that product development is essential for making mass customization happen: "[…] we approach mass customization from a product development perspective. Essentially, the strategy is to include customers in the product development life cycle by proactively connecting customer needs to the capabilities of a company." (Helander/Jiao 2002, p.718).


557

To us, layer manufacturing enables mass customizers for complete customer integration from product design through manufacturing operations. By doing so layer manufacturing can help executing a basic requirement for making mass customization profitable: "Interaction means that the consumers now take part in activities and processes which used to be seen as the domain of the companies." (Wikström 1996, p.361). While there are numerous layer manufacturing methods (see Noorani 2005 e. g. for an overview) not all of these are equally suited for making mass customization real. To our opinion among all layer manufacturing methods laser-sintering is best suited to making customer integration into manufacturing real. While Noorani gives a detailed evaluation of the different methods we want to highlight these points giving laser-sintering (plastics and metal) an edge to our opinion:

Fast and stable processes

Sufficient properties of materials available to laser-sintering in metal and plastics

Huge installed base around the world

Great amount of innovation on systems, software and materials

In the following we want to detail on laser-sintering as an enabler to mass customization manufacturing (Noorani, Stotko, Black 2006). In the following we want to briefly describe the process and then detail on its effect on profitable business models in mass customization. Laser-Sintering Basics Laser-sintering working principle For the purpose of this paper we only want to explain in brief how laser-sintering works. For detailed functional description see e. g. Hopkinson/Dickens 2006 or Noorani 2005. The three dimensional description of a product is converted into a set of slices that each describes a cross section of the part in a defined height (Junior/Shellabear 2003). In state-of-the art EOS systems for plastics laser-sintering this height typically is 0,1 mm in metal 0,02 mm. The laser-sintering machine now builds up theses slices layer by layer to create the desired object. In each layer the powder (metal or plastics) is fused using laser energy. With the help of a scanner the laser energy is "printed" onto the powder layer and by this creates a solidified layer that later will be part of the finished product. The next layer is then produced on top of

558


the first and so on until the build is complete. Figure 2 illustrates the repeating process of building a layer. The process takes around one hour to build between 15 and 25 mm in height in when looking into laser-sintering of plastics. Current EOS innovation has brought up this number significantly with the introduction of FORMIGA P 100, EOSINT P 390 and P 730 systems.

1. Exposure

2. Lower Platform powder / sand hopper

laser scanner lenses

recoater part container

4. Recoating

3. Dispensing

Figure 2: Working principle of laser-sintering.

In the end, you have the solidified products embedded in the powder that was not solidified. Laser-sintering plastics this powder partly can be reused in the next run. It can completely be re-used when laser-sintering metal. In a sieving station you separate the solidified products from the unsolidified powder remnants. In a glove-box the parts are cleaned from adhesive powder remnants by compressed air carrying small glass peens. Then the products are already in a state ready to sell. If a special treatment to the surface was necessary, several post processes are available polishing the surface to almost any quality necessary in order to satisfy customer needs. Building a Business Case on Laser-Sintering When looking into what laser-sintering can do we can distil its value down to: Any shape, anytime, anywhere. "Manufacturing even the 'impossible' to us is the most essential value add when looking into laser-sintering as a mode of

559


manufacturing in mass customization. To us "Any shape" has great value to mass customization and eventually may lead to extreme mass customization.

Laser-sintering value add

Any shape

Anytime

Anywhere

Manufacture even the „impossible“

In every phase of the product life cycle

In every industry

Figure 3: Laser-sintering value added.

Designs that are not suitable for conventional methods (e. g. undercuts, inner structures, cooling ducts, etc. ) can be made possible. Or think of the fancy design the Dutch based designers "Freedom of creation" (FOC) create for the fashion industry as illustrated in Figure 4. Customizing these products is possible at little to no extra costs. Think of varying size and shape of the chain mail, size of the handle or customized accessories attached to the handbag.

Figure 4: Handbag. Courtesy of Freedom of creation (FOC), Amsterdam, The Netherlands.

560


While freedom of design leads to fancy products and extreme mass customization, consolidation of parts helps reducing follow-up assembly costs. In the following example by Hettich the number of parts was reduced from 32 in the conventional manufactured design down to 3 in the e-Manufacturing solution enabled with laser-sintering.

Figure 4: Lavamat. Courtesy of Andreas Hettich GmbH, Tuttlingen, Germany.

By making literally any shape possible, laser-sintering allows for focusing purely on the end product not on manufacturing methods limiting freedom of design. This also helps customization. In this special case, customization occurs according to the region the centrifuges are sold to. All major economies across the world have different sizes and shapes of the blood bags to be centrifuged. Theses differing requirements can be taken into account in product development and be implemented at ease with layer manufacturing. "Anytime" refers to e-Manufacturing employed along the product life-cycle. In product development e-Manufacturing by laser-sintering results in prototypes for market research purposes. Design changes can be made easily and thus liberate decisions from amortization periods of sunk costs (e. g. tools). If the customer


561

requires several iterations on her way to extreme mass customization, she can easily get them. Laser-sintering has become a viable mode of series manufacturing. FOC and Hettich are good examples for series of high value products manufactured by laser-sintering. Sirona as illustrated in Figure 7 poses an example for mass manufacturing of extremely customized dental restorations. When looking into the extreme mass customization requirements of medical implants (e. g. knee, finger, etc. …), it becomes obvious that laser-sintering not only is a manufacturing method for small series of identical products, but for an unlimited volume of mass customized parts. Finally spare parts can be manufactured with laser-sintering if original parts are not available anymore or if a renewed design of the spare part leads to better results than with the original geometry. The following example shows a redefined spare part created by Junior & Tacke for a slideshow tripod with advanced product features compared to the original part.

Figure 5: Tripod spare part. Courtesy of Junior & Tacke, München, Germany.

"Anywhere" describes the ubiquitous applicability of laser-sintering. To us these sectors can benefit most from laser-sintering: Aerospace industry where complex parts are built in rather small quantities. See www.boeing.com/news/frontiers/archive/2006/june/i_ca2.html for a case study by a major aircraft manufacturer. In this case study laser-sintering was used for

562


"…redesign of its 777 interiors from an outside firm", a typical mass customization requirement in the aerospace industry. Medical industry can especially value from laser-sintering because customized implants can be manufactured directly with laser-sintering. One Dircet Metal Laser-Sintering (DMLS) system manufactures up to 250 individual units in about eleven hours. This results in approximately three minutes build time per unit. Operating unattended, one DMLS system can run two manufacturing cycles per day, thus producing up to 80,000 single units per year. The process requires human operation only for loading and unpacking. This makes laser-sintering a true industrial production process where high productivity meets consistent quality standards at reduced costs. Porcelain fused to metal (PFM) restorations manufactured by the DMLS process have consistent quality only achieved with CAD CAM. Dentists will appreciate the consistent fit and margin lines providing more value than simple cast parts. The 0.1 mm spot sized fiber laser of the DMLS system allows for a typical accuracy of +/- 20 micron. Thus, DMLS allows for more precise manufacturing of this critical region than most other metal manufacturing alternatives. Lasersintered metal substructures are 100 % dense. They accurately match the plaster model and are free of internal stresses formed in the casting process. The surface of the copings has been proven to increase ceramic adhesion. The PFM materials offered by EOS are 100 % biocompatible and certified for use in the dental industry.

Figure 6: Left: Knee implant. Right: Dental crowns. Courtesy of Sirona, Bensheim, Germany.


563

Special purpose equipment industry also can benefit to a great extent from lasersintering because extremely mass customized parts are required. The part in Figure 7 – a handle for a wingbox drilling device used in aerospace industry – created by Lübbering utilizes on the benefits of freedom of design. Replacing a previous sheet metal solution, laser-sintering allows for consolidation of parts by integrated air ducts (p = 6 bar) and cable ducts. On top this device is lighter than the previous solution thus contributing to workers' ergonomics. Customization can occur in terms of fitting the handle to the individual ergonomics of the machine operator. Customization can also occur in terms of branding the handle to either the system manufacturer (e. g. Lübbering) or to the product being created (e. g. A 380). Customization can also occur in terms of usage. Imagine one version with an integrated vacuum cleaning off the chips produced while drilling.

Figure 7: Special purpose equipment aerospace. Courtesy of Johannes Lübbering GmbH, Herzebrock-Clarholz, Germany.

Figure 8 shows sieve used in the food packaging industry by an undisclosed OEM saves 90 % on costs compared to the conventional milling solution. Additional value will be created by extremely mass customizing the geometry of this sieve exactly matching the requirements of the food being packaged. Laser-sintering calls for a "round wholes are so yesterday" attitude. This means wholes can have conical shape or differ in size and geometry all over the device. This can lead to customization in the food and beverage industry. Think of having pasta shaped to the latest hype on "Harry Potter" or think of having your customized pasta for the event of a family event like a wedding. Also branding pasta for the event of an international user meeting can become possible and economical feasible by leveraging layer manufacturing in the special purpose equipment industry.

564


Figure 8: Special purpose equipment food packaging.

Success Factors of e-Manufacturing e-Manufacturing with laser-sintering enables for creating convincingly attractive products. To our opinion products are convincingly attractive if they offer value to the customer that results in an additional willingness to pay. With respect to the above shown examples the FOC handbag is a prime example for tickling willingness to pay. Also customized implants result in increased willingness to pay as these better fit to the patient thus allowing e. g. a knee implanted patient taking up his previous sporting activities. The second aspect to a convincingly attractive product is the aspect of reduced manufacturing costs. While FOC and customized implants defy comparison for the lack of adequate manufacturing methods that could match the cost of lasersintering, the other examples lead to reduced manufacturing costs. Hettich not only saved on 29 parts, they also reduced assembly time and cost. On top they produced with less finishing effort than with the previous solution. Lübbering dramatically saved on costs compared to the sheet metal solution that was replaced by laser-sintering. Saving 90 % of the costs of direct manufacturing of the sieve at the food packaging company makes this device extremely attractive to the user. When looking at the common success factors of this limited assortment of casestudies (see www.eos.info/applications.html?L=1 for more) on convincingly attractive products by laser-sintering we can identify these success factors:

Products where specifically designed or re-designed for laser-sintering thus utilizing the freedom of design offered

565


By doing so design was trimmed to material properties matching product requirements

Re-designing parts for laser-sintering was done for parts with a major value contribution to the end product assembly

Extreme mass customization was chosen where it drastically contributed to the product value

When looking into extending the dialogue beyond today’s boundaries and to explore the future of mass customization, laser-sintering has a significant impact on manufacturing processes of customized products. The business case: Convincingly attractive products Customers’ point of view Increased willingness to pay

Reducing manufacturing costs

Creating convincingly attractive products

Manufacturers’ point of view

Figure 9: Convincingly attractive products.

Future Applications There is a plethora of future applications out there that can be addressed by layer manufacturing in general and by laser-sintering in specific. Here we want to sketch out some ideas that might become real in the future by using layermanufacturing in the sense that a unspecified heat source is used for solidifying any material that is provided in layers ranging in thickness from < 1 µm to 200 µm. These applications can be grouped along the three major starting points of mass customization, fit, function and design (form). Fit: The "Fit" requirement can as an example refer to humans. There gloves used in an augmented reality set-up can be tuned to the wearer and manufactured by laser-sintering. Also sports equipment can be tuned to fitting the user. Think of all kind of handles to vehicles (such as bi-cycles, rollers, …). But also think of

566


footwear customized in numerous respects to the demands of the wearer (see www.prior2lever.com). Also protective equipment used in numerous activities (e. g. American football, Hockey …) can be tuned to the fit of the athlete. Freeclimbers might see additional value from customized handles, fixtures or the like manufactured by laser-sintering. Cyclists can benefit from aerodynamic helmets just as apnoe divers. Prosthesis can be customized to a better fit at the patient. "Fit" can also refer to technical fit, e. g. corner blends. These items protect valuable equipment (e. g. navigation device) in un-spaceous environments (e. g. in airplanes, submarines or bi-cycle frames). Function: With "Function" laser-sintering can make applications come true undreamed of today. Think of bionic design that optimizes a function like "strength". A layer-manufactured turbine blade with inner spongiosus structures making up for the required strength covered by a free formed area optimized to aerodynamic requirements. Such turbine blades will help saving fuel as the accelerated mass is reduced. Also functions in numerous springs can be tuned to requirements such as spring-force, size, shape and the like. Design: Regarding "design" think of jewelry being made by layer manufacturing but also think of customizing items like house-hold appliances or furniture handles tuned to customized fit, design, and function. With design also a bionic function can be integrated at additional value (see www.assaashuach.com as an example).

References Cox, M. W. and Alm, R. (1998). The Right Stuff. America’s Move to Mass Customization. Dallas: Federal Reserve Bank of Dallas. Davis, S. (1987). Future Perfect. Reading, Mass.: Addison-Wesley. Helander, M. G. and Jiao, J. (2002). Research on E-product development (ePD) for mass customization. in: Technovation. 22(11): 717–724. Hopkinson, N. and Dickens, P. (2006). Emerging Rapid Manufacturing Processes. In: Neil Hopkinson; Richard J. M. Hague; Phil M. Dickens (eds.). Rapid Manufacturing. An Industrial Revolution for the Digital Age. Chichester: Wiley & Sons. Junior, V. and Shellabear, M. (2003). Enabling Mass Customization Through Laser Sintering. From R&D to Manufacturing. World Congress on Mass Customization and Personalization, München. Noorani, R. (2005). Rapid Prototyping. Principles and Applications. Hoboken, N. J.: Wiley & Sons. Noorani, R. and Stotko, C. and Black, M. (2006). e-Manufacturing using rapid prototyping. Global Congress on Manufacturing and Management, 19.11.06 – 22.11.06, Sao Paulo. Piller, F. T. and Möslein, K. (2002). Overcoming the Efficiency Paradox: Competitive Strategies for the Management Education Industry. in: SMS 22th Conference 2002, Paris.


567

Wikström, S. (1996b). Value Creation by Company-Consumer Interaction. in: Journal of Marketing Management. 12(5): 359–374.

Author Biographies Dr Christof M. Stotko is the global marketing manager at EOS. The company develops Laser-Sintering Technologies and sells e-Manufacturing solutions that enable service providers and OEM’s to efficiently develop and deliver high quality products. Contact: [email protected] Andrew Snow is a Sales Manager with over 20 years experience in new business development and works for EOS of North America, Inc. The company develops Laser Sintering Technologies and sells eManufacturing solutions that enable service providers and OEM’s to efficiently develop and deliver high quality products. As a Sales Manager he is responsible for the sales and strategic planning of systems produced by EOS. Andrew Snow has a B.S. in Management from the Lesley University of Cambridge Massachusetts (1996). Contact: [email protected]

5.3

RepRap: The Replicating Rapid Prototyper: Maximizing Customizability by Breeding the Means of Production Ed Sells Mechanical Engineering Department, University of Bath, United Kingdom Sebastien Bailard Supermeta Fabrication, Canada Zach Smith RepRap Research Foundation, USA Adrian Bowyer Mechanical Engineering Department, University of Bath, United Kingdom Vik Olliver Diamond Age Solutions Ltd., New Zealand

This paper describes progress on RepRap, the replicating rapid prototyper. RepRap is a filament-deposition rapid prototyping machine that has been designed to manufacture the majority of its own parts. All other parts of the machine are standard materials and components available everywhere in the world. RepRap is intended to maximize the customizability of both the products that it makes and also itself. It achieves this by several complementary mechanisms: it is intended for individual (as well as industrial) use, so its users may employ it to manufacture whatever they want; it can make copies of itself, and those copies can be customized; it is extremely low cost, and so ownership can be widespread; and finally it is open-source, so all its designs and software are available for modification. Prototype RepRap machines have been built and are described. These have made parts for themselves and each other, and this is depicted. The design principles and specifications of the machine are given. The paper concludes with a discussion of the possible impacts that the machine may have on personal manufacturing and product customization.

Introduction Consider the wolves that you see being led down the street every day. Their appearance ranges from the whimsical to the grotesque, and their adult body size covers a span unmatched by any other species. This virtuoso and antic variety was 568


569

created by one of humanity’s oldest and grandest technologies: genetic engineering. We have been customizing life since the invention of agriculture in Mesopotamia around 9500 BCE (Bellwood 2005). Nowadays much of that customization is done industrially, though the techniques still retain an important characteristic that they have had over the millennia: they can be done by a single person possessing equipment no more advanced than a breeding pen or a potting shed. Even the latest twist of the helix ─ direct manipulation of DNA ─ requires modest wherewithal well within the resources of an individual (Dyson 2006).

This first machine was fabricated using polymer parts from a commercial 3D printer. The machine is roughly a 0.5 m cube. A 12v line out of an old PC supply is used for power.

Figure 1: RepRap Version 1.0 "Darwin".

570


The exuberant variation that we have achieved in the customization of our domesticated plants, animals and microbes is completely reliant upon one phenomenon: the fact that they can copy themselves. How may we extend that phenomenon, and hence that degree of customization, to the products of engineering? One way would be to design a general purpose manufacturing engine that could also make copies of itself. This is not a new idea ─ Samuel Butler put selfreplicating machines into his novel Erewhon in the nineteenth century (Butler 1872), and John von Neumann did extensive theoretical studies in the middle of the twentieth (von Neumann 1966). For a comprehensive review of the history and technology of self-replicating machines, see the book by Freitas and Merkle (2004). However, nobody has yet made a self-replicating machine with the intention of using it as an everyday piece of production technology. That is what this paper is about. The RepRap Machine RepRap is short for replicating rapid-prototyper. It is intended to be a practical self-replicating 3D printer. Specifically, it is a fused filament fabrication (FFF, Fused Filament Fabrication is sometimes called Fused Deposition Modeling (FDM), but that phrase is a registered trademark of Stratasys Inc. FFF is an open phrase.) rapid prototyping (RP) machine that has been designed to be able to make most of its own parts. Figure 1 shows the first version of the RepRap machine constructed by the authors. At the time of writing about a hundred RepRap machines are under construction or in use round the world. Method and Design As can be seen in Figure 1, the bulk of the machine is a conventional Cartesian robot. This moves heads that extrude the build materials. Right from its instigation (Bowyer 2004) the RepRap project has deliberately subjugated theoretical perfection to the requirements of practicality. In particular, the following principles were adopted: Self-replication is distinct from self-assembly. The fact that all organisms (except viruses) do both is not a reason to conflate these ideas. Machines are much better at making accurate parts than are people; people are much better at putting parts together than are machines. It therefore makes sense to have a collaborative division of labor. Indeed, the proposed collaboration is more than that ─ it is a symbiosis between two replicators. People will help RepRap machines to reproduce by assembling them in return for the other goods that they make. There


571

is an exact analogy between this and the symbiosis between ─ for example ─ flowers and insects: insects help flowers to reproduce in return for nectar. Because of this the RepRap design (at least initially) will concentrate on making its parts; its owner will assemble it. Use some bought-in parts. Laboring extravagantly to have the machine make parts that are already ubiquitously available and cheap would waste better-directed effort. For example, it would doubtless be possible to have the machine make its own fasteners that could be used in place of conventional nuts and bolts. But nuts and bolts are available at insignificant cost even in the poorest and most deprived places, so there is no immediate practical advantage in making replacements. (For the benefits that this technology might bring to developing nations, see the section

Figure 2: The RepRap polymer extruder assembled (left) and its internals (right). In addition to a US$4 geared motor it only has one moving part ─ a threaded drive rod mounted on two half-bearings and rotated by a flexible coupling (a length of steel cable) to allow the motor to be offset. This threaded rod bears against the 3mm diameter polymer supply (which can be seen coming in from the left in both pictures) and drives it down into a melt chamber heated using nichrome wire. The chamber is just another threaded rod that has been drilled down the middle and that ends in a 0.5mm extrusion nozzle. The offsetting is to allow the polymer to run straight down the drive thread. This is not necessary for the polymer used, which is flexible, but will be useful for other stiffer materials. The large grid squares are 10 mm.

572


on Implications below.) Thus the machine will make all the parts that are specific to itself, but things like fasteners, steel rods, stepper motors, and microcontroller chips are added. This means that, while RepRap may not be a philosophically perfect self-replicating machine, it is a practical one. Distribute the control logic. As one of the purposes of the machine is to allow the maximum customization ─ not just of the products that it makes, but also of itself ─ it was decided that each component would have its own microcontroller (a PIC16F648) connected in a token ring along with the controlling computer. This allows the addition of, for example, extra extruder heads very simply. The distribution is taken to the lowest level, with a separate controller for each axisdrive stepper. The Bresenham DDA that generates movement paths for the extruder heads works round the ring with one extra synchronization line between the three axes. Other than that one wire, there are only four other wires that connect everything in the machine (ground, +12v, and two serial data lines).

Figure 3: The RepRap GUI on the host computer. The user loads STL files of parts to be made into the left hand window and uses the mouse to place them in the position and orientation where they are to be built on the build base. Starting the build both sends instructions to the RepRap machine and runs a simulation showing progress in the righthand window (the little inverted U shape is the extruder clearing itself at the start). The host software is written in Java for platform independence.


573

Recently the open-source Arduino microcontroller (www.arduino.cc) has been programd and used by one of us (ZS) to drive the machine as a compatible (at the inter-machine communications level) alternative to the PICs. Open source the project. This sounds like a political choice rather than a design principle. And indeed the initial decision to open-source RepRap was taken because it is potentially a powerful technology, and a good way to make bad things happen with a powerful technology is to put it at the disposal of some people and not of others. But it was almost immediately also realised that it is not practical to attempt exclusive sales of a machine that can copy itself (sales figures would total one), nor was it practical to attempt to protect any "Intellectual Property" in the machine (as it can copy itself, this would just be a recipe for spending lengthy periods in courts of law attempting to prevent people from doing with it the very thing that it was designed to do). Here are the RepRap machine’s specifications:

Working volume: adjustable, but nominally 230x230x100 mm3

Working materials: Polycaprolactone, polylactic acid, ABS and HDPE

Material handling: Two material deposition extruders, user exchangeable

Configuration: 3-axis Cartesian drive using stepper motors

Line and space: 0.5mm and about 0.2mm

Feature size: about 2mm

Positioning accuracy: 0.1mm

Layer thickness: adjustable, but nominally 0.3mm

Computer interface: RS232 (or USB -> RS232) at 19200 baud

Power supply needed: 8A max, 3A continuous at 12V DC

Driving computer and operating system needed: Microsoft Windows, Linux, Unix, or Mac.

Considerable thought, trial, and error went into the design of the polymer extruder head. Figure 2 shows the release design of one of these (left), and a dismantled view (right). Initially the polymer that was used as a build material was polycaprolactone. This is a very tough nylon-like polymer that has the added advantage of melting at about 60 oC. This low melting-point makes it very easy to work with in the machine. But polylactic acid, ABS, and HDPE have now also successfully been used in it. Figure 3 shows the GUI that the user of the machine sees when building objects with it. The only actions required are loading STL

574


files, positioning and orienting them on the build base, and instructing the machine to start. Results Figure 4 shows a close-up of a prototype RepRap machine with the extruder head described above (made in ABS by a commercial RP machine) on the right, and on the left an identical head made by that first head in polycaprolactone. The new head that the machine has made for itself is shown starting to extrude polycaprolactone. Figure 5 shows the first "child" RepRap machine made by a "parent" RepRap machine. The child parts were all made in polylactic acid by one of the authors (VO) on his RepRap Version 1.0 "Darwin" machine. Extensive and detailed reporting of all the many experiments being conducted with the machine and its components (which are too numerous to include in this paper) can be found on the project blog at blog.reprap.org

Figure 4: An extruded extruder extrudes. The RepRap extruder head on the right was made in a commercial rapid prototyper (the transparent plastic shroud is to reduce cooling by air currents). That right-hand extruder then made the extruder on the left, which is shown starting its first test extrusion for itself. Both heads are mounted in a prototype RepRap machine.


575

Future Developments The first release of RepRap was in Summer 2008, after the achievement of selfreplication. All its designs, software and documentation are on the web at reprap.org. The next step of the project will be to add a support material extruder. Overhangs of 45o and steeper can be built without support, but shallower angles require it. We propose to rapid-prototype a paste extruder similar to the one used in the Fab@Home project (Malone and Lipson 2006), except that ours will employ a flexible sack containing the support paste within a pressure chamber filled with air at about 0.5 bar. The pressure will force the paste from the sack through a solenoid clamp on a flexible silicone tube to make a valve, and then finally out of a nozzle. In keeping with the need to use universally-available materials, the sack will be a child’s balloon, the pressure chamber will be a halflitre fizzy-drink bottle, and the compressed air will be provided by a bicycle pump. The support paste will harden on contact with the outside air, and will be water soluble for easy removal. As described below, future developments of RepRap after its first release will depend much more on its user community than on its original creators. However, some work has already been done on a second release of the machine to incorporate a low-melting-point metal alloy in the structures the machine builds for use as an electrical conductor (Sells and Bowyer 2006). This would allow the machine to make electrical circuits (including things like IC holders) in three dimensions in the body of otherwise mechanical parts. Switching from the initial polymer, polycaprolactone, to polylactic acid has important ecological and economic implications. Polylactic acid has a rather higher melting point, but has the advantage that it can be synthesised by fermentation from starch (corn/maize or potatoes, for example). The RepRap machine would make the fermenter, of course. This would mean that anyone with a RepRap machine and a few tens of square metres of land would not only have a self-replicating manufacturing machine, they would also have a self-replicating source of build material. Polylactic acid is (like polycaprolactone) fully biodegradable. This gives the user of such a machine a local route to recycling involving no transport or processing. Old or broken products would simply be thrown on a compost heap to prepare for the next corn planting. The authors are also experimenting with a thermoset resin (EcoComp) derived from plant oils (Sustainable Composites 2008). This sets under the action of UV light, and we shall use the paste extruder described above with the addition of a ring of UV LEDs around its nozzle to deposit the resin. This will be mixed with a glass filler to make a self-supporting paste. Conventional UV LEDs with a

576


wavelength of 400 nm will not set the polymer, but 370 nm ones (for example the NSHU550A) work well. While a biodegradable plant-derived thermoplastic like polylactic acid is carbon neutral, a plant-derived thermoset like EcoComp that is not biodegradable locks away atmospheric carbon, and so is carbon positive. Implications Almost all current manufacturing systems (for example CNC lathes or injection moulding machines or chip fab lines) are geared towards the mass production of identical parts, and all such machines make goods in an arithmetic progression. But a machine that can copy itself can grow its numbers in geometric progression, and the goods it produces can also grow in geometric progression. Any geometric progression, no matter how slow, eventually overtakes every arithmetic progression, no matter how fast.

Figure 5: The first (parent) RepRap machine is on the left. The first child RepRap machine made by a parent RepRap machine is on the right. The child machine made its first grandchild part at 14:00 hours UTC on 29 May 2008 at Bath University in the UK, a few minutes after it was assembled.

But having goods produced in a geometric progression is something that humanity has experienced for a long time. As was mentioned in the Introduction, agriculture goes back twelve millennia, and is exclusively concerned with entities that copy themselves and thus grow in number geometrically. Self-replicating manufacturing technology makes engineering much more like agriculture. But whereas agriculture (traditionally) takes generations to customize its products by selective


577

breeding (and quite a while to do it by recombinant DNA techniques), a selfreplicating manufacturing machine can have a new part designed, then made, then fitted, in an afternoon. Its owner can also customize the products that the machine makes with equal facility. Note that replicating production methods are not necessarily efficient in every aspect of their operation. For example, they facilitate distributed local production, thus reducing transport energy requirements and CO2 emissions. But they implicitly lead to a very large number of production machines, each of which requires its own supply of energy. In this context it is perhaps worth mentioning that the RepRap machine consumes 40 W, and can be run from an old car battery charged by a solar photovoltaic cell. The open-source nature of the RepRap project means that many design improvements will be posted back on the web. Owners of old-version machines will be able to use those machines themselves to upgrade to the latest design. RepRap will evolve, like the wolves of the Introduction turning into dogs, by artificial selection. This evolution will almost certainly be taken in many different directions. Some possibilities are: reducing the number of bought-in parts, increasing the size of objects that can be made, making the machine simpler to assemble, refining its resolution, and increasing the number of materials that can be processed. The authors see one of the advantages of creating a self-replicating machine as being that, once the first design has been released, they can sit back and let Charles Darwin (in the guise of hundreds of thousands of highly-motivated tinkerers) take over the job. The target cost for the bought-in parts and materials for one RepRap machine is US$500, which is well within the resources of a single individual in a developed country. And at this level small communities of people even in the most deprived parts of the world should be able to afford a machine. This should allow them to place one foot on the manufacturing ladder that has made the rest of humanity rich. And their labor costs allow them to undercut everyone else when they make products using the technology, so this should give an economic boost where it is most needed. Such low labor costs often fail to give the advantage that they should because of the high capital cost of starting a manufacturing venture. With RepRap that limitation should not arise. RepRap will put individual cottage industry on a closer footing with conventional big manufacturers. This will be a general phenomenon with consequences for manufacturers with large production runs. These effects will be even more significant in small and niche markets, which currently generate more expensive goods due to lower output and the lack of economies of scale. For example, there

578


are people in both the developed and undeveloped world who cannot afford Braille typewriters, sophisticated limb prostheses, and page-turning machines. Inexpensive 3D printing will make these objects more widely available. The open-source nature of RepRap means that anyone who owns one is free to use it to copy itself and to give those copies to friends. Indeed RepRap etiquette asks owners of machines to do this at cost at least twice. It also means that a company with a machine can double its production rate simply by having their machine copy itself. This strategy can, of course, be repeated. With self copying, self-repair comes free. When someone acquires a RepRap machine one of the first things that they will be instructed to do is to make one each of all its component parts and to put them on a shelf in a cardboard box in a cool dry place. Then, when a part breaks, it can simply be replaced. Another such part would then be made and put back in the box for next time. Alternatively two machines have the ability to self-repair reciprocally. Many of the efforts to stave off the problems of climate change by reducing (or at least not increasing so fast) atmospheric CO2 involve large-scale political collaboration or the general adoption of technologies that are at best marginally economic. The resulting Tragedy of the Commons is becoming plain for all to see (Gore 2006). On the other hand, the widespread adoption of distributed personal manufacturing using a carbon-neutral plant thermoplastic (such as polylactic acid) might harnesses a boundless resource of virtually unstoppable power – the human greed for material wealth – to the task slowing the increase of atmospheric CO2. And if that greed could be directed to the consumption of a plant-polymer thermoset that had a stable lifetime measured on a geological scale (think amber), then permanent CO2 reduction might be possible. Conclusions An ear of wheat is almost unrecognizable when compared with the seeds of the ancient grasses from which people customized it. It is also orders of magnitude more intricate than any machine ever made by people. And yet it is virtually free. The reason for this is that it can copy itself. Self replication leads to an exponential growth in numbers, and large numbers mean that the replicator becomes very inexpensive. At US$500 the RepRap machine starts off inexpensive (when compared to commercial rapid prototyping machines), and that cost can only go down. In addition, self-replication allows RepRap to go beyond the customization of products: it allows the creator of those products not just to customize them, but also to customize the machine that produces them. And in the end, creator and the machine’s user can be the same person.


579

References Bellwood, P. S. (2005). First Farmers. Blackwell Publishing. Bowyer, A. (2004). Wealth without money. Available at reprap.org/bin/view/Main/. Butler, S. (1872). Erewhon. Available from Project Gutenberg at www.gutenberg.org/ etext/1906. Dyson, F. (2006). Make me a hipporoo. New Scientist, 11 February 2006, (issue 2538). Freitas, R. A. and Merkle, R. C. (2004). Kinematic Self-Replicating Machines, Landes Bioscience, Georgetown, TX. www.MolecularAssembler.com/KSRM.htm. Gore, A. (2006). An Inconvenient Truth: The Planetary Emergency of Global Warming and What We Can Do About It. Rodale Books. Malone, E. and Lipson, H. (2006). Fab@Home: The Personal Desktop Fabricator Kit. Proceedings of the 17th Solid Freeform Fabrication Symposium, Austin TX, USA. Aug 2006. Sells, E. and Bowyer, A. (2006). Directly incorporating electronics into conventional rapid prototypes, Proc. 7th National Conference on Rapid Design, Prototyping & Manufacturing, Eds. Bocking, C.E. et al. Centre for Rapid Design and Manufacture, High Wycombe, UK, June 2006. Sustainable Composites (2008). EcoComp UV-L resin at www.suscomp.com. von Neumann, J. (1966). The Theory of Self-Reproducing Automata. (ed. A. W. Burks), University of Illinois Press.

Author Biographies Ed Sells is a Research Officer at the University of Bath. After completing his first degree in Mechanical Engineering he went on to study self-replicating machines, working on the RepRap project for his PhD. His research focuses on the mechanical development of rapid-prototyping techniques to achieve self-manufacture. Contact: www.reprap.org | [email protected] Sebastien Bailard is a developer with the RepRap Self-Reproducing 3D Printer Group (www.reprap.org). He works on this and other projects in Ottawa, Canada. Contact: www.supermeta.com | [email protected] Zach Smith likes to dream big, fail big, and win big. His true passion in life is acting as a catalyst and helping others do amazing things. Whether it is creating open source micro controllers, robot controller software, object sharing websites, or self replicating 3D printers there is one central purpose: to help other people help themselves create an awesome world to live in. He hopes that someday we can create a world that surpasses even the wildest futures portrayed in science fiction. In the early 1970s Adrian Bowyer read for a first degree in mechanical engineering at Imperial College, London, and then researched a PhD in tribology there. In 1977 he moved to Bath University’s Maths Department to do research in stochastic computational geometry. He then founded the Bath University Microprocessor Unit in 1981 and ran that for four years. After that he took up a lectureship in manufacturing in Bath’s Engineering

580


Faculty, where he is now a senior lecturer. His current area of research is self-replicating machines. He is the inventor and developer of the RepRap replicating rapid prototyper. He also works on geometric computing (he is one of the authors of the Bowyer-Watson algorithm for Voronoi diagrams), the application of computers to manufacturing, the biochemistry of smart materials, and biomimetics. Vik Olliver is an IT systems analyst and replicator developer from New Zealand’s Waitakere rainforest, where he teleworks and runs a workshop. Born in the UK, he emigrated to NZ in 1994 with his wife Suz and daughters Kate & Tamara. He has worked on projects ranging from implantable heart monitors to launching satellites, and from urban self-sufficiency to hydroponics in moonrock. Vik’s employers, Catalyst IT Ltd., are very active in the Open Source community and the majority of project work done there uses Open Source software and tools. He is also a committee member of the New Zealand Open Source Society. If not involved in the above, he is probably on call with the Laingholm Volunteer Fire Brigade. Contact: diamondage.co.nz | [email protected]

5.4

Customization of Consumer Goods: First Steps to Fully Customizable Fashionable Ladies' Shoes Marc van der Zande TNO Science and Industry, Eindhoven, The Netherlands Sjors Bergmans Sjors Bergmans Concept Design, Amsterdam, The Netherlands Nico Kamperman TNO Science and Industry, Eindhoven, The Netherlands Bart van de Vorst TNO Science and Industry, Eindhoven, The Netherlands

The production of fashionable footwear has basically been the same for almost a century. It consists of combining elements of various materials together formed around a physical shape – the last – thereby making a product that fits around the foot – the shoe. This process requires different stages in production, transport of semi-finished products, intermediate components etc. In the European Community funded project "CEC-madeshoe" (www.cec-made-shoe.com) one of the innovation goals has been defined as the development of a product concept and underlying production method with which this traditional manufacturing process is radically transformed, the so-called Direct Manufactured Shoe. Production aspects such as transport, complex logistics, long lead times, and components such as lasts, stiffeners etc. are avoided and the possibility to personalize the shoe is included. This goal was met in a research project in which the production of footwear was approached from the Rapid Manufacturing (RM) point of view. Such a new approach to footwear requires a lot of background knowledge of footwear requirements, production processes, material properties and design opportunities and a lot of creative input from a design point of view.

Rapid Manufacturing Rapid Manufacturing (RM) is a new emerging manufacturing process. It originated from Layered Manufacturing Technologies (LMT’s) that are used in Rapid Prototyping (RP). Originally these techniques were developed to aid and speed up the design process by creating prototypes; for instance for fitting and

581

582


aesthetic purposes. Soon developments in process and materials led to the possibility of creating functional prototypes. This line of development is still being continued. At the moment state-of-the-art technology is being used for RM on a small scale for small series, one-off’s and customized products like air ducts for Boeing fighter jets (Wooten 2006), MGX designer lamp shades (www.materialise.com) and Siemens hearing aid shells (Masters 2006). RM offers a large amount of design freedom in the design of products. The technologies and materials used are still at an early stage of development, but at the moment the state-of-the-art machines and materials already produce goods with characteristics that allow them to be used as final products – with some post-processing if desired. Related to footwear, benefits of RM include:

RM offers a great deal of design freedom for the designer, taking away previous restrictions originating from the manufacturing process: e.g. it gives the designer the ability to quickly alter the fit of a shoe without further production consequences; there is no longer a need for new lasts; and development time and cost are reduced.

Due to the nature of footwear, diversity in sizes is required for a full range of shoes; with RM there is no longer a need for a wide range of moulds and lasts, as digital resizing is possible.

The transport and assembly of semi-finished products is no longer necessary.

The current production of shoes is very labor-intensive; with RM this is no longer the case. This makes it possible for the industry to create added value with limited labor costs and therefore to keep production in the western world.

RM gives the designer the ability to quickly change models; the speed to change is only limited by the ability of the designer to anticipate and react to trends & preferences and the functional ability to modify a design in 3D CAD, therefore RM of shoes makes it possible to respond very quickly to the market and to personalize footwear, aesthetically as well as functionally.

New Conceptual Shoe Designs However, if Rapid Manufacturing is used for new designs in footwear, new product designs are required (Figure 1):

New designs should provide the functionality that is offered in common shoes, like shock absorption, stability, comfort and fitting, but realized with completely new materials and production requirements and possibilities.


583

The new production technology offers totally new construction solutions with which unique new functionality can be provided. Conceptual designer Sjors Bergmans in collaboration with TNO Science and Industry developed innovative but market-acceptable shoe concepts. These concepts give an answer to the question of how to provide existing shoe functionality with new Rapid Manufacturing production techniques and associated materials. In this an attempt was made to come with completely new solutions that take fully advantage of the possibilities that these production techniques offer:

Freedom of shape (no need to limit shapes due to production limitations);

Graded properties (from stiff to flexible and vice versa);

Complex structures can easily be produced. In order to freely think about functionality, brainstorming sessions were carried out as well as free design sessions. This lead to a number of interesting solutions to create the basic shoe functionality. After idea selection, exploration and concept development, this converged to a new lady’s shoe model that was produced with the current RM techniques available at TNO Science and Industry.

Figure 1: Functionality needs in footwear.

Head over Heels: A Concept Shoe The aim was to create a fully 3D shoe with complex structures and containing solutions for: adaptable fitting, an integrated closure system, bending of the forepart relative to the back part, support of the mid foot. Based on these points

584


initial concepts were developed and finally the target product was set as a high heeled lady’s shoe, with special conceptual constructions in heel, forefoot and closure, responding to current fashion styles, and producible in selective laser sintering (SLS) Nylon PA12 materials. A flexible element in the sole allows for high flexibility, and integrated elements in the upper provide an integrated closure based on flexible holding of the foot. From January 2006 onwards collaboration was established between a team of TNO experts (including Rapid Manufacturing specialists, CAD specialists, designers and shoe experts) and conceptual designer Sjors Bergmans, in order to develop the so-called Direct Manufactured Shoes (DMS). Important requirements that were set were:

The shoe should be produced in one step in SLS PA12, a nylon material, without using a last;

The shoe should be functional: people can walk in it comfortably;

The shoe should show the unique opportunities of RM techniques but fit in the current fashion style. Requirements that were considered as desirable (preferable but not required) were:

The shoe can be easily customized;

The shoe is commercially viable. Reasons for not taking these desirable requirements into account are that current CAD-systems, materials and production processes are not yet sufficiently developed to the stage that all these requirements can be met. Developments however in the market and at research organizations like the University of Loughborough, the University of St. Gallen and TNO itself on

production processes, e.g. within the EU 6th framework project Custom Fit (10* faster with a process time of 1-2 hours, and a product price 10* cheaper),

materials (more flexible, stronger, more elastic) and

CAD-systems (parametric), will make it possible that within the coming years all requirements are met. Therefore the emphasis was on conceptual development, using the benefits of the production process and using some features of the unique product concepts that were developed during brainstorming. In order to physically create the model, we choose to hand-prototype the first concept, reverse engineer the model and then alter it digitally. A shoe model was built on a special last from polyester. Typical dimensions of the shoe including material thicknesses were chosen that would give the desired flexibility when


585

produced in the final nylon material to be formed with SLS material. This was a complicated calculation since polyester is more brittle and less flexible compared to laser sintered nylon. The model and the last were then scanned with an optical precision scanner and converted into STL-files. The STL-files created of the shoe (toebox and shoemodel) were joined together digitally in such a way that a hinge was created between the two elements at the joint (ball) (Figure 2).

Figure 2: Digital manipulation of the scanned objects.

Following that, the CAD file for the shoe for the other foot was created by mirroring the model digitally. Next a pre-processing stage took place:

Orienting the model in such a way that the production layers are located in a plane that optimizes the strength of the final model, reducing the chance of breakage (due to the layer wise manufacturing, laser sintered parts show anisotropic behavior. There is the possibility of delamination of layers in the building direction. Therefore parts built in upright position tend to be more brittle and less strong than horizontally built parts);

Scaling the model to compensate for shrinkage during production. Subsequently the pair of shoes was created using selective laser sintering in Nylon PA12 with the Electro Optical Systems EOSINT P380 system, available at TNO. Some post processing was required: removing excess powder and some finishing of rough edges, but basically the final product was then available.

Figure 3: Laser sintering of the shoes

586


Figure 4: The first pair of "Head over Heels".

Improvements on Functionality and Looks At that point, after a first wear test, a number of improvements became apparent, of which changing the heel counter shape (which proved to be too tight) and changing some of the thicknesses (especially in the mid-foot zone) were the most important. Usually when designing footwear the shoe/last dimensions are compensated by the designer for the adaptability/deformability of the used materials (leather or similar materials). With the SLS-materials however this is not the case so existing design standards had to be left behind completely. At some areas the shape of the body has to be followed more exactly (especially when the foot shape is non-deformable like in the heel area) and at other parts flexibility has to be increased to be able to follow the foot when walking, as is the case in the mid- and forefoot area. To improve the digital design we wished to work with a better design (CAD) tool for manipulating the 3D model on the screen, but it became apparent that no good solution exists for developing a full 3D CAD model with double curved surfaces (i.e. a concave partially enclosed shape) which is also parametrically modifiable, so work was done on STL-CAD systems. Features were added: ventilation, anti-slip and some fitting and strength modifications. Some new surface designs were developed using novel macro texturing software (to enhance strength and looks).

Figure 5: Features of the "Head over Heels" design.


587

Experiments were also conducted with different types and colors of coatings; we developed three color/surface options (a rough red surface by coloring the nylon itself, a felt-like black surface by spray-glueing a colored micro fibre on top of the surface and a red shiny coating spray-painted on the surface) to show a variety of "visual effects". Using Selective Laser Sintering technology several pairs of the shoes were realized. The different coating techniques realized different look and surface texture sensations. The friction of the bottom surface of the shoe was improved by adding heel and front outsole pieces made of polyurethane using a vacuum casting technique.

Figure 6: Final CAD model.

Finally, at the end of 2006 the final version of the "Head over Heels" shoe was presented: the first fully Rapid Manufactured functional lady’s fashion shoe. This shoe is protected under the European Design and Mark legislation since 23 June 2006. Figure 7 shows the final result.

Figure 7: Head over Heels, manufactured and post processed.

Currently, this project is in the proof of concept state. Future development will include a wider range of models (including a model for men) and an easily

588


scalable design so that in the end a foot scan can automatically be transferred into a custom design. In order to be successful in the marketplace, manufacturing costs for this process have to go down. Today, a pair of rapid manufactured shoes comes with a heavy price tag of 600-700 Euros. But it is expected that production costs will drop to less than 100 Euro within a few years, given the present speed and scope of application of Rapid Manufacturing technologies in many industries leading to more demand. With this larger scale use of the technology, materials and machine costs will become much cheaper. Conclusions At this moment it is possible to use Rapid Manufacturing techniques to produce functional Direct Manufactured Shoes. The developed concept is the first sample of a fully functional mechanical loaded product with a high design value. In essence, the combination of custom fitting and high design value in a functional, and personal product makes it the next step after the development of technical parts such as the Boeing air ducts, customized medical parts such as the Siemens Hearing aids and furnishings such as MGX designer furniture parts. Some post processing (application of rubber outsoles for grip and applying a coating) is required and long time durability needs to be verified. Also long time comfort properties in terms of ventilation, moisture-absorption, temperature regulation and fit still need to be assessed. The CAD design of the "Head over Heals" shoes at present is limited in adaptability. The adaptation of the CAD model to the client’s foot anatomy and personal preferences should eventually be done automatically. Therefore customization software should be developed to a further extent with which the shoe design can be automatically adjusted to the customer’s foot-scan and requirements. In order to direct manufacture high quality shoes, comparable to high fashion shoes in the market, this rapid manufacturing technology is still quite costly to use. Prices will drop due to increased use of this technology in the manufacturing business overall. This will make this type of product more and more acceptable in the market and will open a whole new perspective and design potential in the market of customized shoes.

References Integrated Project CEC-Made-Shoe, contract 507378 , supported by the EC 6 FP Priority IST – NMP, www.cec-made-shoe.com.


589

Masters, M., Velde, T. and McBagonluri, F. (2006). Rapid Manufacturing in the Hearing Industry. In: Rapid Manufacturing, N. Hopkinson, R.J.M. Hague, P.M. Dickens (eds). John Wiley & Sons, 2006: 195– 209. MGX Design products, www.materialise.com. Wooten, J. (2006). Aeronautical Case Studies using Rapid Manufacture. In: Rapid Manufacturing, N. Hopkinson, R.J.M. Hague, P.M. Dickens (eds). John Wiley & Sons, 2006: 233–239.

Author Biographies Marc van der Zande (1962) is currently a Sports Products R&D program manager at InnoSportNL, the Dutch national innovation program on Sports. He is also cocoordinating a European network of companies and knowledge institutes in the field of Sports and Innovation. Before taking up his current position in 2006 he worked since 1989 at TNO, the Dutch non-profit organization for Applied Scientific Research, in various positions related to Product Ergonomics and Innovation in Sports Products. With his background in Biomechanics and Product Innovation, Marc’s focus is on improving the (physical) interaction between man and product. At national and European level Marc has a profound experience in setting up and managing innovation projects, related to biomechanics and ergonomics and on the application of modern design & production tools in the manufacturing of ergonomic products. Contact: www.innosport.nl | [email protected] Sjors Bergmans is an industrial designer. He works in Amsterdam, The Netherlands. Contact: [email protected] Nico Kamperman obtained an MSc degree in Mechanical Engineering at the University of Twente, The Netherlands in 1996. After 5 years of multidisciplinary experience in the industry as senior engineer he joint the Rapid Manufacturing department at TNO Science and Industry. He has extensive experience in the application of layered manufacturing as production technology. He was scientific coordinator and member of the management board of several EU projects related to customized products. Nico Kamperman is currently Manager Metals at the Central Laboratory of Daf Trucks in the Netherlands. Contact: [email protected] After the completion of his study Industrial Product Design in The Hague, Bart van de Vorst BSc. started working at TNO Science & Industry. As a member of the Rapid Manufacturing (RM) department he is involved in several national and international research projects based upon layered wise production with topics like customization, medical applications and 3D Engineering. Besides knowledge and hand-on experience of several commercial Rapid Manufacturing processes he is also involved in the development of new and innovative techniques based upon 3D printing. Working over eight years in the field of Rapid Prototyping and Manufacturing he is well experienced in the design and construction of (end) products made with these digital production techniques. Contact: www.tno.nl/rm | [email protected]


Vol. 2 Applications and Cases

7378.V2.tp.indd 1

10/27/09 5:31:32 PM


Vol. 2 Applications and Cases

HANDBOOK OF RESEARCH IN MASS CUSTOMIZATION AND PERSONALIZATION edited by

Frank T Piller RWTH Aachen University, Germany

Mitchell M Tseng The Hong Kong University of Science & Technology, Hong Kong

World Scientific NEW JERSEY

7378.V2.tp.indd 2

•

LONDON

•

SINGAPORE

•

BEIJING

•

SHANGHAI

•

HONG KONG

•

TA I P E I

•

CHENNAI

10/27/09 5:31:34 PM

Published by World Scientific Publishing Co. Pte. Ltd. 5 Toh Tuck Link, Singapore 596224 USA office: 27 Warren Street, Suite 401-402, Hackensack, NJ 07601 UK office: 57 Shelton Street, Covent Garden, London WC2H 9HE

British Library Cataloguing-in-Publication Data A catalogue record for this book is available from the British Library.

HANDBOOK OF RESEARCH IN MASS CUSTOMIZATION AND PERSONALIZATION (In 2 Volumes) Volume 2: Applications and Cases Copyright © 2010 by World Scientific Publishing Co. Pte. Ltd. All rights reserved. This book, or parts thereof, may not be reproduced in any form or by any means, electronic or mechanical, including photocopying, recording or any information storage and retrieval system now known or to be invented, without written permission from the Publisher.

For photocopying of material in this volume, please pay a copying fee through the Copyright Clearance Center, Inc., 222 Rosewood Drive, Danvers, MA 01923, USA. In this case permission to photocopy is not required from the publisher.

ISBN-13 978-981-4280-25-9 (Set) ISBN-10 981-4280-25-9 (Set) ISBN-13 978-981-4280-26-6 (Vol. 1) ISBN-10 981-4280-26-7 (Vol. 1) ISBN-13 978-981-4280-27-3 (Vol. 2) ISBN-10 981-4280-27-5 (Vol. 2)

Printed in Singapore.

Shalini - Hdbk of Res in Mass Customization.pmd 2

11/5/2009, 4:14 PM

Volume 2

Applications and Cases

v


Foreword & Acknowledgments This book is the third in a series of publications that present the latest advancements in research on mass customization and personalization. Starting with Tseng & Piller (2003) and continuing with Piller, Reichwald & Tseng (2006), we again could collect the thinking of some of the leading scholars and practitioners in the field. In comparison to the previous editions, this is the most comprehensive collection of writings on mass customization ever. This inspired our publisher to name it the "Handbook of Research in Mass Customization & Personalization". The contributions in this handbook were inspired by the 4th World Conference on Mass Customization and Personalization (MCPC 2007), a biannual academic event that gathers the international research and practice community interested in mass customization, held in October 2007 at the Massachusetts Institute of Technology (MIT), hosted by the MIT Smart Customization Group (Mitchell et al. 2007). The conference also included a business seminar held at HEC Business School in Montreal, Canada. The participant roster of the conference represented the interdisciplinary nature of customization and personalization drawing from a wide range of interest from hard core engineering, fashion design, architecture, retail, business strategy to psychology. The papers in this book reflect this richness and scope. Our authors come from diverse schools in leading researech institutions as well as from business practice or consulting firms. Such a voluminous work is not possible without the help from many individuals. At MIT, we sincerely want to acknowledge the support and help by Prof. William Mitchell, Ryan Chin and Betty Lou McClanahan from the MIT Design Lab and the MIT Smart Customization Group. As co-chairs and organizers, they were providing leadership for the MCPC 2007 at MIT, hence paving the way that the research presented in this book could be assembled in the first place. From more than 200 conference contributions, an editorial committee selected the papers included in this handbook. All papers were subject of an additional review process. Along with the feedback authors received in the conference, the manuscripts were modified and edited to the collection of papers presented here. There are too many reviewers to name them here individually, but we want to thank them all for their great service to our community. At RWTH Aachen, Frank Steiner coordinated the editorial and publishing process of this large project as the executive editor and provided valuable assistance to us. Dealing with more than 100 authors and coordinating more than 50 papers is a very demanding and time-consuming task. In addition, we thank our

vii

viii


publisher for their patience and continuous support for this project. It was a real pleasure working with World Scientific Co. on this book. Our final thank, however, deserve the authors and contributors to this handbook. Only due to their willingness to contribute their latest research and thinking, this project has been possible. We thank them for their patience and compliance in addressing all the numerous demands and requests that such a book project demands. We believe that the research presented here provides a comprehensive and rich introduction into the various aspects that make mass customization one of the most promising business strategies for this century. Frank T Piller & Mitchell M Tseng

Contents Volume 1 Foreword and Acknowledgments................................................................................... vii Introduction: Mass Customization Thinking: Moving from Pilot Stage to an Established Business Strategy ...................................... 1 1

Strategic Aspects of Managing Mass Customization & Personalization............ 19 1.1

From Mass Production to Mass Customization: Hindrance Factors, Structural Inertia and Transition Hazard

21

How to Implement a Mass Customization Strategy: Guidelines for Manufacturing Companies

44

Media Market Inertia: A Potential Threat to Success of Mass Customization

65

Operationalizing Mass Customization – A Conceptual Model Based on Recent Studies in Furniture Manufacturing

79

1.5

Beyond Mass Customization: Exploring Features of a New Paradigm

97

1.6

Is the Best Product a Unique Product? Exploring Alternatives to Mass Customization with the Online Community of Threadless

118

Before Pine and Dell: Mass Customization in Urban Design, Architecture, Linguistics, and Food

139

1.2 1.3 1.4

1.7

2

Consumer Studies & Marketing Aspects ............................................................ 159 2.1 2.2 2.3 2.4 2.5 2.6

Typology of Potential Benefits of Mass Customization Offerings for Customers: An Exploratory Study of the Customer Perspective

161


181

Why Consumers Are Willing to Pay for Mass Customized Products: Dissociating Product and Experiential Value

208

Sneakerheads and Custom Kicks: Insights into Symbolic Mass Customization

226

E-Customization: Research and Applications from the Cognitive Learning Theory

240

Modularity as a Base for Efficient Life Event Cycle Management

263

ix

x


3

4

Bundling, Mass Customization, and Competition under Consumption Uncertainty

275

Building the Solution Space: Product & Process Design for Mass Customization........................................................................................ 295 3.1

Towards a Knowledge Support System for Product Family Design

297

3.2

Product Family Modeling: Working With Multiple Abstraction Levels

319

3.3

Market-based Strategic Platform Design for a Product Family Using a Bayesian Game

338

3.4


357

3.5

Change Prediction for Mass Customized Products: A Product Model View

376

Making Manufacturing & Supply Chain Management for MCP Work.......... 401 4.1

An Agility Reference Model for the Manufacturing Enterprise: The Example of the Furniture Industry

403

Overcoming Configuration Process Complexity of Highly Customizable Components

427

4.3

Mass Customization of Responsive Automated Assembly Cells

451

4.4

A Prioritization Algorithm for Configuration Scheduling in a Mass Customization Environment

487

4.5

Procurement Mechanisms for Customized Products

513

5

Rapid Manufacturing for Mass Customization

535

5.1

Extreme Customization: Rapid Manufacturing Products that Enhance the Consumer

537

e-Manufacturing – Making Extreme Mass Customization Real by Laser-Sintering

555

RepRap: The Replicating Rapid Prototyper: Maximizing Customizability by Breeding the Means of Production

568

Customization of Consumer Goods: First Steps to Fully Customizable Fashionable Ladies' Shoes

581

4.2

5.2 5.3 5.4

xi

VOLUME 2: APPLICATIONS AND CASES

Volume 2 Foreword and Acknowledgments................................................................................... vii Introduction and Overview........................................................................................... 591 1

Customization & Personalization of Services .................................................... 601 1.1

How to Master the Challenges of Service Mass Customization – A Persona-Based Approach

603

Mass Customization in Wireless Communication Services: Individual Services and Tariffs

622

Unraveling the Service Innovation Dilemma: The Promise of Network Embeddedness

646

1.4


665

1.5

One Size Fits All, Made-to-Measure, and Bespoke Tailoring: Challenges in Building an Attractive Service Portfolio

683

Mass Customization for Individualized Life-long Learning: Needs, Design, and Implementation

698

A Mass of Customizers: The WordPress Software Ecosystem

717

1.2 1.3

1.6 1.7 2

Beyond Bespoke Tailoring: Mass Customization in the Apparel Industry...... 729 2.1 2.2 2.3 2.4 2.5 2.6

3

Virtual Fit of Apparel on the Internet: Current Technology and Future Needs

731

RFID Diffusion in Apparel Retail: How Consumer Interest and Knowledge Lead to Acceptance

749


771

Developing Considerate Design: Meeting Individual Fashion and Clothing Needs Within a Framework of Sustainability

813

Customized Garment Creation with Computer-Aided Design Technology

833

A Case Study in Personalized Digitally Printed Clothing

852

Mass Customization in Architecture and Construction .................................... 867 3.1


869

xii

4


Mass Custom Design for Sustainable Housing Development

892

3.3


911

Applications of MCP in Various Contexts .......................................................... 941 4.1 4.2 4.3 4.4

4.5 5

The State of the Art of Mass Customization Practices in Finnish Technology Industries: Results from a Multiple-Case Study

943

Opportunities and Challenges of Furniture Manufacturers Implementing Mass Customization

965

Mass Customization in the Ophthalmic Lens Industry: Progressive Addition Lenses for Your Visual Map

977


996

Simulation Models to Demonstrate Mass Customization Strategies

1005

From Mass Customization to Open Innovation .............................................. 1021 5.1 5.2 5.3 5.4

User Innovation and European Manufacturing Industries: Scenarios, Roadmaps and Policy Recommendations

1023

Bridging the Innovation Gap: From Leading-Edge Users to Mass Market

1044

Ordinary Users and Creativity: Fostering Radical or Incremental Innovation?

1059

Modeling and Evaluating Open Innovation as Communicative Influence

1081

Introduction and Overview Frank Piller RWTH Aachen University, Germany Mitchell Tseng Hong Kong University of Science & Technology, Hong Kong

Continuing the introduction and overview provided in Volume 1 of this handbook, this chapter provides a comprehensive overview of the research presented in this volume of the handbook.

Structure of the Book Volume 1 of this handbook was structured along the value chain. After a broader discussion of the mass customization concept and its implementation in industry, the chapters of Volume 1 followed the value chain by looking on mass customization from the perspective of marketing, product creation and design, and manufacturing. The latter aspect was extended by a special focus on rapid manufacturing (3D printing), a technology that can be seen as a key enabler of new solutions for mass customization manufacturing. Volume 2 of the handbook provides a more focused view on applications of mass customization & personalization in diverse industry settings. It contains a number of extensive case studies of specific mass customization implementations. These case studies discuss the findings presented in Volume 1 in an integrated way and discuss how the three bundles of capabilities of a sustainable mass customization system have been applied in different companies. But beyond just demonstrating "best practices" and learning from case studies, the papers in this part of the handbook also provide new conceptual, methodological and theoretical contributions with a distinctive industry focus. While the fashion industry (Chapter 2) has been the focus of research in mass customization since a long period of time, the construction industry (Chapter 3) has only recently become an object of study. The same holds true for the vast area

591

592


of services and intangible products (Chapter 1). This chapter provides insights into specific challenges and methods required to customize services efficiently for a larger customer segment. But also furniture, optical lenses, industrial machinery, or high-end home entertainment are fields of mass customization application discussed in this part (Chapter 4). The final chapter of this book then bridges the topic of mass customization & personalization with a closely related topic, open innovation and customer co-creation in the new product development process (Chapter 5). Chapter 1: Customization & personalization of services Mass customization originated in the manufacturing domain. This origin still dominates the research landscape. But mass customization thinking also is increasingly relevant for providers of intangible goods such as software and services. There is, however, relatively little work on mass customization in the service context until today, despite the dominant role of this sector in the modern society. The papers in this chapter want to close this gap. They include case studies of specific company examples of successful service customization and conceptual or empirical papers addressing particular challenges of customizing a service offering. These challenges effect all three core capabilities of mass customization (Salvador et al. 2009): the requirements towards the design of the service architecture (solution space design), the design of the service production and delivery processes (robust value chain design), and the design of the interaction system with the customer, including an appropriate education of service employees (choice navigation). One key challenge for service customization is to translate information about consumer preferences gained through market research into a format that can be easily used for service modularization decisions and customer-contact personnel training. Michael Haas and Werner H. Kunz address this aspect of solution space design for service mass customization in Section 1.1. The authors propose narrative descriptions of archetype service customers (called 'personas') as a tool for transferring complex market research data into a solution design for service customization. In Section 1.2, Hong Chen and L-F Pau provide a detailed insight into a core application of service customization, the configuration of individual services and tariffs for wireless communication services. Their chapter advocates a user-centric view of wireless service configuration and pricing as opposed to present-day service catalogue options. The focus of the paper is on the design methodology and tools for such individual services and tariffs, using information compression,


593

negotiation algorithms, and risk port-folio analysis. The implications of this approach are discussed based on an end-user survey and model-based calculation. The authors show that users can achieve desired service bundle cost reduction while suppliers can improve significantly their risk-profit equilibrium points, reduce churn as well as marketing costs, and simplify provisioning. Introducing service mass customization always is a major innovation from the perspective of suppliers and customers alike. In Section 1.3, Ikenna S. Uzuegbunam, Satish Nambisan, and Manli Chen, discuss the specifics of the service innovation process. They identify network approaches through which firms engage the "service innovation dilemma"— the problem of diseconomies of scale in a world of increasing demand for services. The authors argue that firms can develop sustainable competitive advantage in services through "real" and "virtual" embedded inter-firm and customer co-creation mechanisms. Section 1.4 provides an in-depth insight in such a service innovation process. Here, X. Hernández and his co-authors from the Universidad Politécnica de Valencia describe the use of emotional design techniques for service personalization. They show how these techniques can improve the design of the point of sale of habitat-related products. The authors share findings from a user experiment and show to which extent a retail store's background has influence on the willingness to purchase a product. The remaining three sections of this chapter describe three specific applications of service customization. In Section 1.5, Hans Björkman discusses the special case of customization at Unionen, a major white-collar trade union in Sweden. He builds on the dilemma that only "bespoke" services can deliver ultimate customer value, but that these services often come at a cost. The section demonstrates how Unionen managed to provide a broad and attractive service portfolio consisting of standardized, mass customized, and individually customized services. Each individual member creates an individual service portfolio through personal choice of information, activities, and services. Linkages between standardized, mass customized, and individually customized services are well defined. Section 1.6 describes a field of service customization that has received a lot of attention recently, mass customization of education. Hermann Klinger and Alexander Benz propose to replace the paradigm of traditional education with its idea of an "economy of scarcity" by an "economy of self-generation". Mass customization provides the conceptual and operational framework for analyzing participants' needs. Andrew Watson finally discusses the WordPress software ecosystem, which is both software and service (Section 1.7). The chapter presents WordPress as a blogging software that is widely distributed and deeply customizable, as a family of products built on a common platform, and as the focus of a vibrant community and a thriving ecosystem. Analyzing these layers, the author provides

594


implications for service customization at WordPress and for mass-customized software more generally. Chapter 2: Beyond bespoke tailoring: Mass customization in the apparel industry While customization in the service sector still is in an emerging stage, mass customization in the apparel industry today can be regarded as an established field. Plenty of research and a wide array of business applications exist in this domain. Accordingly, the papers in this chapter address advanced topics. They often build on the combination of services and tangible product features, providing an integrated experience from the customer's perspective. The first three papers in this chapter (Sections 2.1, 2.2, and 2.3) focus on personalization in fashion retail. Rather then discussing aspects of the customized production of garments for an individual consumer, their perspective is on personalizing fashion retail. A fundamental approach here is technology to match an existing assortment of garments to an individual profile of one consumer, helping customers to better navigate the existing choice. The remaining papers in this chapter (Sections 2.4, 2.5, and 2.6) discuss advancements of the customized manufacture of clothing, i.e. of creating a new assortment for each consumer. The focus is on new manufacturing technologies like digital printing or 3D knitting which translate the promises of rapid manufacturing ("3D printing") for the fashion industry. Section 2.1 is a good example for this new stream of research. Susan Ashdown, Emily Calhoun and Lindsay Lyman-Clarke address personalization in fashion retail over the internet, comparing a match-to-order system, where a standard good is matched to the personal profile of a consumer, with a make-to-order system where the product is produced on demand based on the personal consumer profile. They compare different technologies to support this matching process. Overall, the paper reveals how personalized online recommendation systems can improve consumer confidence in purchasing, and ultimately boosts sales. While Asdown et al. focus on personalizing the online experience, Sanchit Tiwarie and Suzanne Loker cover personalization in an offline retail setting. Here, RFID technology is regarded as a core technique to enable a personalized retail experience in stores. Previously, this research was dominated by technical aspects of making RFID work. In this chapter, however, the crucial question is answered whether consumers will accept these technologies. The authors present results from an empirical study about consumer perceptions of body scanning and RFID technologies and applications.


595

"Discard 'one size fits all' Labels," demand Marie-Eve Faust and Serge Carrier in Section 2.3. While this sounds like a natural claim in a book on mass customization and personalization, the authors provide detailed evidence that this easy statement is not heard in today's fashion industry. They propose a new size labeling system to support order givers, manufacturers, retailers, and consumers. Such a system may become a cornerstone of personalization in fashion retail, providing consumers with fundamental information for matching an existing garment to an individual preference profile. With Section 2.4, the perspective turns back to more conventional mass customization applications. Firsr, Jing-Jing Fang and Chia-Hsin Tien propose an application of computer-aided techniques to customize garment creation. Different from traditional CAD tools on planar pattern design, the authors show how a 3Dbody scan of an individual consumer can be transferred into basic body shapes. The authors also propose an innovative tailoring method to generate a flare sagging style. In Section 2.5, Sandy Black, Claudia M Eckert and Philip Delamore address "The Fashion Paradox" – the economic importance of the fashion industry set against its inherent obsolescence and waste through constant change. They present a new methodology for designers to approach these complex problems and to evaluate the impact of design decisions through the development of personalized fashion products. The approach positions the user at the centre of the design process by applying rapid manufacturing technologies (see Chapter 5 in Part I of this handbook) in the textile industry. Section 2.6 discusses a further case of digital technologies in manufacturing. Philip Delamore and Jennifer Bougourd present a detailed look onto the digitization of the clothing product development process incorporating 3D body scanning, automatic pattern generation, visualization, digital printing, and embroidery. The aim of this project was to introduce custom printing and embroidery into an existing line of mass customized fashion, provided by Bodymetrics Digital Couture of London at Selfridge's and Harrods, two leading UK department stores. Chapter 3: Mass customization in architecture and construction Only very recently, the large field of architecture and construction has seen a systematic discussion of mass customization. As Amir E. Piroozfar and Olga Popovic Larsen observe in Section 3.1, mass customization has been used inadvertently in the building industry since a long time. But only very little and scattered systematic attempts have been made to apply it within the field knowingly. Almost none of these attempts have successfully avoided the

596


predicament imposed by the dominance of either (craft) manufacture interpretation or its predecessor, mass-production. In this section, the authors provide a review of the literature and current attempts to mass customization in the building industry. They investigate a series of projects where mass customization thinking has been successfully applied to the design and fabrication of building envelopes. One reason that mass customization has received more attention in the field of architecture is the recent focus on sustainability and eco-efficiency in housing. Accordingly, Sections 3.2 and 3.3 address this issue. Masa Noguchi and Karim Hadjri discuss how mass customized design can lead to more sustainable housing development. Homes today need to be socially, economically, and environmentally sustainable in response to the wants and needs of individual homebuyers as well as the society. However, existing approaches to housing design in the residential market barely lead to any accomplishment of the sustainability agenda. In their paper, the authors discuss how mass customization thinking can support sustainability in the housing market. This also includes the idea to masspersonalize a home after its first occupancy to meet diverse market demands over the lifetime of a building. In Section 3.3, Amir E. Piroozfar, Olga Popovic Larsen and Hasim Altan extend this discussion. Their paper proposes "modern methods of construction" (MMC), a set of approaches and methods in the building industry in which the notion of customization can potentially be embedded. Findings of a comparative study of two building projects are used to demonstrate how the benefits of a mass-customization based modern construction method may increase sustainability in the built environment. Chapter 4: Applications of MCP in various contexts Chapter 4 is devoted to applications of mass customization in diverse contexts. The papers assembled here provide an insight into the scope and scale of mass customization thinking in different industries. Section 4.1 starts this review with an in-depth view into the state of mass customization practices in Finland. Marko Mäkipää and his co-authors present the results of a multiple-case study conducted in 37 industrial companies. They show that mass customization practices are widely used in Finnish technology industries. Product qualities and production processes are managed well, but still numerous challenges remain, especially with regard to cross-functional cooperation, the deployment of configuration systems, and the integration of different information systems. In Section 4.2, Torsten Lihra, Urs Buehlmann and Robert Beauregard discuss opportunities and challenges of North American furniture companies implement-


597

ing mass customization. Their study sheds light on manufacturers' perception of mass customization as a strategy to react on the growing competition from Asian manufacturers. A survey of furniture manufacturers in Canada, the USA, and Germany showed that developing modularity and agility, integrating the supply chain and pursuing a competitive cost structure are critical elements of a furniture customization system. But while these elements were mastered well by many respondents, limited capabilities of end users when designing their individual product were regarded as the true limit of mass customization in this industry. Section 4.3 changes from a macro to micro perspective and provides an in-depth study of one particular company and its mass customization strategy. Begoña Mateo, Rosa Porcar-Seder and their team of co-authors show how INDO, a Spanish provider of ophthalmic lenses, designed and implemented a new generation of customization in the lenses industry. Traditionally, users have been asked to adapt to progressive lenses that are designed to fit an average wearer. INDO's proposal is that a customized progressive lens that mimics the natural vision can be obtained by measuring the visual strategy of each individual user, defined as the coordination of eyes and head movements. The result represents a major scientific advance and has positioned INDO at the head of the progressive lens' field. An even more specialized application of mass customization is presented in a paper by Richard So, John Au and K.L. Leung (Section 4.4). They discuss how personalization technology can resemble the experience of listening to music played on a theater stage in a consumer's home. Using personalized filtering technology, it is possible to simulate the acoustics effects of a concert-theater for music presented on a pair of earphones. However, such a personalized solution can cost over $2000 and may not be feasible for consumer products. Nonpersonalized solutions, on the other hands, do not work well. This paper discusses the challenges and opportunities of mass-customized solutions in this industry. In Section 4.5, Fazleena Bardurdeen, Haritha Metta and Brandon Stump present an approach to teach mass customization and to educate students about the elements of this strategy. The authors have developed a simulation model to incorporate practical demonstrations of mass customization in order to engage students in active/experiential learning. Their paper presents a simple but versatile simulation that can be used in classroom environments to help participants to understand the concept of mass customization and challenges to implementing the strategy.

598


Chapter 5: From mass customization to open innovation Chapter 5 extends the perspective presented in the previous chapters by connecting mass customization with the concept of open innovation. The term characterizes a system where innovation is not solely performed internally within a firm, but in a cooperative mode with other external actors (Fredberg et al. 2008; Reichwald and Piller 2009). Open innovation is opposed to closed innovation, in which companies use only ideas generated within their boundaries, characterized by big corporate research labs and closely managed networks of vertically integrated partners (Chesbrough 2003). Sources of external information for the innovation process are plentiful, including customers, suppliers, competitors, university labs, and research institutions. In the papers in this chapter, the focus is on the users' or customers' contribution to an innovation project. Integrating the customer into the firm's value chain is a dominating perspective in both mass customization and open innovation. In mass customization, customers are being integrated to utilize an existing solution space, i.e. to configure a product or service by selecting options from this existing assortment of choices. In open innovation, on the contrary, customers are being integrated to create a new solution space or modify an existing one, i.e. to co-create a general product or service offering that then is offered to a larger customer base. While the context of customer integration is different, customer integration in mass customization and in open innovation share a number of common characteristics, as the papers in this chapter reveal. In Section 5.1, Philine Warnke, Karl-Heinz Leitner, François Jégou and Wolfram Rhomberg provide a good macro-level overview on the state of customer innovation in Europe and its relevance for the European manufacturing industry. In times of increasing relocation of manufacturing to low wage production sites, production strategies that place a large part of the value chain close to the customer are becoming increasingly attractive to keep jobs within the country. As a result, many governments recently have launched initiatives to explore how to benefit from user innovation and to support companies in their adoption. However, the authors show that to achieve this goal, tailored and efficient actions are required that can align research and innovation policy with measures from other realms such as IPR and regulation. In Section 5.2, Philippe Duverger and Salah Hassan provide a micro-level analysis of contributing customers. They start with their analysis with leadingedge users in a market that are ahead of the general trend and have a high motivation (and ability) to solve an existing problem on their own. But lead users only correspond to a very small segment of customers in a market segment. Thus,


599

the authors look further how firms can bridge this "innovation gap" between leading-edge users and the mass market. Building on the lead user and customer co-creation literature, their paper demonstrates the strategic pertinence of involving a firm's users and defectors in generating new innovative ideas. Peter R. Magnusson, Per Kristensson and Christiane Hipp extend this perspective in Section 5.3. They look whether the contributions of ordinary users to a firm's innovation process foster radical or incremental innovation. Their paper provides a better understanding of how users contribute in ideation, using the example of mobile telephony services. They conduct a quasi-experimental study to evaluate the innovation potential of ordinary users. Based on the results of this study, the authors derive a number of managerial implications how to obtain ideas from ordinary customers. The final section of this chapter provides an integrated framework and methodology to perform open innovation (Section 5.4). Building on a large literature background, Jouni Similä, Mikko Järvilehto, Kari Leppälä, Harri Haapasalo and Pasi Kuvaja propose a facilitated innovation process model which links a company's internal and external innovation process, including also the role of an external intermediary brokering the interaction between the company and external actors. Empirical evidence from three cases of SMEs in an early idea generation phase is used to evaluate the proposed method. References Chesbrough, H. (2003). Open innovation: the new imperative for creating and profiting from technology. Boston: Harvard Business School Press. Fredberg, T., Elmquist, M. and Ollila, S. (2008). Managing Open Innovation: Present Findings and Future Directions. Working paper. Chalmers University of Technology, Gothenburg. Reichwald, R. and Piller, F. (2009). Interaktive Wertschöpfung [Interactive Value Creation]. 2nd edition. Wiesbaden: Gabler. von Hippel, E. (1998). Economics of product development by users: The impact of "sticky" local information. Management Science. 44(5): 629–644.

Author Biographies Prof. Dr. Frank Piller leads the Technology & Innovation Management Group at RWTH Aachen University. He also is a co-founder of the MIT Smart Customization Group at the Massachusetts Institute of Technology, USA. Before entering his recent position in Aachen in spring 2007, he worked at the MIT Sloan School of Management and has been an associate professor of management at TUM Business School, Technische Universitaet Muenchen (1999-2004). His research focuses on mass customization, open/user

600


innovation, and methods to increase the efficiency and effectiveness of the innovation process. As a founding partner of Think Consult, a management consultancy, he helps his clients to serve their customers better by using truly customer-centric strategies. Contact: mass-customization.blogs.com | [email protected] Prof. Mitchell M. Tseng, Ph.D. is Chair Professor and Director, Advanced Manufacturing Institute, Hong Kong University of Science and Technology. He also is an Adjunct Professor of MIT-Zaragoza Logistic Program. Prof. Tseng started his industrial engineering career in developing key enabling manufacturing technologies for IT industry. Some of them, such as configuration systems for computers, diamond machining for polygons in laser printers, are still widely used in industry. After serving in industry for two decades, he joined HKUST in 1993 as the founding department head of Industrial Engineering. He is an elected fellow of the International Academy of Production Engineers (CIRP), and American Society of Mechanical Engineers (ASME). Professor Tseng is internationally known for his research in Mass Customization and Global Manufacturing. Sponsors of his research include AT & T, Astec-Emerson, Esquel, Honeywell, Lucent Technologies, Intel, SAP, Rockwell International, Liz Claiborne, Motorola, Nokia, GAP, Ford Motor, Norvullus, Tecton, Synocus, Yuesan, OOCL, Novellus, Ove ARUP, HK Air Cargo Container Limited, and various government agencies in Hong Kong, Mainland China and the EU. Contact: ami.ust.hk | [email protected]

1

Customization & Personalization of Services

601


1.1

How to Master the Challenges of Service Mass Customization – A Persona-Based Approach Michael Haas A.T. Kearney GmbH, Berlin, Germany Werner H. Kunz College of Management, University of Massachusetts Boston

Whereas efficient and effective practical implementation of mass customization in goods industries is a well-researched topic, a deeper understanding of service mass customization is still missing. Particularly service specific challenges have been hitherto unaddressed. These characteristics affect the requirements towards the service design as well as the appropriate education of service employees. One key challenge for service mass customization is to translate information about consumer preferences gained through market research into a format that can be easily used for service modularization decisions and customer-contact personnel training. As solution to this "translation problem" we propose "personas" as a tool for transferring complex market research data into narrative descriptions of archetype customers. Such archetypes convey customization needs in a way that makes it easy for decision makers and employees to understand and to communicate effectively with each other. Along with the concept we introduce a structured approach for developing and using personas in service mass customization projects and provide managerial implications as well as an outlook for further research.

Introduction Many marketing scholars have portrayed the application of mass customization strategies to services as a viable approach to reconcile the tension between cost reduction targets and the need for providing a highly personalized experience (e.g. Gilmore and Pine 2001; Piller 2001; Pine 1993). Hence, mass customization enables companies to offer new solutions to their customers and differentiate itself successfully against the competition. Extraordinary successful business success stories based on the effectively use of the new technological possibilities can be observed over the last ten years (e.g. Dell, Audi). The reason for this is based on experiences from goods industries where it has been amply demonstrated that by use of product modularization and customer orientated product configurators it is possible to generate high customer benefits while maintaining a high efficient production process (Franke and Piller 2003; Piller and Müller 2004). 603

604


Whereas efficient and effective practical implementation of mass customization in goods industries is a well-researched topic, a deeper understanding of the economics of service mass customization is still missing. Knowledge is particularly scarce with regard to service specific implementation challenges arising from the customer integration into the service delivery process and the intangibility of services (Zeithaml and Bitner 2000). A better understanding of these issues becomes especially important with regard to services that are delivered by a customer contact employee and not via a human-to-machine interface (e.g. web sites). Here mass customization becomes an ongoing configuration process with direct involvement of the customer and professional advice by the service personal. Moreover, service personnel needs to understand configuration rules and know them by heart because the possibility to pre-define the configuration rules and to completely embedded them into machines is limited. The main objective of this article is to present a concept that facilitates the development and implementation of service mass customization processes. For this purpose we propose the use of "personas", a concept that has its origins in software design but is nowadays also widely applied by web designers and marketing agencies (Chon 2006; Cooper 1999). Personas have been shown as a tool well suited for transferring complex market research data into narrative descriptions of archetype customers. Such archetypes convey customization needs in a way that makes it easy for decision makers and employees to understand and to communicate effectively with each other. Hence we perceive personas as adequate mechanism to overcome the specific challenges in developing and delivering mass customized services. Based on a literature review and analysis of existing case studies of persona applications we conceptualize a theoretical framework for the persona development process, which can be used by firms to support their service mass customization process. In the following we provide at first a brief overview on the specific challenges of mass customizing services. The next section introduces the persona concept and demonstrates its value as solution to the previously highlighted implementation problems. Following the introduction to the persona concept we provide a framework that supports the development of personas and their application to mass customization projects. Finally we provide an outlook on further empirical research and managerial implications of persona-based service mass customization. Specific Challenges to Service Mass Customization One major difference between the provision of goods and services is the fact that services are delivered along a process in which the customer is directly involved.

605


At first glance mass customization seems to blur this difference. After all, mass customization actively involves the customer as it allows them to influence how the end product looks like. Yet a closer look at mass customization processes reveals that the difference between goods and service is still of relevance. Firstly, during mass customization of goods the production process is adapted based on configuration choices or rules that have been fixed prior to the start of production. The customer can influence the product only along a predefined choice menu. The content of the choice menu is, in turn, constrained by the limits of the production technology and cannot be influenced by the customer at all. In most cases the choice menu is hard-coded into a machine-to-human interface and thus the customer has not the opportunity to question the available choices. Secondly, once the production process for a mass customized good has started it is executed without any further customer involvement. Thus, the customization cannot be flexible adapted to the specific needs of the customer during the production process. The general differences between mass customization of goods and services are summarized in Figure 1 and elaborated in the following paragraphs.

Mass Customization of Goods

Service Mass Customization

Configuration primarily based on human-tomachine interaction Configuration rules and choice menu are hard-coded into the “machine”

Customer integrated into the service delivery process Heterogeneity of the service outcome Often to be delivered personally

Configuration settings cannot be flexibly adapted

vs.

Configuration is an ongoing process with direct involvement of the customer Prior fixed configuration settings are limited

Figure 1: Differences between mass customization of goods and services.

Customer integration in the production process In contrast, in the case of services customers are not simply consuming the outcome of the production process but are an integral part of it. This direct integration of the customer in the service delivery process implies both an opportunity and a challenge for the mass customization of services (Kaplan and Haenlein 2006). On the one hand the company has a continuous contact with the customer, which enables it to respond to the customer’s reaction actively and can be used for the personalization of the service delivery. On the other hand, the

606


integration of the customer in the service delivery process implies an inherent heterogeneity of the process’s outcome, which makes it difficult to maintain standardized service modules. Firstly, as customers directly interact with service personnel already in the beginning of the delivery process it might become necessary to modify the available choice menu as customers re-negotiate the details of the pre-defined service options. Secondly, being involved in the production process customers will also have the opportunity to re-configure the services although its production has already started. This opportunity to exert influence strongly limits the possibility for a customization based on prior fixed configuration settings (Wirtz and Bateson 1999), particularly when the service delivery process is of critical value for customers (e.g. restaurant, hospital). The necessity/opportunity to customize the service while it is produced has a significant consequence: service customization becomes an ongoing process and it results out of the continuous interaction between frontline contact employee and the customer, where the professional advice by the service personal is needed. Hence, as Gwinner et al. (2005) already stressed, the ability and the willingness of the frontline contact personnel to apply customization strategies "on-the-fly" is a key success factor for mass customization of services. Gwinner et al.’s (2005) study further shows that a clear understanding of customer needs is a key prerequisite for contact employees in order to successfully adapt a service offering to the customer’s customization needs. A clear understanding of customer needs is particularly required as during customization, according to the literature on mass customization of products, the customer must be provided with a set of configuration options that strike a perfect balance between flexibility and complexity (e.g. Piller 2001). Moreover, depending on the customer’s experience with configuration options the "choice menu" may vary in its degree of complexity (Dellaert and Stremersch 2005). This implies that a frontline customer contact employee must not only be able to understand the customer’s principal customization needs but must also be able to assess to what extent the menu of configuration choices may get more complex or not. To put it differently, the service staff has to adapt the choice menu according to the customer’s reaction and balance between too few and too many options flexibly. However, in order to avoid exploding service costs resulting from fulfilling every wish a customer has, service personnel should also not be too responsive towards customers for further personalization. Conversely, responding to a customer’s request for customization should be based on a thorough assessment of the costs and benefits of fulfilling the customer’s request. Service personnel should


607

consider what aspect of the service customization is valued most by the customer and what would it cost to satisfy the customer. All customization requests that are too costly to fulfill given the benefits they create should be turned down. Being able to take such a decision quickly while still interacting with the customer requires again an in-depth understanding of the customers' customization needs. In addition, a well thought through design of the service can ease the decision making process when it provides pre-defined customization options which are prioritized based on the value these options generate. Designing a service in this way, in turn, needs also be grounded on a thorough understanding of customization needs as otherwise the provided customization options might not create the desired value. Intangibility of the service offer In addition to the integration of customer in the delivery process also the intangibility of services affects the mechanism of mass customization. Intangibility reduces the observability of the service a priori (Zeithaml and Bitner 2000). This poses a challenge to consumer’s ability to imagine the "product" and it is more difficult for the service provider to communicate the advantage of a new offer (Lievens et al. 1999; Mittal 1999). If a rather intangible service is modularized, the customer will be confronted with a larger number of abstract service modules. Consequently, it will be relatively more difficult for the customer to differentiate between various service modules and to decide, which one is important for him. For instance, compare the resulting choice complexity based on modularization of health care services or consulting services by a lawyer and with the complexity involved in the configuration of footwear. Hence, the intangible nature of services relatively increases the complexity of the configuration process per se. Consequently, in order not to confront the customer with a too complex set of configuration options, extra care is required in the management decisions that lead to an understandable modularization of a service (Heim and Sinha 2001). A well-grounded management decision about the modularization, again, increases the role of an in-depth understanding of customers customization needs (Zipkin 2001). Therefore, empathy and deep understanding of consumers' customization needs is key for flexibly adjusting the choice menu and creating a superior service experience. Service employee must be empathic to a consumer’s customization need. This means for instance, that they need to develop a feeling for a customer’s familiarity with configuration choices and his expectations concerning choice variety. Service employee must be able to flexibly adapt the choice menu and to support the customer during the act of configuration. Thus, they need to reduce

608


the available choices in cases a consumer seems overwhelmed by configuration complexity or expand the choice menu in case customer expects more variety. The key challenge – translating consumer needs into customization concepts and guidelines Both the integration of the customer in the act of customization and the intangibility of the service offer emphasize the need for a clear design of the service customization options as well as the appropriate education and support of the service employees (Gwinner et al. 2005; Meuter et al. 2000). However, to effectively support the design of a mass customized service and corresponding training of service employees two translation tasks have to be fulfilled. Firstly, market research must enable the identification of what Frei (2008) called "customer operating segments". Customer operating segmentation seeks to identify common needs with regard to service quality across traditional segments rather than highlighting differences between customer groups for the purpose of creating better communication messages. In the context of mass customization of services taking a customer operating segmentation approach means to identify common consumer needs towards customization options and processes. The gained understanding of what consumers would consider a good customization offering and experience can then be used to guide the design of a customization concept, i.e. the decisions about how a certain service offering should be modularized and what customization options should be made available to the customer.

Real Consumer Needs translated into

Customization Concept

Enabling sales personnel to flexibly react to the consumer’s customization needs

translated into

Configuration Guidelines Figure 2: Translating customers needs into customization concepts and guidelines.


609

The second challenge is to translate the developed customization concept in guidelines and descriptions for the front end service personnel, that help them to apply customization strategies "on-the-fly" considering the specific customer needs and expectations as well as reacting flexibly and efficiently to create a superior service experience. "Personas" as Enabler for Mass Customization of Services – Insights from an Explorative Case Study As solution to these "translation problems" we propose to use "personas", a concept that has its origins in software design but is nowadays also widely used by web designers and marketing agencies (Chon 2006; Cooper 1999). Personas have been applied by leading consumer brand such as Analog Devices, Amazon, BBC, Best Buy, Chrysler, Discover, FedEx, Fidelity, Ford, MINI Cooper, Microsoft, SAP, Sovereign Bank, Staples, Unilever, or Whirlpool. We will first provide an introduction to the persona concept, then give an overview on how personas have been put to use by various industries and finally discuss the advantages of using personas in mass customization projects. Introduction to personas Essentially a persona is a narrative and explicit description of a particular customer who represents an entire customer segment. This description is vivid and easy to grasp and includes various key attributes, goals, and needs of the particular person. Thus, personas convey information about preferences of a customer segment regarding price sensitivity, product design, or communication tools through the description of a "real" person. This description is enriched with background stories, history, likes and dislikes, which makes it more vivid but also in some sense more fuzzy. An example of such a persona description of the segment "young professional" of a global telecommunication provider is illustrated in Figure 3. The persona "Martin" has (1) a name, face, job, and home, (2) evokes empathy with rich details as well as pictures, (3) enables design decisions and (4) calls out key attributes and high-level goals. The persona describes a fictitious person, but this description is derived from real facts about customer needs gained through qualitative market research (focus groups, customer observation etc.). It is important to stress, that personas are not an average description of a marketing segment. Such a description would be not specific and vivid enough to deliver enough empathy for the customer. Conversely, personas condense rather complex and "dry" market research data into the

610


description of a fictitious person which vividly represents the needs of the respective customer segments.

Martin – A young professional and mobile phone customer “Hi, I’m Martin. I’m 24 and I work for a bank in Boston. I like my job, it is interesting and I don’t work too much hours. I’m usually out the door by 8 a.m. and back by 7 or 8 p.m. This is my first apartment I’ve lived in alone. It is a lot better when you have no roommates.” “Steady girlfriend – well, not now, but I am always open for a flirt. I want to look good for the ladies – that’s why I spend a lot of my free time working out and doing all kinds of sports. At least when I’m not chilling out with my friends.” “When me and my friends go clubbing we stay constantly connected with people. This way, we don’t miss a thing. Just a quick text message away and we know where the action is. That’s when I ask myself how people ever were able to meet without mobile phones.” “My mobile tariff is great for that, too. I get lots of free SMS and with the cost control function I can keep track on how much I spend. Once in a while I download a new ring tone or a song. My new phone has all kinds of extra features and is really up to date. It lets me surf the Net and check my e-mail. But I still visit electronic stores once in a while to keep up to speed on what’s new. I can’t afford to be out of style, after all.”

Figure 3: Example persona used by a mobile telecommunications provider.

Application fields of persona – a brief review In the following paragraphs we assess to what extent and for what purposes the persona concept has been put into practice. Aggregating the extant literature on the usage of the persona concept highlights two broader fields of application: Firstly, personas as means for facilitating a user friendly design of customer interfaces and touch points. Secondly, personas as a tool to support the visualization of "traditional" customer segmentation approaches. Although both application fields rely on personas as tools of clearly communicating customer needs there are subtle but crucial differences between the two. User-friendly design of customer interfaces and touch points Applying personas in order to improve user-friendliness of customer interfaces and touch points is strongly rooted in Cooper’s original intention of the concept. Cooper (1999) envisioned personas as tool to improve the ease of graphical user interfaces and described its application to the development of an in-flight


611

entertainment system. In a similar vein is the case study of a rear seat entertainment systems provided by Brechine (2002), an employee of Cooper’s design consultancy. Further examples of software designers relying on personas to enhance the ease of use of graphical interfaces are the development of the Microsoft Internet Explorer (Buss 2006) or SAP products. A slight adaptation of the original concept occurred when web designers began to use personas in the development of online sites. In the case studies described by authors such as Cunningham (2005, design of a university web site) or Head (Head 2003, redesign of BBC’s website) shifted the focus of the persona concept from user needs with regard to the handling of software applications to the information needs of users. The application of personas to the design of online sales channels can be considered a combination of both approaches: When Whirlpool re-designed its online sales portal it used personas that conveyed information needs of potential customers with regards to products as well as their needs with regard to the handling of the purchase interface (Wasserman 2006, see also Buss 2006 for an account of HomeBanc Mortgage’s website redesign). Best Buy left the realm of software applications and used personas in the redesign of shop floor layouts (Buss 2006). Common to all approaches described above is that they take "customer operating segments" as starting point (Frei 2008). Persona development followed an customer operating segmentation approach in the sense as it was aimed on identifying common needs with regard to the handling of graphical user interfaces, information content, or sales touch points (Brechin 2002). Once the common needs of consumers with regard to the usability of user interfaces, web site content, or shop floor layout had been identified personas facilitated the empathizing with the consumer needs and their sharing across all relevant stakeholders of the design process. This, in turn, contributed to a better outcome of the design process. Visualizing customer segments with the help of personas Screening of the existing literature also shows that personas are not only used within the context of designing customer interfaces and touch points. Companies such as Delta Airlines, Unilever, or Zippo rely on personas to help their marketing and sales teams to better understand what stands behind established customer segments (Buss 2006, Wassermann 2006). For instance, Unilever created the persona "Katie" for the promotion of one hair care product. Understanding that female target customers matching Katie’s personality favored certain TV programs, Unilever focused its media buying on the respective shows and tailored its marketing messages according to these TV shows' target customers (Wasser-

612


man 2006). Delta Airlines used "Ted" to familiarize its flight attendants and ground personnel with the needs of Delta’s primary business customer segment (Buss 2006). In contrast to persona-based design processes which revolve around customer operating segments, these approaches have their starting point in "traditional" socio-graphic or psycho-graphic customer segmentations and are aimed on fostering the creation of well-targeted marketing messages. In this way personas are used to condense customer segmentation data into vivid description of fictions personalities in order to make the complex data much easier to grasp. The most extreme departure from Cooper’s original concept is the case of Chrysler which created so-called persona rooms (Chon 2006). Persona rooms recreate the living rooms of personas which in turn represent certain target customer segments. Rationale for building a living room for a person who does not exist is to provide marketers as well as their marketing agencies with a tangible context that helps them to better understand on an emotional level target consumers' brand preferences and purchase behavior. The "typical" buyer of fuel-guzzling off-road vehicles, for example, displays certain socio- and psychographic peculiarities which might also be reflected in his life-style. In his living room you might rather find the remainders of a call-a-pizza and empty beer cans, whereas organic food and yoga equipment would be rather part of an environmentally friendly hybridcar driver’s living room. Being exposed to the traces of a certain lifestyle marketers will find it easier to create life-style appropriate communications and to select the right media channels to communicate with potential buyers (Buss 2006; Manning 2006). Additionally, as personas or their rooms can be easily shared without ambiguities or information loss they facilitate as well the information sharing between marketers and marketing agencies which are in the end responsible for creating the marketing messages (Manning 2006). Personas and services – insights from qualitative interviews A common characteristic of all case studies presented by the extant literature is that the applications of personas took place primarily in the context of goods and neglected the application for services. Additionally, all case studies were very one sided in the sense that they were full of praise for the persona concept and did not provide information with regards to downsides or risks associated with the concept. In order to gain a more balanced view and to better understand how personas were applied in the context of services we conducted qualitative interviews with marketing managers from a globally operating mobile telecommunications group.


613

Rationale of the interviews and the selection of interview partners We selected mobile communications as research object because it can be considered an industry that offers highly customized services which is reflected in the broad range of tariffs plans consumers can choose from. Moreover, the variety of and differences between tariffs plans is in most cases quite complex requiring significant support of the customer in his decision making. Trade news of mobile operators reported that the persona concept received a lot of attention by mobile service providers and has been applied in this industry already several times (Wireless Business Forecast 2006). The results of this qualitative study is gained by interviews with a global mobile communications group that had successfully implemented personas. From this group we interviewed different managers for segment marketing at the international group headquarter level as well as the local subsidiary level. Status, merits and risks of persona application in mobile telecommunications With regard to the field of application we found that the group was following a hybrid approach: On the one hand, personas were used for internal communications of target segments which had been developed through a "traditional" customer segmentation approach. Each of the identified segments was represented by one persona and the different personas, in turn, were showcased through various internal communication means. The names of the personas and the associated segment classifications thus became widely known throughout all major subsidiaries. On the other hand, the persona descriptions transported rather product design relevant than marketing message relevant information of the target customers' needs. In this sense, although persona did not play currently a role in product development processes, the foundation for using them to guide product design had been laid. However, a modification of the current persona set towards a stronger focus on facilitating life-style and psychographic oriented marketing communications seemed also likely based on one of the manager interviews. Personas as such were considered as a "great tool to vividly communicate the key attributes and needs of target segments to all employees". Additionally, our interview partner stressed that personas were a superior tool to facilitate thinking in target segments which must be considered quite a new way of thinking in the mobile industry as it has just recently lost its status as growth industry and entered a stage of market saturation. With regard to risk and downsides of the concept our interviewees stressed that for certain segments a single persona might be "too pointed". When a single target segment displays quite a broad range of variances, which is not unlikely particu-

614


larly with regard to tastes, fashion preferences and social habits, it can not be adequately represented by a single persona. For example, the target segment "teens" can encompasses an age group from 14-24 years, but the persona of an 18 years old would never be adequate for representing the personality of 14 years old customers. Benefits of using personas in the development of service mass customization concepts The preliminary insights from the literature review and the qualitative interviews suggest that personas could fruitfully be applied in the context of mass customization of services. Using Personas in the development of mass customization concepts for services may help companies particularly in three ways (Guenther 2006; Head 2003) which are visualized by and further elaborated below:

Development of a shared understanding of the customer

Focus on realistic customer needs for a solution

Guidance for the frontline service employee

The major advantage of personas is to avoid misconceptions about the customer needs during the design of the configurator. Although in any design project the "user" should be in the focus, the term remains usually rather vague and imprecise defined. Cooper talks about this as an "elastic user", where everyone has an own understanding how the user looks like and what he need (Cooper 1999). Moreover, there is no guarantee that different people, e.g. marketing and service development departments, involved in developing a mass customized service share actually the same vision of the user. This is because marketing and development departments usually talk about service options and requirements in different terms. Consequently, it is not seldom the case that the designers formally fulfill the given specification but nevertheless fail to satisfy effectively the needs of the target customer marketing had in mind. In contrast, personas are very real, precise and vivid descriptions of target users and their needs. Therefore, people from different departments can easily gain a shared understanding of user needs when they talk about personas instead of abstract requirements (Cooper 1999). Developing personas requires thinking about and in-depth analysis of customer needs, and triggers therefore customer centric thinking in all integrated departments. Creating a persona forces marketing, management, innovation and design department to think about and to analyze customer needs in particular. These "realistic" descriptions of a person can be helpful for managers and designers to make better-informed modularization decisions. Personas ensure that the

615


customization strategy is based on real existing market segments with clear definition of the needs and therefore personas can be used to prioritize the requirement of the configuration options – what is essential and what is discretionary. To put it differently, using personas as design guidance ensures that only those configuration options are offered that are of major importance for a real user (Guenther 2006).

Customization Concept Developers Guides design of the customization concept

PERSONAS

Vivid and easy to grasp description of the key attributes, goals and needs of the most important customers

Guides design of the configuration rules

Service Personnel (Concept Users) Furthers understanding on what has guided the designers Enables flexible modification of the configuration rules Shared understanding of customer needs

Figure 4: Benefits of personas in mass customization of services.

Finally, Personas are a perfect tool for spreading information about customer needs across the whole organization as the convey information about customer needs in an easy to grasp way. A key challenge for implementing customer centric thinking is to transfer the knowledge about customers from the marketing department to other units such as call centers and points of sale. Being introduced to a "real" person supports customer contact personnel but also product developers in empathizing with customer needs and responding more adequately to these needs. Personas are helpful to educate the customer contact personnel appropriately to ensure the quality of the delivered service. This description conveys the key attributes and the key customization goals of the customer in a way that makes it easy for employees to understand and to empathize with. Rather than being confronted with complex and abstract market segmentation data, employees are introduced to a "real" person. Personas give the customer "face" back (Cooper 1999). A clear understanding of the target user’s needs also provides a clear yardstick when it comes to evaluating the achieved quality of the customized service. The ultimate goal of any customization should be to satisfy the customer’s needs (Gwinner et al. 2005). Personas help the contact employee to

616


identify the right customer type at the beginning of the interaction with the customer. The Persona-Based Concept of Service Mass Customization In the following paragraphs we conceptualize a theoretical framework for the persona development process, which can be used by firms to support their service mass customization process. The concept applies the advantages of personas in a wide organizational context (from the corporate culture development to the more tangible aspects of mass customization of services) and is based on our literature review and analysis of existing case studies of personas application. The process of persona development has six steps and can be divided into two main stages a) persona creation and b) persona implementation and is illustrated in Figure 5.

Goal Setting

Segmentation

Business Mission

Market research

Internal Goal

Detect segments

Stakeholder goal

Identify segment needs

Persona identification

Persona verification

Persona Embedding

Persona application

Identify individual as role model for personas

Consolidate in Persona sets

Introduction of the selected personas to all relevant areas

Interviews

Identify overlaps

Consistent usage of the selected personas in Management, Design, and

Focus groups

Prioritize personas

Detect individual needs

Ensure the knowledge and acceptance of the persona

Customer contact Verify Personas regularly

Persona Creation Phase

Persona Implementation Phase

Figure 5: The persona development process.

Persona creation Crucial for the beginning of the persona creation process is to ensure that the persona development is aligned to business goals and conforms to best practice. This includes that the goals and objectives of all relevant stakeholders like management, innovation, marketing as well as support and front end employees are considered. Internal goals need to be clarified in order to provide the general objectives and constraints of the upcoming mass customization projects. The next step includes the elaboration of a profitable and useful market segmentation, which targets the most promising customer segments and detects their major needs and wants. Here are the classic data analysis tools of market research for


617

segmentation used. Objective of this step is to draw upon existing segments and to generate the commercial framework in which personas have to fulfill their purpose After all, persona creation is about studying the user and his needs. The efforts of persona development, therefore, must be directed to target users who represent attractive customer segments and ultimately provide viable business opportunities. The previous process step has clarified from which market segments target users should be selected. This step now prepares the actual user studies. For each customer segment individual users who can be used as role models for the segment need to be identified and gained as participant for the user study. These studies can either take the form of focus group interviews or the form of observational field studies (Goodwin 2002). In the former case, target users are invited to a test facility for an interview about when, where, how, and why they engage with a product or a channel. In the latter case, the researchers visit the homes or offices of the selected target users and observe as well as interview the users as they try to achieve their goals in their normal work environment (Fogliatto and Silveira forthcoming). After the data gathering stage, data analysis will be executed to uncover clusters of key customization needs (Piller and Müller 2004). Those that are relevant for designing the configuration concept build the foundation for the creation of the personas. The results of this step are persona descriptions that are based on primary research with real people. Every element can be traced back to research. The description should include a name, face, job, and home, call out key attributes and high-level goals. Furthermore, the persona should easily be recognized by service front end personnel and be formatted as vivid narrative. The narrative evokes empathy with rich details and focuses on enabling design decisions (Goodwin 2001). This means for instance that service personnel feel that such a kind of person regularly walks into the shop or the call center agents think that they have just talked to such a kind of person. At the end, every customer segment should be linked to one persona as representative. This does not necessarily mean that each customer segments has its "own" persona. The clustering of the customization needs might reveal common needs that stretch across the boundaries of "traditional" segments thus leading to personas which are "shared" by more than one socio- or psychographic segment. The clustering of needs across segments notwithstanding, typically the data and its analysis lead to more than one persona. Quite often these personas have conflicting goals and needs. However, it is not viable to attempt to accommodate these diverging needs with a single configuration design. The management has to

618


consider the individual needs, identify overlaps of the personas and finally prioritize the personas and select the primary ones. The configurator setting should be focused first and foremost on the primary personas. Ideally, satisfying the primary persona does not create barriers for other personas, but make them happy as well (Guenther 2006; Head 2003). Persona implementation In the fifth step the created personas will be implemented by embedding them into the organization and applying them to the mass customization project. The objective of personas is to remind employees of and to focus them on the needs of the most important customers. In that way personas also provide a common denominator for aligning the whole organization. Obviously, to serve this purpose, personas need to be known and accepted by the employees. Therefore the newly created personas need to be communicated within the organization. It is very important to have a live meeting when first introducing personas in order to cheerlead, persuade, and address any concerns people have. A crucial factor for the success of these internal marketing campaigns is the backing of the top management. It must be clear to all employees that the needs of the target customers and personas as their representation are not negotiable. The final step is the application of the persona concept to the delivery of the mass customized service offers. Applying personas means that service personnel embrace personas as helpful tool for them and use it to offer a better and more customized service experience for the customer. Furthermore, a consistent usage of the selected personas in management, design and customer contact should be regularly verified. The detailed description of a persona should serve as a lasting reference and should be quickly accessible whenever and wherever a customization decision needs to be made. Any format such as posters, persona t-shirts, cups or laminated placemats that can easily be shared are helpful for this purpose (Goodwin 2006). Summary Mass customization of personally delivered services raises specific challenges both for the implementation of a mass customization concept and the actual delivery of the mass customized service. With this article we have elaborated the specific challenges of mass customizing services and proposed to tackle these challenges with a Persona-based approach. We demonstrated the value of the Persona concept in various industries and also verified in a first qualitative study its benefits in the context of service mass customization.


619

In order to master the specific challenges of mass customizing services it is required to solve a two-staged "translation problem": At first, consumer needs have to be translated into a viable customization concept, which supports service design. The customization concept, in turn, needs to be translated into customization guidelines, which support customer contact personnel during service delivery. Personas are a well suited tool for overcoming these translation problems as they condense complex market research data into narrative descriptions of archetype customers. Such archetypes convey customization needs in a way that makes it easy for decision makers and employees to understand and to empathize with such needs. The main advantage of personas is that a customer-oriented implementation of the mass customization concept into the entire service organization is ensured. Firstly, developing of personas triggers customer centric thinking within the company. Secondly, once created personas provide a common language and understanding about the customer, which supports the communication between all related departments (Marketing, Design, Management, Service Personnel) during mass customization projects. This implies that the concept cannot only be used to facilitate the development of the mass customization concept, but is also suited to support the education of the customer contact personnel in a way that improves the quality of the service delivery process. Moreover, personas focus the customization strategy on real existing market segments. This ensures that the service mass customization strategy stays profitable and the configuration options are relevant for the target group. Especially in industries, where consumers have a lot of different options, like the mobile services industry, this is a very important aspect for the mass customization decisions. Various case studies from different industries and explorative interviews support the successful applicability and value of the Persona concept. The case studies show those personas are used both for improving design processes as well as for supporting internal communications of customer needs. Building on these case studies and interviews we provided a framework that supports the development of personas and their application to mass customization projects. This framework consists of a six-stepped Persona creation process that can be used by managers to develop personas systematically for their service mass customization projects. It can be seen as a general approach for developing service mass customization strategies and is widely applicable. However, this article shows only first experiences with the application of personas for service mass customization based on qualitative research. Further research

620


based on quantitative data is necessary to verify also the success of personas in general for the service industry. The first insights are promising and with the proposed framework a fast application of persona for service mass customization can be initiated.

References Brechin, E. (2002). Reconciling Market Segments and Personas, Newsletter Cooper.com. Buss, D. (2006). Reflections of Reality, Advertising Age. 77 (23): 10–11. Chon, G. (2006). Chrysler’s Made-Up Customers Get Real Living Space at Agency "Persona Rooms" Help Marketers to Understand How Target Drivers Think, The Wall Street Journal, Jan 4, 2006. Cooper, A. (1999). The Inmates Are Running the Asylum: Why High Tech Products Drive us Crazy and How to Restore Sanity. Indianapolis: SAMS. Cunningham, H. (2005). Designing a Web Site for One Imaginary Persona that Reflects the Needs of Many. Computers in Libaries, October 2005, 15–19. Dellaert, B. G. C. and Stefan S. (2005). Marketing Mass-Customized Products: Striking a Balance Between Utility and Complexity, Journal of Marketing Research. 42 (2): 219–27. Fogliatto, F. S. and G. J. C. da Silveira. Mass customization: A method for market segmentation and choice menu design, Int. Journal of Production Economics. 111 (2): 606–622. Franke, N. and F. T. Piller (2003). Key research issues in user interaction with user toolkits in a mass customization system, International Journal of Technology Management. 26 (5/6): 578–99. Frei, F. X. (2008). The Four Things a Service Business Must Get Right, Harvard Business Review, April 2008, 70–80. Gilmore and Pine (2001). Markets Of One. Boston: Harvard Business School Press. Goodwin, K. (2002). Getting from Research to Personas: Harnessing the Power of Data., Newsletter Cooper.com (Nov.). Goodwin, K. (2001). Getting Perfecting Your Personas, Newsletter Cooper.com (Jul). Goodwin, K. (2006). Taking Personas Too Far, newsletter cooper.com (33). Guenther, K. (2006). Developing Personas to Understand User Needs, Online. 30 (5): 49–51. Gwinner, K. P., M. J. Bitner, S. W. Brown and A. Kumar (2005). Service Customization Through Employee Adaptiveness, Journal of Service Research. 8 (2): 131–48. Head, A. J. (2003). Personas: Setting the Stage for Buidling Usable Information Sites, Online. 27 (4): 14. Heim, G. R and K. K Sinha (2001). A product-process matrix for electronic B2C operations: Implications for the delivery of customer value, Journal of Service Research. 3 (4): 286–302. Kaplan, A. M. and M. Haenlein (2006). Toward a Parsimonious Definition of Traditional and Electronic Mass Customization, Journal of Product Innovation Management. 23 (2): 168–82. Lievens, A., R. K. Moenaert and R. S. Jegers (1999). Linking Communication to Innovation Success in the Financial Services Industry: a Case Study Analysis, International Journal of Service Industry Management. 10 (1): 23–47. Manning, H. (2006). Persona Rooms: What, Why, And How, in Best Practices.


621

Meuter, M. L., A. L. Ostrom, R. I. Roundtree and M. J. Bitner (2000). Self-Service Technologies: Understanding Customer Satisfaction with Technology-Based Service Encounters, Journal of Marketing. 64 (July): 50–64. Mittal, B. (1999). The Advertising of Services, Journal of Service Research. 2 (1): 98–116. Piller, F. (2001). Mass Customization. Ein Wettbewerbsstrategisches Konzept im Informationszeitalter. Wiesbaden: Gabler. Piller, F. T. and M. Müller (2004). A new marketing approach to mass customization, International Journal of Computer Integrated Manufacturing. 17 (7): 583–93. Pine (1993). Mass Customization: The New Frontier In Business Competition. Boston: Harvard Business School Press. Wasserman, T. (2006). Unilever, Whirlpool Get Personal with Personas, Brandweek. 47 (34): 13. Wireless Business Forecast (2006). Can Carriers Really Go Granular?, Wireless Business Forecast, 200603-23, 8. Wirtz, J. and J. E. G. Bateson (1999). Consumer Satisfaction with Services; Integrating the Environment Perspective in Services Marketing into the Traditional Disconfirmation Paradigm, Journal of Business Research. 44 (1): 55. Zeithaml, V. A. and M. J. Bitner (2000). Services Marketing: Integrating Customer Focus Across the Firm. Boston et al.: McGraw-Hill. Zipkin, P. (2001). The Limits of Mass Customization, Sloan Management Review. 42 (3): 81–88.

Author Biographies Dr. Michael Haas is a consultant in the Berlin office of the A.T. Kearney GmbH, Germany Contact: [email protected] Prof. Dr. Werner H. Kunz is an Assistant Professor of Marketing at the College of Management in Boston. His research interests are E-Services, Social Networks & Media, Innovation, Communication and Service Research. His professional activities are split up in academic contributions for journals, associations, and conferences. Further, he is doing consulting projects and executive teaching courses on Social Media, Innovation, Internet Marketing, Service and Quality Management to private and public sector organizations. Contact: [email protected]

1.2

Mass Customization in Wireless Communication Services: Individual Services and Tariffs Hong Chen Rotterdam School of Management, Erasmus University, The Netherlands L-F Pau Copenhagen Business School, and L.M. Ericsson AB, Denmark

This chapter presents results on mass customization of wireless communications services and tariffs. It advocates for a user-centric view of wireless service configuration and pricing as opposed to present-day service catalogue options. The focus is on design methodology and tools for such individual services and tariffs, using altogether information compression, negotiation algorithms, and risk portfolio analysis. We first analyze the user and supplier needs and aspirations. We then introduce the systematic design-oriented approach which can be applied. The implications of this approach for users and suppliers are discussed based on an end-user survey and on model-based calculations. It is shown that users can achieve desired service bundle cost reduction, while suppliers can improve significantly their risk-profit equilibrium points, reduce churn as well as marketing costs, and simplify provisioning.

Introduction This chapter presents the progress and challenges of the research and needed support tools in mass customization of mobile/wireless communication services. Recent research in mass customization has primarily focused on the customization of physical products such as clothing (e.g. footwear, garments, prêt à porter), consumer electronics (e.g. watches, laptops), etc., and on user design toolkits for such customizations (Abou-Jaoude and Kung 2005; Au and Goonetilleke 2005; Franke and Pille 2004; Ogawa and Piller 2006; von Hippel 2005). Services differ from physical goods in their intangibility, perishability, heterogeneity and in the inseparability of production from consumption (Zeithaml et al. 1985). Mass customization of services is rather limited and applications are spread across different industries such as education, finance, health care, etc.; the corresponding research publications have focused on mass customization strategy (Gabriel et al. 2006; Grenci and Watts 2007; Lampel and Mintzberg 1996), customization 622


623

frameworks, processes and customization approaches under specific contexts (Huang and Lin 2005; Winter 2001), and customer value perception (Prahalad and Ramaswamy 2004). Current practices of mass customization in the mobile communication industry mainly focus on the industrial design of the wireless terminals, on user interfaces and sometimes on software based feature selection; see e.g. Sigala (2006) for customers' perceived value dimensions (e.g. functional, social, and emotional) of mass customized "mobile phone + services" and implications for suppliers in conducting a customer value-based market segmentation. Some mobile terminal suppliers have made mass customization of their designs their key intellectual property rights (IPR) based business model. This is accomplished via common design platforms that help third parties produce competing customized products (e.g. Ericsson Mobile Platforms AB, Modelabs, and Qualcomm). With regard to public mobile services which involve the network infrastructure for the provisioning of such services, mass customization remains limited despite mobile users numbering in excess of 2 Billion. Public operators usually offer 5-10 generic service bundles which are sold "unmodified" to anyone and include only basic service functions that most customers are expected to use, such as voice communication, SMS; each bundle corresponds to a set of tariffs. User chooses a bundle, signs a contract (usually for 1 or 2 years) and pays a fixed monthly fee, which may cover a maximum usage of voice minutes, SMS, MMS or a certain amount of data traffic, plus limited customer support and wireless terminal amortization. Costs for usage exceeding the default service usage limits vary among bundles. There is no interaction between the supplier and the user, other than when a user is choosing from a supplier’s defined catalogue of services (with the exception of a few options on customer management). Different suppliers offer almost the same bundles except for prices, support functions and upgrades. Personalization of the services is limited to a few discrete user choices relating to the same bundles. Scope of the Paper In this chapter and other related papers (Chen and Pau 2006; Chen and Pau 2007a; Chen and Pau 2007b; Chen and Pau 2007c; Pau and Chen 2006), we advocate for a much bolder user-centric view of wireless service configuration and pricing. The goal is to let all users who wish to manage service and content access themselves, define which functions, content, contract duration and price they will accept, waiting for competing suppliers to bid for it. Once a supplier has been chosen and

624


has accepted, he must provide such a mass customized service to this user for a duration chosen by the user. Obviously not all users want to take this individualized customization into their own hands, in that many will just be happy with the generic catalogue bundles and the lack of transparency attached to the provisioning costs and quality of service (QoS) of such generic bundles. Thus, the real challenge behind mass customization of mobile communication service and content bundles is, akin to general governance concepts, how much the user as a stakeholder needs/wants to know, and how much the supplier is willing to reveal, about the infrastructure characteristics and operations. A second challenge is to establish how the laws of large numbers can actually benefit the supplier who is willing to offer individual bundles and tariffs, by increasing user loyalty over time, enhancing the service demands, and reducing the financial and operational risks. Therefore, we will in this paper systematically analyze the user and supplier needs and aspirations, benefits and risks, as a result of a proposed negotiation process supported by the corresponding tools. The implications of mass customization of mobile services are studied both via user surveys and via quantitative tools. The text is organized as follows: Section 3 analyzes user and supplier needs and expectation from mass customized mobile services; Section 4 summarizes the progress of a series of projects we have conducted under the title "individual tariffs", which provide methodology and models to design mass customized wireless services and tariffs; Section 5 discusses the implications of individual services and tariffs for both the users and the suppliers; the analysis is based both on the results of a survey and on model based calculations in a mobile music service case; Section 6 extends the previous results by showing that even with standard generic service bundles, just letting the user specify a duration and an acceptable cost to him, bring also value to the operator; Section 7 discusses the open research issues; Section 8 concludes the chapter. User and Supplier Needs and Expectations from Mass Customized Mobile Services User needs and expectations As mobile wireless becomes the dominant communications access technology worldwide, mobile services, especially value added services are growing in an explosive manner. User’s needs and expectations from mobile services are changing. According to a worldwide survey conducted by the Rotterdam School


625

of Management (Chen and Pau 2007c) with the participation of the International Telecommunications User Groups (INTUG):

Users are demanding specific service bundles that meet their unique demands; there is a wide variety of needs but also significant overlap (in terms of proposed applications and services) amongst individual users;

Users want to reduce cost and simplify their service bundles when given the opportunity to personalize their service bundles and tariffs. It is highlighted that such requirements change over time quite often, thus leading to short or long contracts with different service characteristics and prices. This is reflected in the answers of 51% of the respondents who consider personalized service bundles and tariffs to be a cost reduction feature; while the rest consider them as a life style feature (28%) and productivity feature (10%). Regarding the pricing, 47% of the respondents would pay less than 50% of the original price when the number of service features in the service bundle is reduced by 50%; 40% of the respondents will pay exactly half; and 13% of the respondents are willing to pay more than half;

Users are demanding greater flexibility in contract length. The average ideal contract length is 6.8 month when the respondents can define the contract length themselves;

Users are willing to spend time on personalizing a service. The majority of the respondents chose 30 minutes (22%) and 1 hour (41%).

Supplier needs and expectations At the same time, the public suppliers in the mobile communications industry are facing:

A continuously increased competition which was brought about by deregulation, privatization and liberalization (Geddes 2000; Noam 1983; Pau 2002; Xu and Pitt 2002). Competition is both technological, in that e.g. fixed telephony is decreasing in proportion vs. other access technologies, as well as market based in that marketing costs and price pressure have major impacts while the number of suppliers increases;

The fundamental layering introduced by the 3rd Generation Partnership Project (3GPP) consisting of access networks, transport networks, and service layers, so that one given supplier may not have to own and operate all three with their related huge complexities.

626


Largely saturated markets for wireless communications in early adopter geographical zones, with huge marketing costs, high churn rates and high customer retention costs as pitfalls (Wieland 2006).

A technology and customer push for converged services; here the meaning of convergence is twofold: the convergence amongst various access technologies (2G, 2.5G, 3G, B3G, WiMax, fixed) and the convergence of access, content and applications;

High pressure on overstaffed incumbent operators to reduce operational processes and costs, even outsourcing many wireless service operations to systems suppliers or integrators;

Insufficient investments in service creation, commoditization of key infrastructure elements, in content access rights, etc … The segmentation of mobile services into limited generic bundles can no longer satisfy the needed actions from the above trends, as they fundamentally have diminishing returns from ever increasing provisioning complexities. It is high time that user-centric mass customization, with user-supplier interactions, is introduced in mobile communication services; it will fundamentally allow the suppliers to achieve a higher granularity of their service and access components and thus finally to clinch more stable revenues from managed services rather than from managing network capacity exposed to user churn. Such a mass customization via negotiation on the basis of stated user requirements is also a fantastic dynamic data mining and targeting capability allowing to reduce general marketing costs.

Modeling the User, the Supplier and Their Interactions in the Context of Mass Customized Mobile Services and Tariffs Definition of individual services and tariffs Under the context of mass customization of mobile services, we define individual service and tariffs as the regulatory protected ability for an identified user to obtain from a service provider, by a bilateral specific contract, a set of service and related content specific prices corresponding to a user request specified with a service demand profile and some duration. In short, the user will have tariffs unique to him depending on the services he requires in terms of quality and quantity. The research on individual bundles and tariffs can be divided into three major areas:

The study of users from the perspective of mass customized consumer behavior in services (different from products) (Chen and Pau 2006).


627

The study of suppliers exposed to mass customized services and tariffs which focus on service and content bundling models, service provisioning, and on churn reduction problems (Chen and Pau 2006; Chen and Pau 2007a; Pau and Chen 2006).

The study of the interactions between a user and a supplier for individual service bundles and prices. The focus is on negotiation models, with all refinements including risk analysis; and on the operational environment for the deployment of such models in view of the large number of negotiations (Chen and Pau 2006; Chen and Pau 2007a; Chen and Pau 2007b; Pau and Chen 2006). We take a utilitarian approach when modeling the users and the suppliers.

Modeling the users The user, instead of being fully rational, has "bounded rationality" (Simon 1957). He tries to optimize his utility but in a simple way. We propose that the user builds his utility function based on a set of "perceived attributes" of a mobile service. The "perceived attributes" are a reduced, e.g. 3 dimensional mapping of the e.g. 20+ service attributes defined by the service provider using full service characteristics in the "service design space". The mapping is determined by user surveys, previous customer information, and/or actual service usage as communications networks offer unique self-monitoring capabilities not found in other application areas for mass customization. We introduced a computational method to design such service specific perceived attributes, using principal component analysis adapted in a specific way (Pau and Chen 2006). In the reduced "perceptual space" constructed by the perceived attributes, the user’s utility function is defined in an easily understandable way as the Euclidian distance between a point ideal to him (target point) and the supplier’s service bundle offer. The user maximizes his utility by minimizing the distance, subject to his social and economic constraints. He uses satisfying rules in decision making. Modeling the supplier(s) The supplier is considered to be a profit maximizing firm, with detailed modeling of an incremental customer’s usage of infrastructure, content, management and other resources. The supplier makes decisions in the "service design space". His decision variables are primarily related to parameters of the technical specifications and configuration of a service (e.g. parameters related to QoS). These technical details are often beyond the interests or understanding of an ordinary user. The supplier’s utility function is defined as the incremental profit from

628


serving the demand of an additional user, which is the user’s bid price minus all non-financial costs corresponding to the user-requested customized service and content characteristics. The utility function of a supplier and the applicable constraints and configuration options can be very complex; it has a lot nonlinearities due to the underlying cost items and technology jumps involved in providing the service (Chen and Pau 2007a). Interaction between individual users and a supplier around individual tariffs and bundles The service personalization and tariff negotiation processes can be modeled by different multi-stage non-cooperative games (Chen and Pau 2006; Pau and Chen 2006). The user and supplier exchange technical, service, duration and price attributes of a mobile service bundle as requested initially by the user, in a "service design space". They take turns to optimize with constraints their own utilities in each stage of the game. We define equilibrium as a situation in which both the user and the supplier are satisfied with the negotiation result and agree to sign a contract. Stackelberg games are used when the user and the supplier have different bargaining power; the leader (the user) has the advantage of moving first. It can be a 1-on-1 or 1-on-N game when the supplier is negotiating with one individual user or a user community (Chen and Pau 2007a; Pau and Chen 2006). In practice, such a Stackelberg equilibrium determination cannot obviously be carried out for any user with any supplier. But each operator will normally have all the information needed as well as the computation power to generate, e.g. in look-up table format, most of the equilibrium cases, or non-contract cases, to be expected in a given market. Risk analysis As they are driven by mass diversity in needs and willingness to pay, the userspecific service bundles and tariffs bring uncertainty to the supplier’s profit: some of them generate profits; others may bring losses. This uncertainty can be reduced if the supplier pools together a large group of users whose preferences follow certain distributions. Note that revenue assurance techniques inspired from the insurance industry can be developed and used. In our research, we apply such statistical methods and techniques in analyzing the consequences and implications when the supplier is offering certain services to specific groups. For example, in the numerical case we designed (see below for details), we calculated the expected profit, value at risk (VaR) and percentage of contracted equilibria using the Monte Carlo method; the basic assumption is that a random user has consistent character-


629

istics to a surveyed group. Such analysis is necessary to give the supplier basic business guarantees to enter individual tariff agreements. Furthermore, the supplier can adjust the parameters in the decision rules which are used in the negotiation; the adjustment can be based on profit vs. risk from the obtained risk statistics (Chen and Pau 2007a). If the supplier has a risk adverse profile, the above calculations produce estimates of the proportion of potential customers that no contracts will be signed with because such customers might have requests that represent the risky service bundles or usage profiles; in such a case, laws of competition should allow such users to find a competing supplier with a different risk profile. What the above capability means in a practical context, can be explained by a short example. If one mobile operator, such as an MVNO (mobile virtual network operator) wants to differentiate itself while winning loyal customers, he will set a relatively speaking high minimum threshold on the proportion of signed contracts, and select, from amongst candidates for individual tariffs, those who meet the constraint.If another mobile operator, with high debt, prioritizes higher the profit from incremental subscribers, he will put a relatively high thrshold on the minimum average incremental profit,and take on less new customers with individual contracts to reduce the loan exposure. Implications for Users and Supplier In the previous section we provided a systematic design-oriented approach regarding "how to achieve mass customized individual service bundles and tariffs". In this section, we give estimations of the implications, benefits and costs, when the users and the supplier commit to individual services and tariffs. These estimates are based both on the end-user survey reported in (Chen and Pau 2007c), and on model based equilibrium and risk calculations from very extensive simulation computations (as explained before):

There were 13 questions in the survey; each focused on different aspects of user preferences regarding individual tariffs. The questions were also chosen in such a way that by nesting them, some estimates of key implicit indicators could be computed. Overall there were 102 respondents worldwide; the majority of them were end-users and researchers.

With respect to the model-based calculations, we use a quasi-realistic operator model with non-linear wireless technology provisioning options, as well as user utility determinations from a survey group of 600 persons. The model was estimated from actual wireless operator analytical data. Furthermore the

630


model calculations include full Monte Carlo simulation, value at risk determination and realistic discretized decision variables on service attributes (e.g. contract duration is in full integer months). Total traffic change Total traffic change can be estimated as the joint effect of the changes in communication patterns and changes in the number of service bundle features due to personalized configurations. Changes in communication patterns are measured in the number of contacts, communication frequency and duration. The survey shows that about one-third of the respondents will not change their current behavior; about 60% of the respondents consider a slight increase up to 50%. On average, the increases will be 16% (number of contacts), 20% (communication frequency) and 34% (communication duration). These increases will lead to increased service bundle usage and network traffic demands. If we assume network traffic changes linearly with the changes in communication patterns, Ceteris paribus, then the increases in the number of contacts, communication frequency and communication duration will lead to an increase in the network traffic demand of: (1+ 16%) * (1+20%) * (1+34%). i.e. 187%. Table 1: Different scenarios regarding the number of service bundle features and traffic change. Relative changes in the number of service design features Scenarios

Low

Medium

High

Respondents want simplification (70%)

25%

50%

75%

Respondents want to keep service design unchanged (20%)

100%

100%

100%

Respondents want more features (10%)

125%

150%

175%

Expected change in the number of service bundle features

50%

70%

90%

On the other hand, personalization with service design (put eventually in the hands of the users) allows for simplification of the services. If we assume the amount of network traffic is proportional to the number of features of the personalized service configuration, simplification could lead to a decrease in network traffic. The survey result indicates that 70% of the respondents considered that an individual service should be simpler when compared to a complete

631


version of the service bundle; 20% respondents would keep it unchanged and 10% wanted more features. The expected change in the number of service features can be calculated from the weight of each group. We assess three scenarios, where the expected changes in the number of service features in the personalized service bundle are low, medium and high. In Table 1, 25% means that the individual service configuration will keep 25% of the features of the original service, 100% means unchanged, 175% means the personalized service configuration will add 75% more features to the original service bundle. Such probabilities are nominal indicators of the different profiles. As it is shown in Table 1, in the low scenario, the result of personalized service configurations is a reduction of 50% of the original service bundle features. Table 2: Total traffic changes under different scenarios. Scenarios

Low

Medium

High

Traffic changes due to changes in communication patterns

187 %

Traffic changes due to changes in the number of features after service design personalization

50%

70%

90%

Total change in network traffic

93%

131%

168%

Given the above estimations, the total traffic change can be estimated under different scenarios. In the low scenario, the total amount of network traffic will decrease about 7%. In the other two scenarios, there will be a growth in network traffic (Table 2). If we assign a probability to the each of the scenarios from Table 1, i.e. feature-poor (0.3), medium (0.4) and feature-rich (0.3), the expected network traffic will be 131%, which is a 31% increase from the present day situation which does not allow service design configuration and no individual tariffs. Of course, the above analysis of traffic change is based on surveys, and actual behaviors may differ from declared ones. If this effect has to be taken into account, it can be in the risk analysis part, by putting a safety margin on the survey results leading to decision rules. Churn rate Churn rate is defined as the percentage of subscribers who leave the contractual relationship with one supplier during a given period (e.g. 1 month). "Net churn" measures the overall loss of subscribers, while "Gross churn" measures the absolute loss. Net churn (%) = [Gross churn (%)-New contracts (%)] (see

632


en.wikipedia.org/wiki/Churn_rate). Here we only consider "Gross churn". By allowing flexibility in the length of individual service bundle contracts, the supplier may have a shortened average contract length from the users compared to current dominant 12 or 24 months terms (with or without new mobile terminal). On one hand, it becomes easier for the users to switch network operators. But on the other hand, personalization can greatly improve user satisfaction, which in turn will reduce churn. This could mean that more users would like to prolong or renew the contracts with the same supplier of an individual service bundle when the previous contract expires, as such users know that this operator provides the flexibility they need. According to Wireless Intelligence (Wieland 2006), and spot data from wireless operators worldwide, the churn rate for mobile subscribers or prepaid users usually falls in the interval 1.5% - 3% /month, with operator, bundle and country specifics. We estimate the gross churn rates under two scenarios: 1.5 % and 3 % monthly churn rates (Table 3). The corresponding yearly gross churns are calculated for reference purposes. Using the result from the survey which indicated an average individual tariff contract length of 6.8 months, the 7 month gross churn rate (rounded up from 6.8) can be calculated. These data can serve as indicators for individual tariffs: by introducing mass customized service personalization for service and content bundles, and individual tariffs, the users accepting such a contract would NOT churn at all within 7 months. This means that for that sub-category of users, churn rate is certainly less than 10.4 % under a low churn rate scenario for all users, and less than 19.2 % under a high churn rate scenario for all users. Table 3: Churn rate estimations. Scenarios

Scenario low churn

Scenario high churn

Monthly gross churn rate

1.5%

3%

Yearly gross churn rate

16.59%

30.62%

7 months gross churn

10.04%

19.20%

Changes in "price/bit" Changes in price per bit can be approximated by the following formula around a given point: Derivative (price/bit) = [d (price)*bit – price * d (bit)] /bit2 = d (price)/bit - price*d (bit)/bit2


633

Consider the scenario that as a result of the personalization of the service bundle configuration, the number of service features is reduced to 50% of the complete version of a service and content bundle. According to the processed answers from the survey, the expected price the user is willing to pay will change to 0.4450 of the original price (of the complete bundle). The traffic under this (low) scenario will change to 93% (Table 2) of the original service and content bundle traffic. A rough estimation of the changes in (price/bit) ratio by compounded derivatives is: [(-0.5550*price/bit –price* (-0.0643) bit/bit2 = - 0.4907* (price/bit)]. Thus price/bit is decreasing by almost half in this scenario to the benefit of the user if individual tariffs are adopted. Service configuration & tariff negotiation From the survey, 23% of the respondents will pay less when they are offered to do the service design personalization themselves, as compared to the same service design managed by the operator. Suppliers can earn extra revenue by offering service configuration assistance for this group of subscribers. About half of the respondents would like to be actively involved in service configuration and tariff negotiation. The high willingness to spend about 1 hour on self-configuration shows eagerness to differentiate and save if cost savings can be generated and/or or unique service features can be chosen. Suppliers can also save costs from those who configure services for themselves. User and supplier benefits and risks from a case with quantitative analysis

The case: We designed a service bundle in the mobile music area and applied the above mentioned methods and models for service personalization and tariff negotiation. We developed a tool to automate the negotiation process for services and tariffs, along with the risk assessment.

User benefits: Our numerical results show that users achieve a 370 % average gain in their utilities in the mass customized mode compared to the initial utilities derived from the selection of a current generic service bundle (Chen and Pau 2007a; Pau and Chen 2006).

Supplier benefits: At the same time, the supplier is able to adjust parameters to achieve a higher average profit than with non-negotiable tariffs according to the risk he is willing to take. Figure 1(a), (b), (c) give a simplified illustration on how the adjustment of the parameter "minimum profit threshold" affects the percentage of contracted equilibria, the expected profit and the VaR. The effects of the adjustment of this parameter "minimum profit

634


threshold" on the supplier’s total expected profit and total risk are shown in Figure 1(d). An optimal solution where the supplier can achieve maximum profit with minimum risk may not exist. The supplier must then base his decision on an equilibrium compromise between the total risk he is willing to take and the total expected profit corresponding to that revenue. In practice, the shape of the curves can be much more complex, see (Chen and Pau 2007a; Pau and Chen 2006) for further details.

Figure 1: Supplier can adjust the parameter "Minimum profit threshold" to achieve different risk / profit results.

Benefits and risks To summarize, mass customized wireless individual services and tariffs can help the users reduce cost, which is the main concern of more than half of the respondents to the user survey. For the supplier, individual services and tariffs may lead to an increase in traffic demand but with less complexity in over- or under-provisioning. Furthermore, individual tariffs offer the supplier flexibility to achieve better control of their goals in terms of market share, profit level and risks by a customer portfolio approach. Additionally, the supplier can benefit from


635

extra revenues and reduced costs by helping, as an additional cost on top of the bundle, users who cannot themselves configure their individual service attributes. Regarding the churn rate, only the upper limit can be estimated, and it shows a significant reduction vs. present churn for generic services; the actual churn rate should be even lower because of the improved customer satisfaction and loyalty. On the issue of what operational measures to take in order to adjust the minimum profit threshold, for a given risk, this question goes beyond the scope of this paper but fall into what is called "revenue assurance" in communications. It is addressed by wide spectrum of techniques such as customer payment track records, to long term or flexible sourcing of content or ressources, to loyalty programs for communications usage, to add-on services tied to the one’s chosen by the customer. Extension to Mass Customized Generic Wireless Services A typical mobile operator offers 5-30 service bundles: each bundle has a set of bounded standardized maximum service usages and a tariff for some duration, plus excess usage conditions. The bundle may cover usage within ceilings for voice minutes, SMS, MMS and/or a certain amount of data traffic, plus limited customer support. A user chooses a service bundle, signs a contract (minimum duration is often fixed by supplier lasting 1 or 2 years), receives possibly an upgrade to an existing mobile terminal, and pays a fixed monthly fee corresponding to the chosen bundle. Additional tariffs for usage exceeding the "bounded" ceilings of the chosen bundle vary slightly among bundles, but usually an overprice applies to the excess usage, unless sometimes unlimited usage of a certain service is offered within specified hour/date/destination combinations. The limited 3-30 type segmentation often leaves a significant number of demands from the users unsatisfied. Generic mobile services differ from value-added mobile services by usually having a much larger number of customers, so that the pre-existing installed base is much larger than the ones used in the previous section. The result of this research setting can be used by users who want only individualized generic services, or it can be used in connection to the negotiation of a value-added service which uses some of the generic service attributes, as in the above. Behavioral models In the same way as with the user behavioral model before, the user is characterized by having both economic and social concerns when designing an individual

636


generic service bundle. He has bounded rationality and uses a simplified decision model with satisfying decision rules. To reduce the complexity for the user in service design, we introduced earlier the concept of a service perceptual space, which is a reductionist mapping of a service design space. The reductionist mapping is, however, not necessary when the user and supplier jointly design a generic mobile service bundle for the user. Most of the time, the service bundle consists of straightforward and well understood services such as voice, SMS and data. Other attributes of the service bundle include contract length and price per month. The number of service attributes to be individualized is only between three and five; therefore the service perceptual space is not used here. The supplier and user share the same generic service bundle design space for the negotiation, which is also their decision space. However, like in Section 4, the supplier will still have for his internal use a large number of engineering attributes for this generic service bundle. The supplier is a mobile operator characterized by seeking a maximum profit with a minimum risk. Computational model The computational models of the user and the supplier are similar to those presented before. The approximated supplier’s model and prototype mobile music implementation can be used directly with minor modifications, which are mainly to the traffic estimation and congestion aspects. It is used however with a far greater number of users in the installed base than in Section 4: typically, the additional user will be an increment to an existing base of many millions. The supplier usually publishes a list of generic service bundle offers. Each bundle has consumption ceilings of the generic services, contract duration and a unique bundle price. Let the bundle that is closest to the user’s wish (targeted situation) be the initial public offer for the negotiation. The closeness is determined by the Euclidean distance from the offer to the user’s target point. If the user’s target point has equal distance to two offers, he randomly picks one offer. Negotiation algorithm The negotiation algorithm is similar to the algorithms introduced before, and the user and the supplier use the same decision rules. The main difference is that in the generic service bundle negotiation, the user’s constraints, maximization and decisions are all in the same service design space as the supplier.

637


Case results The key difference with current practices as described before is that now there is no need to choose between discrete bundles of the same basic services (e.g. voice, SMS, MMS, data), but the user negotiates instead with a supplier a tailored bundle meeting his individualized usage, price and duration demands. This means that unused service quotas because of discrete bundle design and choices can be vastly reduced and that the user does not pay for things he does not use. Also a dynamic adaptation can take place, as the user can adapt his service demand ceilings each time a new contract is renegotiated (e.g. a contract for the holiday months and another one for the usual working months). Furthermore, the user also can avoid tying himself up to long fixed durations like one or two years; however, he may still choose to do so if the supplier has a built-in incentive in the offer. In the analysis below we eliminate the effect of a possible upgrade or supply of a new mobile terminal to the user. This does not change the approach, as the possible amortization of terminal costs borne by the supplier, or other costs for the same terminal borne by the user, are traffic and usage independent. The supplier’s offer consists of three generic services: voice, SMS and data (typically Web access). Let the decision variables for an individual bundle of such generic services be X = [x1, x2…x5], and each corresponds to an individualized bundle design attribute: x1, max voice minutes for contract duration (discretized) x2, max number of SMS for contract duration (integer) x3, max amount of data download in MB for contract duration (real value) x4, contract length in months (integer) x5, price in € paid to the bundle for contract duration (discretized). Table 4: An operator’s public offer list of generic mobile service bundles (terminal not included) (Vodafone, 2006). Price plan

Max. voice minutes

Max. number of SMS

Max. data download (MB)

Contract length (month)

Price per month in €

Total contract cost excluding excess usage / €

1

75

100

0.5

12

20

240

2

200

150

0.5

12

30

360

3

500

200

2

12

40

480

4

800

250

4

12

50

600

5

1200

250

10

12

75

900

638


Based on the information retrieved Vodafone’s website (with no terminal included) on 3rd December 2006, we compile a public offer list. What needs to be emphasized is that these discrete bundles give the possibility of only an acceptance or rejection decision by the users, and the contract duration is also a supplier condition with no choice. User’s constraints The user' constraints for an individualized generic service bundle are mainly of two types: the budget constraints, and the time constraints (e.g. he cannot spend all his time talking on the phone). There are also boundary constraints which are the user’s region of tolerance regarding the value of a design attribute (e.g. a user wants between a and b voice minutes as max ceilings). Supplier’s constraints The supplier has constraints on the price he charges for the contract duration, i.e. the contract revenue. He may also want to differentiate incentives between the different service categories, by giving better treatment e.g. to data traffic vs. voice traffic. He may also want to add in contract duration-based incentives by offering cheaper voice, SMS and data download when the user is willing to sign a longer contract. Furthermore, his offer in each round of negotiation should not be radically different from the user’s request. All in all, the supplier will have to better adapt the service mix and usage request to his own technical and operational efficiencies, as well as making better use of his competitive advantages in these subjects. Numerical results The generic services are offered to a large population of users: their preferences for the service design attributes follows a distribution, which could be found out by a user survey. What needs to be pointed out is that in reality, in a competitive environment, user surveys may be of limited use, as survey results are only valid until a competitor undertakes a new campaign, which happens frequently. Due to lack of user survey results of generic service usage and conditions, we assume the user preferences for "voice minutes", "data download" and "price per month" to follow a multivariate normal distribution. Although real user preference distributions can only be conjectured to differ from it, the means/variances of the user preference multivariate normal distribution are approximated by the means/variances between the discrete bundles offered by a supplier (Table 4). It should be stressed that in practice the assumption of a multivariate normal

639


distribution is far from correct, but it is used here for lack of a better and simpler option. We assume the user’s preference for contract length/ and SMS usage to be independent of other design attributes and that each of them follows a normal distribution. The mean and variance of the preference distribution for SMS are approximated from the public discrete bundle in Table 4. For the mean and variance of the preference distribution for contract duration, we use the data obtained from the global user survey presented before (mean =7, standard deviation =5). We randomly generated some user preferences obeying the distributions described above. The generated values are mostly monthly based due to the nature of data in Table 4. We approximate a user’s total demands of each generic service by multiplying the generated user preference values by the generated individual contract length. Table 5: Individualized user requests, initial fixed bundle offers, and negotiation results of individualized generic service bundles. User a

b

c

d

Service design attributes x1 x2 x3 x4

x5

Utilities user supplier

Results user supplier

Request Public

3486 6000

1140 2400

18 24

6 12

264 480

0.92

103.86

win

lose

EQ Request Public

4235 10656 14400

1400 1488 3000

20.7 108 120

7 12 12

329 708 900

0.59

41.82

0.54

19.21

win

win

EQ Request Public EQ Request Public EQ

11700 2295 6000 1900 5873 9600 5600

2990 1160 2400 960 1393 3000 1295

117 25 24 14.6 56 48 51.1

13 5 12 4 7 12 7

754 195 480 164 399 600 434

0.65

37.42

0.51 0.86

103.86 19.06

win

lose

0.54 0.89

2.15 60.23

win

lose

We feed the randomized user preferences into the prototype implementation of the negotiation algorithm described before. We assume the subscriber base N = 106. We set the number of sub-intervals for the continuous variables in X, which is a discretization related parameter, to 5. We set the supplier’s "minimum profit threshold" θ to 0 (Figure 1). The supplier will not sign a contract if the incremental profit from the individual user is below θ. Some randomly selected individualized users' requests (a, b, c, d), public offers and negotiation results are shown in Table 5. Note that the public offer is selected from Table 4 as the one

640


Supplier utility: expected incremental profit from the user

that is closest to the user’s request. The value of each service attribute is shown for the total contract length. 80

40

6

4

60

2

7 5 3

20

public offer

0

-20

-40

initial point user operator final point 1

-60

0.35

0.4

0.45

0.5

0.55

0.6

0.65

0.7

- (distance to target point)

User utility: e

Figure 2: The Stackelberg negotiation game between a user and a supplier on a generic mobile service bundle

In the negotiation game, the user maximizes his utility by minimizing the distance to his target point. The supplier maximizes his utility, which is the incremental profit from that user. When there is a game equilibrium, which means an agreement is reached, we compare the utilities of the user and the supplier at the final equilibrium with their initial utilities from the corresponding initial offers. The user always achieve gains; the supplier in some negotiations makes less profit than in the public initial fixed bundle situation, but nevertheless, he still makes a positive incremental profit in all three negotiations. If we decrease the threshold θ, the supplier may have losses from some users. This will be discussed in the next chapter. The negotiation process of user-b is shown in Figure 2. The result is a win-win situation for both the user and the supplier.


641

Open Research Issues Mass customization support tool The users want to be passively or actively involved in service personalization and tariff negotiation, but in all cases it is necessary for the supplier to have a set of tools to automate the process. A tool should function either in a simplified mode without much involvement by the users (e.g. users only provide simple service feature, price and contract duration wishes) or in an interactive mode where users actively participate in the service design personalization and negotiation. Furthermore, the tools should all provide risk assessments for the supplier when selecting different negotiation parameters and decision criteria. We have developed and implemented a prototype of such tool (Pau and Chen 2006) but it is not yet complete. The open issues are:

How to implement the tool in a complex operational environment coupled to the OSS and rating engines;

How to integrate the tool seamlessly into the supplier’s service provisioning infrastructure;

How to provide customer support to users of the mass customized service bundle once contracted.

User behavior From the survey result, still half of the respondents prefer a flat rate for all the services; it will be valuable to find out the main reason behind this preference. We have done some basic research regarding the "simplification" concerns of the users when they do not select a flat rate. The solution is a "service design space" for the supplier and a "perceptual space" for the user, the latter being a reduced mapping of the former (Pau and Chen 2006). However, there is no immediate solution for what should be included in the design elements so that the perceived attributes of a service meet the user’s demands. The perception of individual services and tariffs as a cost reduction / life style / productivity improvement feature implies different user behaviors during the service configuration and tariff negotiation process. Our previous study used a coarse method by assuming all the users have a similar form of utility function, or that their preferences belong to a common utility distribution (Chen and Pau 2006). Further research can be done by identifying different utility functions for different categories of users.

642


Churn Given the averaged ideal contract length for a mass customized service bundle and individual tariff (i.e. about 7 months), what still needs to be investigated is the specific churn rate for users of such individual services and tariffs alone, once compounded back to a monthly churn rate for such users, over a number of cycles of individual tariff contracts. This analysis should clarify how much smaller the specific churn rates would be for these specific users than the observed ones under the two scenarios before. Some early reports from industry (Wieland 2006) show that by allowing very preliminary levels of personalization, operators can already greatly reduce churn. Dynamics The above analysis is static in the sense that it does not reflect radical market and pricing innovations and their effect on how individual tariffs would operate. This would be quite hard to do anyway ex-ante. One such a case would be if an operator offers an unlimited service usage at a very low bundle price. The computational framework above, both allows to respond to such an offer, but also to estimate how large a risk such a competitor takes in so doing. He may win subscribers but serve them all at a loss. Risk There are a lot of distribution uncertainties to be addressed when using individual tariffs:

Uncertainty that comes from user preferences;

Uncertainty that comes from user characteristics, which are hidden and driven by social issues such as affordability;

Uncertainty that comes from the supplier’s service provisioning cost model due to lack of data and technical system failures. Classic statistical analysis such as value at risk (VaR) may not provide good estimate of events with small probabilities but big impacts. Currently, we are trying to embed extreme value theory into the tools, which may safeguard the supplier from such risks.

Extensions to other fields The above tools and methods could be extended to other types of services where mass customization reduces wasted ressources while increasing loyalty and value to the end customers. Potential such application areas include public transport


643

tariffs on user-selected repetitive routes, electronic media subscription plans (like in book clubs), mass customized insurance (across health, property and savings), and investment services (mixing transactions and advice on specific user-defined classes of questions). Conclusions This chapter contributes to the research about mass customization of services by introducing the methods and models which can be applied in the mass customization of mobile communications services and tariffs. The analysis of the generic service bundles above demonstrate the scalability of such results to bases of N=1 Million users. The model-based quantitative analyses show that the adoption of personalized service bundles and tariffs can be beneficial to both the users and the suppliers. As the end-user survey has already indicated, there is a critical mass of the population who is willing to adopt mass customized services and tariffs, it is up to the suppliers to take the challenges to bring it into reality. In consequence when mobile operators will offer better responsiveness to their customers in terms of operations transparency, as well as governance to them as stakeholders, the sustainable financial performance of these operators will gain from mass customized service bundles and tariffs. Acknowledgments The authors would like to express their appreciation to the 22 members of the international discussion group "Personalized pricing of mobile service bundles" as well as to the people and organizations that helped with the distribution of the survey, especially the "International Telecommunication Users Group (INTUG)". The authors also thank the respondents who spent their time filling out the survey. References Abou-Jaoude, George and Samuel Kung (2005). The science of fit in garment MC, in 3rd Interdisciplinary World Congress on Mass Customization and Personalization. Hong Kong. Au, Yim Lee Emily and R. S. Goonetilleke (2005). Comfort characteristics of ladies' dress shoes, in 3rd Interdisciplinary World Congress on Mass Customization and Personalization. Hong Kong. Chen, H. and L-F Pau (2006). Individual tariffs for mobile communication services, in 16th Biennial Conference of the International Telecommunications Society, Tingjie Lu and Xiongjian Liang and Yan Xu (Eds.). Beijing: ITS Chen, H. and L-F Pau (2007a). Individual tariffs for mobile services: Analysis of operator business and risk consequences, in International Conference on the Management of Mobile Business (ICMB'07), Norm Archer and Khaled Hassanein and Yufei Yuan (Eds.). Toronto, Canada: IEEE Computer Society.

644


Chen, H. and L-F Pau (2007b). Individual Telecommunication Tariffs in Chinese Communities, in Global Mobile Commerce: Strategic Perspectives and Implementation Cases Wayne Huang and Y. L. Wang and John Day, Eds. Hershey, USA: Idea group. Chen, H. and L-F Pau (2007c). Personalized pricing of mobile service bundles: Survey design & key findings. Rotterdam: Rotterdam School of Management. Franke, Nikolaus and Frank Pille (2004). Value creation by toolkits for user innovation and design: The case of the watch market, Journal of Product Innovation Management. 21 (6): 401–15. Gabriel, R., M. Gersch and P. Weber (2006). Mass customization as an adequate strategy for educationservices, in World Conference on E-Learning in Corporate, Government, Healthcare, and Higher Education, T. Reeves and S. Yamashita (Eds.): Chesapeake, VA: AACE. Geddes, R. (2000). Public utilities, in Encyclopedia of Law and Economics, Geerit De Geest and Boudewijn Bouckaert, Eds. Vol. III The Regulation of Contracts. Cheltenham: Edward Elgar. Grenci, Richard T. and Charles A. Watts (2007). Maximizing customer value via mass customized econsumer services Business Horizons. 50 (2): 123–32. Huang, E. Y. and C. Y. Lin (2005). Customer-oriented financial service personalization, Industrial Management & Data Systems. 105 (1): 26–44. Lampel, J. and H. Mintzberg (1996). Customizing customization, Sloan Management Review. 38 (1): 21– 30. Noam, E. M. (1983). Telecommunications regulation today and tomorrow. New York: Law & Business. Ogawa, S. and F. T. Piller (2006). Reducing the risks of new product development, MIT Sloan Management Review. 47 (2): 65–71. Pau, L-F (2002). The communications and information economy: Issues, tariffs and economics research areas, Journal of Economic Dynamics and Control. 26 (9–10): 1651–75. Pau, L-F and H. Chen (2006). Individual tariffs for mobile service bundles: A negotiation calculation tool in 2006 IEE/ACM Intl conference on e-Business ICE-B 2006, Joaquim Filipe and Thomas Greene (Eds.). Setubal,Portugal: INSTICC Press. Prahalad, C. K. and V. Ramaswamy (2004). The future of competition: Co-creating unique value with customers. Boston, MA: Harvard Business School Press. Sigala, M. (2006). Mass customization implementation models and customer value in mobile phones services: Preliminary findings from Greece, Managing Service Quality. 16 (4): 395–420. Simon, Herbert A. (1957). A behavioral model of rational choice, in Models of man. New York: Wiley. von Hippel, Eric (2005). Democratizing innovation. Cambridge & London: The MIT Press. Wieland, Ken (2006). The customer retention challenge, in Telecommunications International Vol. 40. Winter, R. (2001). Mass customization and beyond–evolution of customer centricity in financial services, in International NAISO Congress on Information Science Innovations (ISI'2001), M Sebaaly (Ed.). Dubai, United Arab Emirates: ICSC Academic Press. Xu, Y. and D. C. Pitt (2002). Chinese telecommunications policy. Norwood: Artech House. Zeithaml, V. A., A. Parasuraman and L. L. Berry (1985). Problems and strategies in services marketing, Journal of Marketing. 49 (2): 33–46.


645

Author Biographies Hong Chen obtained his PhD from Rotterdam School of Management. His research focused on individual tariffs and service personalization. He received a MSc degree (cum laude) in computer science in 2003 from University of Twente, the Netherlands. He studied Telecommunications Engineering in Beijing University of Posts and Telecommunications from 1995 to1999. From 1999-2001, he worked in Huawei Technologies. Since August 2008, he had joined Altran CIS as a consultant specialized in telecom business. Prof. L-F Pau is Professor of Mobile business at Rotterdam School of Management, Adjunct Professor of Mobile business at the Copenhagen Business School, and Technology Director/CTO at L.M.Ericsson. Earlier he has been CTO for Digital Equipment Europe, and on the faculties of Danish Technical University, E.N.S. Télécommunications (Paris), M.I.T, and University of Tokyo. He is or has been on the Boards of IEEE Standards, OMG, RIO. Besides other areas of work in high tech, his work on mass customization spans since the early 1980’s mass customization in mechanical CAD /CAM, personalization of wireless services, and customization technologies in semiconductors. Contact: [email protected]

1.3

Unraveling the Service Innovation Dilemma: The Promise of Network Embeddedness Ikenna S. Uzuegbunam Gatton College of Business & Economics, University of Kentucky, USA

Satish Nambisan Lally School of Management & Technology, Rensselaer Polytechnic Institute, USA Manli Chen Lally School of Management & Technology, Rensselaer Polytechnic Institute, USA

Service innovation, unlike product innovation, is not easily scalable in the production process. In general, as firms attempt to grow, one potential direction for growth is through a firm’s ability in applying the same processes and resources used for a single unit of production to larger volumes, thus saving costs through economies of scale. Whereas product-oriented companies can easily achieve cost reduction through scientifically tested and validated operational processes for product innovation, the peculiar nature of services do not permit such formal processes to be applied to service offerings. In this chapter, we identify network approaches through which firms engage the "service innovation dilemma" — the problem of diseconomies of scale in a world of increasing demand for services. We argue that firms can develop sustainable competitive advantage in services through "real" and "virtual" embedded inter-firm and customer co-creation (market) mechanisms. We conclude with some implications for theory and practice in services and related innovations.

Introduction The notion that service innovation is significantly different from product innovation has long been established. In general, service innovation differs from product innovation in terms of diseconomies of scale in service innovations, intangibility and perishability of service offerings, simultaneity between production and consumption during service innovation, co-production of services with customers (e.g., Nambisan 2001; Rust and Chung 2006). At the core of the disparities between product and service innovation is an overarching productionlevel dilemma for service-oriented firms. The present chapter focuses on this "service innovation dilemma" in service production firms — the problem of diseconomies of scale in an increasing service 646


647

economy. The notion of service innovation dilemma is similar but distinct from the description of the innovator’s dilemma (Christensen 1997). While the innovator’s dilemma implies that established firms are focused too much on their current customers, thereby missing new opportunities from niche markets; the service innovation dilemma refers to the following paradox: the inability of firms (on their own, without their networks) to achieve economies of scale in services under current conditions where the increasing integration between products and services and digital connectivity presents many opportunities to increase scale economies. We investigate this problem because among the factors stated previously that differentiate service innovation from product innovation; nonscalability of services can profoundly affect the ability of service firms to earn economic rents as they innovate for their customers. Put differently, the inability of a firm to cost effectively scale up its service offerings to meet the heterogeneity of their customers' rapidly changing demands may impair the firm’s competitive advantage in the service economy. As a consequence of this, service firms often transact with other firms and even with customers to deliver service innovations to the customer (Vries 2006). In fact, firms can embed themselves in social attachments that can create behavioral expectations that are different from the atomistic view of market transacting (Uzzi and Lancaster 2003). The underlying norms for these attachments tend to be on the basis of trustful, cooperative behavior that results in mutual exchange between both parties. Firms that have a number of these embedded exchanges can increase their ability to innovate in service environments. Hence, in the rest of this article, we concern ourselves with "how the embeddedness of firms within two main aspects of their networks can enable them develop customizable service innovations for their customers?" Background Our study comes at the heels of recent developments in the rapidly growing service related sector in many developed economies around the world. In some instances, it is estimated that the service sector accounts for approximately 70-80 percent of the gross domestic product (GDP) in these countries (Chesbrough and Spohrer 2006; Rust and Chung 2006; Vries 2006). Coupled with the exponential growth of the service sector in these macro environments is an equally significant change in the demand microenvironment for service related offerings. Specifically, on the one hand, customers (whether individual consumers or businesses) do not want more choices, but more precise choices (Gilmore and Pine 2000). This is because excessive number of choices often times confuse them and present

648


challenges to make informed decisions. They often know what, when, where, and how they want it. On the other hand, there exists significant service opportunities for firms that are strategically engaged in service innovations because of technological advancement, socio-economic changes, coupled with a simultaneous rise in the idiosyncrasies of customer demands. It is plausible that the concomitant change in the supply and demand conditions for services is based on shifts in the socio-economic environments (Tidd, Bessant and Pavitt 2001). These shifts place an overwhelming impetus on managers to create sustainable value through innovating in networked and interconnected forms (Prahalad and Ramaswamy 2003). As commented by foremost evolutionary economists "… there is now sufficient evidence on the role of networking in innovation to postulate that the typical pattern of nineteenth-century innovation (the inventor-entrepreneur) and of the twentieth-century innovation (the in-house corporate R & D department with good external communications) is now increasingly giving way to a pattern of networking collaborative systems of innovation in the twenty-first century" (Freeman and Soete 1997). The consequence of this increasing economic relevance of services for firms and customers in the networked economy is one that is non-trivial and which deserves more research attention. To address this issue, the present chapter applies a network embeddedness framework in understanding how firms are able to achieve "economies of repetition", a reasonable proxy for economies of scale (Davies and Brady 2000) under a mass customization regime. In this vein, we develop a theoretical framework that incorporates a network embeddedness rationale in explaining innovation success in a service economy riddled with sophisticated demand conditions and associated problems of scalability. The basic premise is that firms can gain competitive advantage in service innovation when they develop an embedded network of partnering firms and customers that enables the deployment of customized sets of service solutions repeatedly to different customers in their demand space. We, therefore, revisit embeddedness theory and elucidate previously unexplored aspects of the theory. We argue that the case of services provides us with a rich context to demarcate these new propositions of embeddedness. The rest of this chapter proceeds in the following way. In the next section, we discuss some fundamental reasons why service innovation differs from product innovation. Within this frame, we identify two recent socio-economic sources of disruption to service oriented firms that create the necessity for a network model of innovation. Next we offer an embeddedness model of innovation in services in


649

its full ramifications. Finally, we discuss some implications and caveats of our framework for the nascent literature on service innovation. A Dilemma in Service Innovation The changing landscape of industrial competition The competitive landscape for many organizations has been changing at an unprecedented pace. Recent technological advances in digitization and connectivity increasingly blur industry and firm boundaries and accelerate the release of ample flows of information as well as the potential of circulating capacity and competence among firms. In this way, many factors of production have become mobile as they can move from one firm to the next (Chesbrough 2003). This has fundamentally changed the ways that businesses use to compete against one another. Intensified global competition, shrinking profit margins and the pressure to be highly efficient in operations also impel firms to find alternative costeffective and sustainable strategies for developing innovative offerings. Our study identifies two significant socio-economic shifts in the firm environment that lead to a service innovation dilemma. The first shift is the increase of complexity and convergence in products and services, as well as in their design and production (Gomes-Casseres 1994; Nambisan 2001). In most cases, the contemporary offering is an embodiment of several specialized skills that integrates across product and services. This is regardless of whether this offering is either business-to-business or business-to-customer; either technology or nontechnology based. Consider the example of the services provided by the healthcare system in the United States. The attendance, evaluation, diagnosis, treatment, referral, and follow-up of a patient and his/her ailment are processes that involve varying degrees of specializations and different levels of product and service components. Thereby, such a patient may need to undergo treatment processes that often times cannot be described as strictly product or service. Another good example is IBM’s recent migration to the on-demand strategy—a diversification strategy that moved IBM from the computer hardware industry into the business services consulting realm. This integration of products and services occurred because of IBM’s realization of the available and profitable market space in services and integrated solutions. Thus, the demand conditions for service and product innovation are not clearly distinct from each other, and may co-exist. Further, some arguments have been made that we exist in an intertwined environment; an environment characterized by an overarching demand for experience innovation, which can be an embodiment of both product and services (e.g., Prahalad and Ramaswamy 2003; Nambisan 2001).

650


The second major shift responsible for this dilemma in service innovation is the digitization of technologies and the consequent rise of the Internet (Prahalad and Ramaswamy 2003; Tidd, Bessant and Pavitt 2001). The prevalence of digital technologies creates both new opportunities and challenges for both product and service companies. The changing nature of product and service distribution to customers is one such example. Offerings that were typically distributed "offline" through physical stores now face disruptions from online models of these offerings. For instance, traditional book distribution vendors who previously sold books off the shelves of bookstores, now have powerful facilities online that enable them to reach and serve their customers globally. These developments in this industry were necessitated by the evolution of new online industries occupied by new entrants such as Amazon.com. Digitization of technologies also entails that firms need to reconfigure their production processes. For instance, IBM’s transition to "on demand" business strategy now offers its customers customizable services solutions at the front end of their respective innovation processes. The new configuration at IBM involves an e-business strategy that depicts virtual work and marketplaces. In such a scenario, one can conceive of ubiquitous value co-creation where suppliers and customers would meet and conduct business transactions around the clock in a networked environment (Farhoomand 2005). The aforementioned socioeconomic shifts are by no means the only changes that create problems for service firms. Also, the intensity of regulatory involvement in a particular sector (Hobday 1998) can lead to substantial changes in the ways firms can successfully innovate. According to Gomes-Casseres (1994), constellation strategies prevail because of the necessity to link markets, combine skills, and build momentum, reduce costs and to share risks. Therefore, in our model, we propose that firms can overcome these socio-economic shifts and succeed in service innovation through creating embedded networks of mutually benefiting actors, which ultimately leads to successful mass-customized service innovation. What is the dilemma? The neoclassical Penrosian mass-production perspective argues that one obvious avenue for a firm’s growth rests on the ability of the firm to apply the exact same processes and resources used for a single unit of production to larger volumes, thus saving costs through economies of scale and scope (Penrose 1959). As such, whereas product-oriented companies can easily achieve cost reduction through scientifically tested and proven methods for developing and improving products, the peculiar nature of service offerings do not permit such formal processes to be


651

applied to service development (Thomke 2003). In service development processes, "envisioning, energizing and enabling" capabilities is identified as one of the fundamental and critical aspects for new service innovation (e.g., Menor, Tatikonda and Sampson 2002). As customers become more mobile and their needs evolve to be more complex, it becomes increasingly difficult to meet every single customer’s needs fully by any firm in and of itself. Again, consider the case of the healthcare system. Many customers (i.e. patients) of the U.S. healthcare system have idiosyncratic needs that often times require a variety of inputs from various healthcare professionals and service providers. Typical problems associated with managing the U.S. healthcare often stem from the exorbitant costs that hospitals incur when critical and scarce resources such as medical personnel are diverted to unforeseen or ambiguous circumstances. The reason for this sort of ambiguity or staff shortage could be partially due to the fact that every patient demands a different type of treatment compared with other patients. Consequently, operations and workflow is not as clearly defined as in mass-production regimes as to how to efficiently deal with the idiosyncratic needs of thousands of healthcare consumers. Lapses in the hospital service production floor could easily exacerbate the costs of healthcare for each individual customer. Essentially, service providers' capabilities of being flexible and responsive to these needs are critical for success in the increasingly competitive landscape. As a result, designing appropriate service permutations as standard, but customizable service offerings for customers in an "on demand" fashion would be of strategic benefit to service firms. In the emergent service economy, competition will increasingly center on personalized co-creation experiences, resulting in value that is truly unique to each individual (Prahalad and Ramaswamy 2003). In order to make the mass customization and continuous innovation in the service sector a "win-win" situation in the increasingly dynamic environment, firms have to strive for new strategies and sources for innovation through partnerships with other firms that have complementary resources and competencies embedded in their respective established networks (Rothaermel 2001; Tripsas 1997). Such multi-partner collaborations have become normative in some industries leading often to group versus group competition. In most cases these networks are constructed to reap the rewards of economies of scale (Gomes-Casseres 1994). An example from the airline industry can be used to illustrate this point further. The alliance network comprising of Swissair, Delta, Singapore and SAS airlines was brokered at a time where these companies needed to increase their scale in airline bookings on transatlantic and European-Asian flights (Gomes-Casseres 1994). The impetus to collaborate in

652


services or even in the product area does not exist ab initio but is likely as a result of substantial changes in the globalized economy. The key in the airline example above is that firms need to be strongly linked to other firms in a way that trustful interactions and unique value can be created within the network of firms. These sorts of embedded relationships are critically important for competitive advantage because many of the firms involved in the same networks are loyal to other firms in them and will be more likely to forebear risks on behalf of the other firm(s). Further, they may also be willing to offer their distinctive factors of production only to firms within their embedded network thereby offering these firms a distinctive (unique) advantage over competitor firms (Uzzi 1997). Similarly, collaboration can also be established between firms and their customers. Quite often, interfirm collaboration and collaboration with customers is not distinguishable. In these cases, the customer of the service or product could be a firm (as opposed to an end user) who allies with the focal firm in question to develop an architectural innovation. Architectural innovation refers to innovation that involves changing the relationships between component innovations. This notion has been elucidated by Henderson and Clark (1990) and was initially suggested by Abernathy and Clark (1985) in their classification of innovation. This paper acknowledges that architectural innovation can encompass product components as well as service components. The web of relationships that firms often get entangled in, in their quest for architectural innovation does not often allow for a clear demarcation between a partner that is a firm in contrast to a partner that is a customer. At other times, the customer can clearly be demarked as the end consumer of the services such as in the case of a patient in a healthcare system, a passenger on an airline, or an individual customer at a local bank. However, firms are faced with the challenge of establishing constant and close interactions with their customers. This issue of connectivity is important for advancing customer coupling in innovation processes beyond the programmatic level. The Internet and communication technologies provide an opportunity for firms to establish sustainable connections with their innovation partners. Therefore, one can conjecture that networks can be either "real" based on offline interactions or "virtual" based on interactions facilitated by the Internet and other digital technologies. Considering these different dimensions of embeddedness in service production we develop our framework of network embeddedness as it applies to service innovation, which accounts for "real" and "virtual" embedded relationships between a focal firm and other partnering firms in their networks as well as


653

between the focal firm and their customer networks. Overall, our thesis suggests that firms can enhance economies of repetition by engaging optimal embedded configurations of both their "real" and "virtual" environments for innovative and efficient service production. In addition, the framework takes into account that some firms may employ both "real" and "virtual" embedded relationships simultaneously in network-centric relationships (Figure 1).

Figure 1: The effects of embeddedness on service innovation.

A Network Model of Embeddedness in Service Innovation Real embeddedness Network embeddedness in this study mainly refers to the extent to which a social community (e.g., firms and their customers, or firms and their firm-partners) operates on the basis of shared norms of cooperation, trustful interaction, and "untraded interdependencies" as distinct from competitive, individualistic, "arm’s length exchange" and hierarchical norms (Cooke 2001). Further, the argument of embeddedness can also be viewed as one, which supposes that the behavior and economic action of social units and institutions are constrained by cognition,

654


social structure, other institution and culture (Uzzi 1997) and more broadly by ongoing social relations (Granovetter 1985). Thus, the deeper a firm is embedded in a set of useful relations with other firms and its customers, the more likely the firm gains and optimizes the external resources required from these entities to develop innovative service solutions to its customers. Since the customers have idiosyncratic needs, the greater the heterogeneity of inputs available to the firm within its cumulative embedded networks, the higher the probability that the firm will be able to successfully integrate a meaningful set of service offerings to the customer. However, it is also important to acknowledge that network embeddedness does not always have a positive effect for the embedded firms (Uzzi 1997). For instance, if a network of embedded firms grows to include co-opeting firms (i.e. firms that simultaneously collaborate and compete), it is possible that the effect of embeddedness will diminish at this point. Therefore, our framework as shown in Figure 1 also accounts for negative effects as well as limitations of embeddedness on a firm’s ability to be efficiently innovative in developing various service offerings. Virtual embeddedness The mainstream network embeddedness literature has typically investigated interorganizational relationships without considering if there are significant differences in the effects of the medium through which these relationships are brokered. This is especially important if we consider how the Internet has changed the face of the industrial competition. Recent accounts suggest that virtually embedded ties have multifaceted value that often times complement real (socially) embedded ties (Fowler, Lawrence and Morse 2004; Lawrence, Morse and Fowler 2005; Morse, Fowler and Lawrence 2007). For instance, Morse et al. (2007) argue that virtual embeddedness enables entrepreneurial establishments to overcome the liabilities of newness usually associated with new ventures. In this study, we take the occurrence of the service innovation dilemma to elucidate how virtual embeddedness can enable innovation. For instance, "In 1996, IBM announced its e-business strategy, which mapped the future of the company as well as the industry, in general. The new strategy suggested that all companies would operate within virtual marketplaces" (Farhoomand 2005). Such virtual interconnections between IBM and its partners can enable IBM design and develop easily customizable service solution in a modular fashion. We do not foresee that virtual embeddedness will have a negative effect on economies of repetition after a certain level of embeddedness. This is because of


655

the network externalities effect that exists in virtual environments. The externalities amplify substantially as the number of participants increases without bounds because of the absence of human element in these interactions. However, a firm that employs both "real" and "virtual" embedded networks in service innovation is likely to encounter the inverted U shape effect mainly associated with the influence of real network embeddedness on innovation success. Customer co-creation in experience innovation Creating an engaging consumption experience in a personalized way for customers has become an increasingly critical strategy that companies use to differentiate their products or services from those of the other players in the market and keep them from the "commoditized trap"; instead, charging premiums from satisfied customers. Goods and services are no longer enough. Experiences are the foundation for future economic growth (Pine and Gilmore 1999) and value creation through profitable growth can only come from innovation (Prahalad and Ramaswamy 2003). Different from the conventional traditional firm-centric and product-centric perspective, Prahalad and Ramaswamy (2003) propose that the next practices of innovation must shift onto experience environments – supported by a network of companies and consumer communities — to co-create unique value for individual customers. Customer involvement in innovation process is multi-faceted. Firms can engage customers in the innovation process in almost all the stages of the innovation process including (1) the ideation stage of innovation (2) co-creation of the ideas (3) product testing prior to market introduction and (4) providing end user support for product users (Nambisan 2002). This customer engagement in innovation is not passive in nature; it usually involves active involvement of the customer in the process of innovation, which, often times, creates difficulties for traditionally insular innovation teams in these firms (Nambisan 2002). As the customer becomes involved in this process, the success of innovation is not based only on customer involvement in and by itself; it is often times based on the links that connect customers to the focal firm as well as the links that connect the focal firm in question to other firms in their network. Therefore, the locus of innovation is neither firm centric, nor customer centric, it is usually network-centric (Prahalad and Ramaswamy 2003). Of the most cited reasons for under-utilization of customers in the innovation process is the lack of connectivity to the customers during new product development (Nambisan 2002). This issue of connectivity is important in order to advance customer coupling in innovation process beyond the rhetoric level.

656


Therefore, the application of information and communication technologies (ICT) in coupling customer perspectives has recently received deserved seminal attention (e.g., Nambisan 2002; Sawhney, Verona and Prandelli 2005). For example, Sawhney et al. (2005) describe how two notable companies, Ducati Motorcycles of Italy and Eli Lily Pharmaceuticals succeed in innovation based on their ability to involve the customers through virtual customer communities (VCC) of innovation. In a similar vein, IBM on-demand strategy in service innovation requires IBM’s ability to seamlessly integrate customized modules of service solutions through deploying a global community of interfirm innovators. The effect of virtual customer environments/communities (VCEs/VCCs) High technology advancement in telecommunications and digitization enables the fast development of various communities of interest for products and services. In particular, the Internet plays a critical facilitating role in resolving this longstanding dilemma in the production of service – the difficulty in achieving economies of scale in the production of services (Nambisan 2001). The firms that do well in the Internet era are those firms that exploit the potential of this technology in co-innovating with their customers. In the increasingly networked society, Virtual Customer Communities (VCCs) are playing more important roles in facilitating and shaping the deployment of distributed innovation models that involve varied customer roles in the innovation process (Prahalad and Ramaswamy 2000). As a result, organizations should carefully consider the nature of customer interactions that underlie particular roles and should incorporate appropriate Virtual Customer Communities (VCCs) design features to enhance the potential for customer value creation (Nambisan 2002). Success in leveraging customer knowledge depends on the firm’s systems and the processes that enhance the integrative capabilities of the firm (Kogut and Zander 1992; Verona 1999). Advanced technologies may represent a high level of connectivity and powerful tools to engage customers in the service innovation process; the effective deployment of the VCCs requires close attention to customer-firm interaction patterns and contexts. Specifically, in this computermediated and community-oriented setting, the pattern of customer interactions and the amount of value creation vary with the roles that the customers play and the nature of the mechanisms in this process. Furthermore, customers' motivations to be part of the VCCs and their anticipation of future interaction may potentially affect their behavior and efforts made into the VCCs, which inevitably impact the value that the network receives (Walther 1994; 1997).


657

In a service innovation context, VCCs can potentially generate valuable knowledge and feedback by integrating their experiences and knowledge about existing services/products with their knowledge about a potential usage or application context. Organizations can enhance their customers' abilities to participate in knowledge creation and their contributions to the firm’s network efficiency by increasing their awareness and knowledge about related technologies and networks with the potential uses, and complementary products and services (Nambisan et al. 1999; italics added). According to Leonard-Barton (1995), enhanced product/technology awareness will expand the boundaries of customers' cognitive processes and trigger innovative or creative ideas, which, in turn, facilitate better outcomes of a firm’s service innovation endeavors. Based on VCCs' facilitating role of knowledge transfer and knowledge conversion proposed by Nambisan (2002), we now propose it has an important role to play in our framework. Role of facilitating knowledge transfer One important model of knowledge acquisition is the network model, which indicates the role of person-to-person mode of knowledge transfer. It provides access to knowledge that resides within individuals through establishing direct links among people. Systems to support such network models of knowledge acquisition involve knowledge directories or knowledge maps that contain pointers to the knowledge source (i.e., people) — not the knowledge itself (Nambisan 2002). Role of facilitating knowledge conversion Two types of knowledge conversion identified by Nonaka and Takeuchi (1995) are closely related to a VCC context because they emphasize the process of enabling a firm and its customers to create and share knowledge related to a product/service and its use: (1) conversion of explicit knowledge to explicit knowledge (combination) and (2) conversion of tacit knowledge to explicit knowledge (externalization). Customers may synthesize new knowledge (e.g., new product features) by combining multiple explicit knowledge elements. Nonaka and Konno (1998) describe a "cyber ba" as the place of interaction in a virtual world that supports combining explicit knowledge with explicit knowledge to generate new knowledge. Thus, firms need to provide tools that aid customers in viewing and mapping multiple knowledge elements in textual or graphic form. For example, multimedia technologies enable customers to experiment with

658


different design choices or to understand the impact of different product architectures or features on usage (Dahan and Hauser 2000). Customer innovations that involve the conversion of tacit knowledge into explicit knowledge, however, underline the importance of the emerging notion of individual and distributed cognition systems (Leonard and Sensiper 1998; Sharda et al. 1999). For such innovations to occur, customers should be able to make multiple interpretations of a given product or technology, as well as exchange them with other customers or community members. A distributed cognition system supports such interpretation and dialogue among customers by providing richer forms of self-reflection and communication. This relates to the "interacting ba" described by Nonaka and Konno (1998), where individuals share mental models but also reflect and analyze their own. Such processes can lead to innovative outcomes that spring from the combined tacit knowledge base of the customer community. As noted earlier, the nature of knowledge acquisition and conversion varies with customer-NPD role. For example, in the first three roles (i.e., as resource, cocreator, and product tester), a repository model of knowledge acquisition assumes high importance, given the customers' focus on acquiring factual knowledge about a product or technology. In the product support role, while both repository and network models are important, the network model has more significance, given the focus on customer-customer interactions as the primary mechanism for knowledge acquisition. Similarly, it may be argued that in the initial two customer-NPD roles (i.e., as resource and as co-creator), there is significant focus on capturing customers' tacit knowledge about a product/service and the application context and making such knowledge explicit so that it can be used by the internal NPD team. Hence, there is greater relevance for VCCs features related to individual and distributed cognition systems, although features related to knowledge combination may also assume importance in particular contexts. In product support, much of the customer interactions involve usage-related problem solving, and a vast majority of such problem solving requires knowledge creation through the combination of explicit knowledge. In summary, given the important role and impact of VCCs on the efficiency of the firm’s networks and ultimately the quality of the service innovations generated, we argue that VCCs influence the relationship between the embeddedness of a firm and the competitive advantages of its service innovation. Therefore, companies have to pay close attention to provide the appropriate level of support


659

and develop important elements in this process to enhance the potentials for customer value creation and service innovation success. Implications This chapter offers an elucidation of the role of an organization’s social context in the ability of a firm to successfully implement a customized service innovation strategy in the contemporary economy. In particular, we suggest that serviceoriented firms should be assessed in terms of their abilities to successfully repeat solutions as opposed to the dominant logic of "economies of scale". Also, because service-oriented firms are characterized by project driven relationships (Davies and Brady 2000; Nambisan 2001), it is important that they are embedded in interfirm relations that can easily execute and replicate projects under different scenarios. A firm employing such embedded relationships should also have effective knowledge management mechanisms in order to derive valuable learning across projects in their experience space. Theoretical contributions and implications Our framework presents a new perspective of service related innovation in relation to idiosyncratic customer demands and experiences. The primary theoretical insight from our model provides a comprehensive view of the role of network embeddedness in enhancing a firm’s likelihood of success in a serviceoriented economy. Although, we conjecture that the peculiar nature of services is the reason why economies of scale in service innovation are difficult to be accomplish, our model demonstrates that services rarely exist in isolation from products. Consequently, the suppositions of framework in figure 1 are farreaching to the evolving literature on service innovation as well as the burgeoning academic press on product innovation. In essence, our model implicitly suggests useful implications that are applicable to the literature in both domains. Furthermore, this paper advances the nascent literature on service innovation at the mass-customized level through an explicit analysis of the impact of network theory in these discussions. We integrate the marketing literature on customer involvement in innovation with the popular press in interfirm network innovation strategies to provide a demand and supply perspective of competitive advantage in a service oriented economy. Managerial implications Managers are under increasing pressure to search for the next (or best) practices for innovative actions to enhance competitive advantages through making their

660


customers more satisfied with their service experiences. The theoretical framework developed in this article attempts to provide some insights on how service firms can compete more effectively by integrating customers' insights, enlisting them as key participants, and utilizing the complementary resources and competencies of its network partners into the value creation chain and process. In other words, managers should start modifying their long-standing firm-centric business thinking and equip themselves with a shift from a perspective of exploiting customer knowledge by the firm to a perspective of knowledge co-creation with the customers (Sawhney and Prandelli 2000) due to the changing dynamics between the firm and its customers. With this updated mindset and by managing the customer-firm interactions efficiently, firms can innovate and establish powerful and high-quality network environments and expand their capacities to fulfill a variety of customer needs and consumption experiences via highly adaptive and flexible capabilities and expanded resources enabled by embedding within the compatible networks. In summary, the following practical implications suffice from our article:

New approaches are suggested in which firms can navigate their networks in the process of creating value through customized service innovation.

Some prescriptions are made on how firms can employ technologies in the process of value creation in services. Specifically, we suggest that knowledge generation, transfer and conversion in service innovation can be further enhanced through superior levels of deployment of virtual customer co-creation environments, and subsequently enables firms to effectively fulfill the idiosyncratic customer needs and further improve the firm’s competitive advantages.

Conclusions The role of networks in innovation is undeniably well established. However, the nature and effects of specific aspects of networks are still not well understood (Ahuja 2000). The trend of mass customization in both the product and service worlds forces the entire business networks to redefine their strategy in terms of the type of capability to improve to most effectively serve their customer base. The world of service innovation is not excluded from the "network effect", and the changing customer demands lead to a changing business network, caused by all the strategic actions taken by the individual organizations. This paper attempts an integration of user induced network effects on service related innovation with the strategic management of this innovation process. We offer a combination of


661

supply and demand perspectives in examining an increasingly dominant phenomenon in organizational environments and competitive landscapes — the cocreation of value between firms and customers. We deduce that a firm’s position and embeddedness in its business network and customer co-creation network, as well as its virtual environments serve as effective network mechanisms for establishing competitive advantage and enabling value creation in innovative service offerings in the new service economy.

References Abernathy, W. J. and Clark, K. M. (1985). Innovation: Mapping the winds of creative destruction. Research Policy. 14: 3–22. Ahuja, G. (2000). Collaboration networks, structural holes and innovation: a longitudinal study. Administrative Science Quarterly. 45: 425–455. Baldwin, C. and Clark K. B. (1997). Managing in an age of modularity. Harvard Business Review. 75(5): 84–93. Chesbrough, H. (2003). The era of open innovation. MIT Sloan Management Review, Spring issue: 35– 42. Christensen, C. M. (1997). The innovator’s dilemma. Boston: Harvard Business School Press. Cooke, P. (2001). Regional innovation systems, clusters and the knowledge economy. Industrial and Corporate Change. 10(4): 945–974. Davies, A. and Brady, T. (2000). Organizational Capabilities and Learning in Complex Product Systems: Towards Repeatable Solutions, Research Policy. 29(7): 931–953. Djellal, F. and Gallouj, F. (2001). Patterns of innovation organization in service firms: Portal survey results and theoretical models. Science and Public Policy. 28(1): 57−67. Edvardsson, B., Haglund, L. and Mattsson, J. (1995). Analysis, planning, improvisation and control in the development of new services. International Journal of Service Industry Management. 6(2): 24−35. Farhoomand, A. F. (2005). IBM’s "On demand business" strategy. The Asia Case Research Center, The University of Hong Kong, Ref. 05/257C. Fitzsimmons, J. A. and Fitzsimmons, M. J. (2000). New service development: Creating memorable experiences. Thousand Oaks: Sage. Fowler, S. W., Lawrence, T. B. and Morse E. A. (2004). Virtually embedded ties. Journal of Management. 30(5): 647–666. Freeman, C. and Soete, L. (1997). The economics of industrial revolution. 3rd Edition. Cambridge, MA: The MIT Press. Gallouj, F. and Weinstein, O. (1997). Innovation in services. Research Policy. 26(4–5): 537−556. Gilmore, J. H. and Pine, B. J. (2000). Markets of one. Boston, MA: Harvard Business School Press. Gomes-Casseres, B. (1994). Group versus group: How alliance networks compete. Harvard Business Review. 72(4): 62–74. Granovetter, M. (1985). Economic action and social structure: The problem of embeddedness. American Journal of Sociology 91(3): 481–510.

662


Henderson, R. M. and Clark K. (1990). Architectural innovation: The reconfiguration of existing product technologies and the failure of established firms. Administrative Science Quarterly. 35: 9–30. Hobday, M. (1998). Product complexity, innovation and industrial organization. Research Policy. 26: 689–710. Johne, A. (1993). Insurance product development: Managing the changes. International Journal of Bank Marketing. 11(3): 5−14. Johne, A. and Storey, C. (1998). New service development: A review of the literature and annotated bibliography. European Journal of Marketing. 32(3/4): 184−251. Kogut, B. and Zander, U. (1992). Knowledge of the firms, combinative capabilities and the replication of technology. Organization Science. 3: 383–397. Lawrence, T. B., Morse E. A. and Fowler, S. W. (2005). Managing your portfolio of connections. Sloan Management Review. 46(2): 59–65. Leonard-Barton, D. (1995). Wellsprings of knowledge. Boston: Harvard Business School Press. Lievens, A. and Moenaert, R. K. (2000). Project team communication in financial service innovation. Journal of Management Studies. 37(5): 733−766. Menor, L. J., Tatikonda, M. V. and Sampson, S. E. (2002). New service development: Areas for exploitation and exploration. Journal of Operations Management. 20(2): 135−157. Morse, E. A., Fowler, S. W. and Lawrence, T. B. (2007). The Impact of virtual embeddedness on the new venture survival: Overcoming the liabilities of newness. Entrepreneurship Theory & Practice. March: 139–159. Nambisan, S., Agarwal, R. and Tanniru, M. (1999). Organizational mechanisms for enhancing user innovation in information technology. MIS Quarterly. 23: 365–395. Nambisan, S. (2001). "Why service businesses are not product businesses. MIT Sloan Management Review. 42(4): 72–80. Nambisan, S. (2002). Designing virtual customer environments for new product development: Toward a theory, Academy of Management Review. 27: 392–413. Nijssen, E. J., Hillebrand, B., Vermeulen, P. A. M. and Kemp, R. G. M. (2006). Exploring new product and service innovation similarities and differences. International Journal of Research in Marketing. 23: 241–251. Prahalad, C. K. and Ramaswamy, V. (2000). Co-opting customer competence. Harvard Business Review. 78(1): 79–87. Prahalad, C. K. and Ramaswamy, V. The new frontier of experience innovation. MIT Sloan Management Review. 44(4): 12–18. Rothaermel, F. T. (2001). Incumbents' advantage through exploiting complementary assets via interfirm cooperation. Strategic Management Journal. 22: 687–699. Rust, R. T. and Chung, T. S. (2006). Marketing models and relationships. Marketing Science. 25: 560– 580. Sawhney, M. and Prandelli, E. (2000). Communities of creation: Managing distributed innovation in turbulent markets. California Management Review. 42(4): 24–54. Sawhney, M., Verona, G. and Prandelli, E. (2005). Collaborating to create: The Internet as a platform for customer engagement in product innovation. Journal of Interactive Marketing. 19(4): 4–17. Shimizu, S., Ishikawa, H., Satoh, A. and Aihara, T. (2004). IBM Journal of Research & Development. 48(5/6): 751–764.


663

Thomke, S. (2003). "R&D Comes to Services: Bank of America’s Pathbreaking Experiments." Harvard Business Review. April: 71–79. Tidd, J., Bessant J. and Pavitt, K. (2001). Managing innovation: Integrating technological, market and organizational change. Chichester, England: John Wiley & Sons Ltd. Tripsas, M. (1997). Unraveling the process of creative destruction: Complementary assets and incumbent survival in the Typesetter Industry. Strategic Management Journal. 18: 119–142. Uzzi, B. (1997). Social structure and competition in interfirm networks: The paradox of embeddedness. Administrative Science Quarterly. 42(1): 35–67. Uzzi, B. and Lancaster, R. (2003). Relational embeddedness and learning: The case of bank loan managers and their clients. Management Science. 49(4): 383–399. Verona, G. (1999). A resource-based view of product development. Academy of Management Review. 24: 132–142. Vries, E. J. (2006). Innovation in services in networks of organizations and in the distribution of services. Research Policy. 35: 1037–1051. Walther, J. B. (1994). Anticipated ongoing interaction versus channel effects on relational communication in CMC. Human Communication Research. 20: 473–501. Walther, J. B. (1997). Group and interpersonal effects in international computer-mediated collaboration. Human Communication Research. 23: 342–363.

Author Biographies Dr. Ikenna Uzuegbunam is an Assistant Professor in Management at Gatton College of Business & Economics, University of Kentucky. He earned his doctorate in management from Rensselaer Polytechnic Institute (RPI). He also holds a Bachelor of Engineering degree with high honors from the University of Nigeria, Nsukka and a Master of Science degree in Technology and Innovation Management from the University of Sussex. His research interests focus on topics related to network-centric strategies that foster innovation and entrepreneurship in high technological environments. He was recently a Visiting Scholar at the Center for Technology Management, Cambridge University. His industry and consulting experience spans a number of industries including medical devices, oil & gas, electronics, software development and the non-for-profit sector. Contact: www.gatton.uky.edu | [email protected] Dr. Satish Nambisan is a globally-recognized researcher and thought-leader in the areas of innovation management and technology strategy. He has done pioneering research work in the areas of network-centric innovation, customer co-innovation, and IT-enabled product development. His research publications have appeared in several premier management journals including the Harvard Business Review, Management Science, Academy of Management Review, and MIT Sloan Management Review. His new book The Global Brain: Your Roadmap for Innovating Faster and Smarter in a Networked World was published by the Wharton School Publishing in October 2007. Dr. Nambisan is an associate professor of technology management & strategy in the Lally School of Management at Rensselaer Polytechnic Institute. He has held visiting faculty appointments

664


at the at the Kellogg School of Management, Northwestern University and at the Institute for Entrepreneurship & Innovation, Vienna University of Economics and Business Administration, Vienna, Austria. Contact: www.rpi.edu/~nambis | [email protected] Manli Chen is a current doctoral candidate at Lally School of Management and Technology at Rensselaer Polytechnic Institute in Troy, New York. Her research focuses on innovation management that has an emphasis on user involvement and consumer dynamics in social media as well as methods to enhance consumers' attitude formation and adoption of innovative products and services. In addition to presenting research papers on various topics of customer engagement and innovation adoption at academic and industry conferences, she is a co-author of the Handbook of Qualitative Research Methods in Marketing. Contact: [email protected]

1.4


Ximena Hernandez Institute of Biomechanics of Valencia, Universidad Politécnica de Valencia, Spain R. Lahuerta Institute of Biomechanics of Valencia, Universidad Politécnica de Valencia, Spain M. J. Such Institute of Biomechanics of Valencia, Universidad Politécnica de Valencia, Spain J. Navarro Institute of Biomechanics of Valencia, Universidad Politécnica de Valencia, Spain A. López Institute of Biomechanics of Valencia, Universidad Politécnica de Valencia, Spain C. Soler Institute of Biomechanics of Valencia, Universidad Politécnica de Valencia, Spain S. Redondo ER&SI, Spain

The EMOCIONA initiative has demonstrated that the application of Emotional Design Techniques can serve to improve the design of surrounding settings used in the point of sale of habitat-related products, and to measure to which extent a retail store’s background has influence in the willingness to purchase a product. Around eighty people had participated of this experience through a pilot emotions measurement laboratory in which the emotional profile of the users was determined and its purchase attitude was registered, in order to extract concepts associated not only with the piece being evaluated, but also to establish the influence that different scenarios had in their perception of it. The results showed an increase of the purchase intention of a piece of furniture when it was settled in users favorite environment, and also the emotional component the "desired environment" concept responds to in terms of style, global preferences, et cetera. Furthermore, the results had permitted the corroboration of the possibilities that the utilisation of Emotional Engineering techniques, as part of User Oriented Development methodologies, have in the area of personalization of a retail store environment; allowing its adaptation for a determined population and the effective communication of a brand image.

665

666


Introduction Consumer behavior in relationship with a product has been studied mostly focusing on how the users' evaluate the intrinsic characteristics that a product has. Nowadays, store interior design and ambiance are being studied, and the buying experience is subject of tremendous attention among marketing strategist and retail managers as a way to differentiate from the competition and to improve customer loyalty. But literature is scarce on how the ambiance that a product is set on influences consumers preference and purchase intention. For some products the setting that surrounds them may not be among the key factors leading to purchase, but that may not be true in the case of furniture shopping. Shopping for furniture can be regarded as a complex process for the consumer due to the amount of money invested in its purchase and the uncertainty of how the item purchased it is going to fit on the consumers' habitat. For example when buying context independent products, such as wrists watches or mobile phones, the choice is made based on consumer’s preferences, previous brand experience and word of mouth recommendations, but when the situation implies choosing a product that is going to be used in a determined set, the user is faced with the effort of abstracting in order to "see" how it is going to fit in the home decoration and available space. According to some researchers (Regueiro et al. 2003), the purchase experience has an implicit stress component associated with the decision making process. As the selection of products to choose from increases, the visualization of the outcomes gets more and more complex, therefore the possibility of consumers negative affect raises (Luce 1998). Taking in consideration that, for every piece of furniture the buyer is faced with a myriad of possibilities and that "not knowing all the options" may delay purchase, it is important to make the right kind, and amount, of information available for the consumer in the channels that he or she may use as help in making that decision. The purchase process has been characterized as a set of steps that the consumer takes when shopping: problem recognition, information search, evaluation of alternatives and product choice (Solomon et al. 2002). Problem recognition is characterized for being the stage at which the consumer perceives a difference between a present situation and a desired or ideal situation. The information search phase begins when the person decides to change that present state into the desired one. When the search phase is finished, then the consumer is ready to evaluate the options and select the one that befits the selection criteria. Despite the increasing use of the Internet as a source to gather information and to buy consumer goods, in the specific case of furniture the store remains as the key search, selection and purchase channel in the mind of the consumer (Lihra et al.


667

2007). Although information is important mainly on the search and evaluation phases, consumers want to see and touch the piece of furniture before deciding on buying it (Lihra et al. 2007). As a result, the store is the key point that manufacturers have to interface with its clients, so special attention must be paid on how to design the communication and presentation of a product to suit customers' expectations. According to Reardon and Mac Corkle (cited by Lhira et al. 2007) in using a distribution channel, consumers consider primarily time versus psychological benefits e.g. like having a pleasurable shopping experience. In the experience economy, shopping is an activity that increases perceived product value as it is regarded as part of an experience that goes beyond the mere act of purchasing goods (Gobé 2001). Nowadays it is highly accepted that consumption is associated with hedonic pursuits such as fun and pleasure (Holbrook et al. 1982) and that every product carries a symbolic meaning (Levy 1980), therefore to design a retail environment, the meanings and emotional implications of the consumer are essential. Moreover, making the shopping experience seamless and enjoyable is very important in the specific case of the purchase of a complex product taking into account that furniture shopping can be overwhelming and frustrating. In order to provide the consumer a fulfilling shopping experience, the complexity level must be taken to minimums not only by providing information on the product, but by really helping consumers to make a choice adapted to both his, or hers, emotional and functional needs. To truly accomplish this, the diversity in users' style and aesthetic preference must be taken into consideration. Therefore the need for a customized design of the settings surrounding a piece of furniture can be a solution as to how to attract a wide range of users, not by personalizing the product, which can be expensive for the manufacturer and difficult to accomplish by the user, but by presenting the piece of furniture in a setting that matches users' preferred decoration style. Antecedents and Objectives of the Present Study As specified earlier, there are several factors of key importance that influence users perception of a piece of furniture, whereas price, perceived quality, brand image, and advertising campaigns are important, the store remains as the main point of contact between consumer and product. Adding to this, the uncertainty that the difficulty to imagine the outcomes entitles, and that the shopping experience has a hedonic component, the fact of facilitating the information

668


needed by the customer at the point of purchase is important to provide a pleasurable shopping experience. Previous research on the subject of perception of furniture confirmed that the presentation of a piece of furniture influenced the perception that the consumer had on it (Such 2004). In that particular study, users' evaluated office furniture in both formats of presentation; in a neutral environment and in a realistic environment, showing that some of the emotional factors correlated with the purchase intention were stimulated in a stronger way when the piece of furniture was presented in a realistic environment. One way to establish consumers shopping experience is measuring customer satisfaction. Nevertheless, several authors point out that the measurement of consumer satisfaction is a weak indicator regarding consumer loyalty or willingness to pay (Barlow et al. 2000; Chaudhuri 2006), so measurement of other indicators is necessary. To establish the degree of influence that the environment in which a product is settled has in consumer’s decision making process, an experimental experience was designed: the EMOCIONA initiative. This endeavour was developed to gain knowledge about why the way a product is displayed evokes certain emotional responses that can affect positively or negatively the willingness to purchase of a potential consumer. In the present study the emotional perception of a piece of furniture on the retail environment was measured and the purchase intention was used as an indicator of the degree of influence that the emotional factors have on customers' willingness to buy. Design of Experiment It is highly known that previous experience, brand image and store reputation, are some of the factors that influence the choice of a product to buy (Herr et al. 1991). Furthermore, it is highly known that customers' choice of a product among others is based on emotional components mainly based on likes and dislikes, so: Can the perception of a product be modified in a way that positively influences the purchase intention? As the objective of the evaluation was to measure the emotional factors involved in user perception of a product in the retail environment, it was decided that the experimentation should take place where the shopping actually occurs: on the store. This offered a double advantage, first the buying experience was measured in its "natural" environment, and second, the actual clients of the store were the users participating in the study.


669

As personal style and preference take a big part in deciding which piece of furniture to buy, a first approach of in store profiling was made by classifying users in one of two categories:

Design Users.

Classic Users. This classification of the users allowed to establish the user profile as the control variable of the experimentation and allowed the comparison between profiles. To establish which profile does the users belonged, a profiling questionnaire was designed, in which users were asked about his or hers style preferences. Acknowledging the fact that users may be innovative in some of their consumption habits and classic on others, complementing the questionnaire with information related directly with users' furniture consumption preferences was necessary. As a way to approach this, users were asked to browse through a catalogue of products of the same style as the piece of furniture participating in the experiment, and then asked if they would buy something from that style or not. The catalogue used was the one provided by the manufacturer and reproduced furniture of the same style as the one taking part in the study. The experimental variables are related to the objectives and can be summarized as this:

To measure users perception of the product in a neutral context.

To measure users perception of the product in six different contexts or scenarios.

To measure purchase intention.

Stimulus design and experimentation For conducting the described research, a series of elements were necessary in order to set up an emotion’s measurement laboratory. A furniture retail store (Figure 1) placed in the centre of the city was the chosen setting, where several scenarios representing different styles of interior design were build. To ensure the affluence of people several promotional actions were taken. Six Scenarios were designed, all of them as background of the same console table (Figure 2). The scenario zero, also called the "neutral scenario", was settled on the first floor of the store, whereas the other ones were on the second floor. The design and style of the scenarios was selected according to the main design styles detected in the city were the experiment was going to take place. In this way, the range of styles represented was high enough so users could feel represented by at least one of the scenarios designed.

670


Figure 1: Furniture retail store placed in the centre of the city.

Figure 2: Object of the experimentation.

For the Scenarios 1 to 6 other elements were used as follows:

Flooring (wooden, ceramic tiles, porcelain, et cetera)

Back wall (brick, concrete, color painted, wall paper)

Illumination fixtures (table lamps, wall fixtures)

Ornamental elements (vases, figurines, et cetera)

In the next section, each scenario is briefly described and pictures showing concept design and final results are showed. Neutral scenario: The neutral scenario was implemented as a way to establish the perception of the customers of the table in itself, without any other stimulus but the store regular decoration, in this way the degree into which the other scenarios influenced user perception could be evaluated. This scenario was composed of the table displayed on the store without any context surrounding it. Minimalist scenario: This scenario was composed of the table in a minimalist setting (Figure 3). Decoration, light and the materials used where those of a minimalist style, predominating the whites and a few elements of ornamentation. Modern scenario: This scenario was designed taking into account the main elements reflecting a modern style. Light wooden floors and red walls were used to create a sensation neutral modernity (Figure 4). Design scenario: This scenario used high contrast and shines as the main elements associated with a setting with a high component of design (Figure 5). Classic scenario: The classic scenario’s main characteristics were its wooden floors and soft lighting, and cream colored walls (Figure 6).


671

Neo-Baroque scenario: This scenario’s main characteristic were its sumptuous wall paper, ornaments and light fixtures with shapes inspired in nature (Figure 7). Ethnic scenario: This scenario was designed using earth colors, natural materials and ethnic ornaments (Figure 8).

Figure 3: Minimalist scenario (conceptual design & real scenario).

Figure 4: Modern scenario (conceptual design & real scenario).

672


Figure 5: Design scenario (conceptual design & real scenario).

Figure 6: Classic scenario (conceptual design & real scenario).

Figure 7: Neo-baroque scenario (conceptual design & real scenario).

673


Figure 8: Ethnic scenario (conceptual design & real scenario).

Experimentation To insure the obtainment of feasible data from the experimentation two participation modes were set; although similar, they allowed the investigators the obtainment of data that permitted the evaluation of the whole experimental designed stimulus: One group of users evaluated the table in the neutral scenario. Another group of users evaluated the table in their favorite scenario. Also, a group of users were asked to evaluate all the scenarios. For the evaluation, three different questionnaires were designed using a list of twenty-one emotional terms obtained by the Semantics Differential Methodology (Osgood et al. 1957) on a five point Likert type scale. The emotional terms used were validated in previous studies related to emotional evaluation of furniture (Such 2004). Figure 9 shows the list of terms used to evaluate perception of the table and of the scenarios. ELEGANT

VANGUARDIST

OF DESIGN

INFORMAL

IMPERSONAL

ETHNIC

SOBER

CLASSIC

HOMOGENEUS

COSY

PRACTIC

ROBUST

INNOVATIVE

SIMPLE

LUMINOUS

EXOTIC

MINIMALIST

COMFORTABLE

MODERN

SPACIOUS

Figure 9: Emotional terms used in the experimentation.

674


Before the evaluation of the scenarios, users were asked if they would buy the table by showing them a catalogue were the table was depicted in a neutral scenario. In this way two groups of users evaluated the table; those willing to buy the table and those unwilling to buy it. This would allow researchers to establish if the purchase intention was modified by the contexts in which the table was presented.

The first group of users, who evaluated the table in the neutral scenario, established the perception of the table in itself, this group was composed of users of both profiles; classic and design. And of users that were predisposed to buy the table and those who wouldn't buy it. This allowed the obtainment of two set of emotional profiles of the table, the one form users unwilling to buy the table and the one form users willing to buy the table.

The second group of users took a look around all the scenarios on the second floor, and was asked to evaluate the table in their favorite one. In this way an emotional profile of the table in the favorite scenario was established. At this instance users were also asked if they would want to have that piece of furniture at home.

Users Sample The target population of the study were citizens from a largely populated Spanish city. It is important to state that not selection whatsoever was made concerning the users participating in the study. Over a hundred participants took part of the study, but only the data provided by users that had completed all of the questionnaires, was the one used to complete the study. The final classification of users resulted in the following:

40 users predisposed to buy within the same style of furniture

40 users unwilling to buy a piece of furniture of the same style

Sample size was estimated under the criteria of having enough variability to test differences in products' emotional perception patterns when changing the surroundings. Simplified test for sample size estimation in F-test was used (Cohen 1977): n=

Ya +Yb 16 + 2 and d 0.05 = s d 0.05

Given the recommendations in the literature when using a 5-point Likert scale, and 1.06 as standard deviation for the emotional pattern of a furniture in a of 5point Likert scale (Such 2004), the sample size should be N=20. Hence, with our

675


sample of 40 participants for each profile we feel to have sufficient statistical power (80-90%) to detect significant differences.

Major Findings Users classification As a first step in the data analysis, users were classified into two groups according to their profile as: "classic users" or "design users". This classification was made according to observations and to the analysis of a survey on trend knowledge and fashion preferences. The purpose of this categorization was to differentiate users more predisposed to the furniture style from users less predisposed to it.

Emotional factors affecting user’s preference A first analysis of the obtained data was made to determine what emotional terms influenced the preference variable (Figure 10), based on the evaluations made by users about the table in the different scenarios. Classc profile users

Design profile users

Comfortable Cosy Elegant Sober

Homogéneus

Simple

Innovative

Clássic Figure 10: Shared emotional terms influencing the purchase intention variable for each profile as well as the emotional terms that differ for each one.

Results show that although some terms that influence the preference variable are common for both profiles, there are some terms specific for each type of profile. The shared terms comfortable, cozy and elegant can be regarded as the emotional functionalities that users demand despite of their profile, whereas the terms specific for each profile are aesthetic or style requirements related to the preferred decoration style.

676


Classic users' emotional profile Figure 11, shows the emotional terms stimulated by the Axil table, in the neutral scenarios for the classic profile users. The terms appearing in the horizontal line are the ones related to the preference variable for the classic profile users, and the values appearing on the vertical axe represent the value assigned to them on a 5point likert-type scale (being 2 completely agree and -2 completely disagree).

perception

Figure 11: Emotional profile of the Axil table.

Figure 12: Emotional profiles for the table in favorite and in neutral scenario.

Significant differences were found for the term elegant, which is more stimulated by the favorite scenario and is one of the determinants of the purchase intention (Figure 13).

677

VOLUME 2: APPLICATIONS AND CASES Test statistics a,b,c

Chi-square df Asymp. Sig.

Comfortable

Elegant

Innovator

Simple

Classic

Exotic

Purchase Intention

,102 1 ,749

4,050 1

,532 1 ,466

2,244 1 ,134

,002 1 ,965

,052 1 ,819

3,595 1 ,058

b.

,044

a. Kruskal-Wallis test Grouping variable: Scenario tipology (Neutral vs. favourite) c. User classification = Classic

Figure 13: Kruskal Wallis test (classical users). For the classic users, only one emotional concept (Elegant) shows significant differences between the neutral scenario and the favorite one.

Design users' emotional profile A Kruskal Wallis test shows that only the "exotic" emotional concept presents significant differences between the neutral scenario and the favorite one; for the design users (Figure 14). Contrast statistics a,b,c Comfortable Chi-square df Asymp. Sig. d. e. f.

1,752 1 ,186

Elegant ,241 1

,623

Innovator

Simple

,117 1 ,732

1,611 1 ,204

Classic 2,340 1 ,126

Exotic 4,141 1 ,042

Purchase Intention ,901 1 ,342

Kruskal-Wallis test Grouping variable: Scenario tipology (Neutral vs. favourite) User classification = Classic

Figure 14: A Kruskal-Wallis test shows that only the "exotic" emotional concept presents significant differences between the neutral scenario and the favorite one; for the design users (Figure 9).

Modification of user’s purchase intention Once the emotional aspects demanded by each one of the groups were established the question about the modification of the users' purchase intention could be studied in detail. The first step was to verify that the inclusion of the selected table in a modified environment changed the purchase intention in order to quantify that

678


fact. As emotional concepts depend on the user’s profile, the variation of the purchase intention was studied according to the user profiles determined earlier. A non parametrical analysis on the emotional terms valuated and on the preference variable (Would you like to have this piece of furniture at home?), showed which variables represented meaningful changes in the evaluation of the table in the neutral scenario against the evaluation of the table in any of the favorite scenarios by a classic or a design user. For the design user, we found significant differences were detected when evaluating the table in a neutral scenario or in a preferred scenario. The difference was verified by the use of this emotional term: exotic. In this particular case as none of the concepts had influence on the purchase intention, no significant changes were observable in the preference variable (Figures 14 and 15). For the classic user evaluating the table either in the neutral scenario or in a preferred one, a significant variation of the emotional term elegant was found. This concept presents a significant influence on the purchase intention. According to this it can be inferred that an increase of the valuation related to the elegance concept causes an increase of the purchase intention on the users responding to this profile (Figures 13 and 16).

Figure 15: Contingency table, showing scenario (neutral or favorite) vs. purchase intention ("Would you like to purchase this furniture element?") for the design user profile.

Figure 16: Contingency table, showing scenario (neutral or favorite) vs. purchase intention to the classic user profile.

It is important to highlight that the contextualization of a piece of furniture in a context appealing to the user modifies purchase intention. The profile of the users was detected to influence the preference variable in relation to which emotional terms were more important to each user’s profile. Although for both profiles the purchase intention was modified, a greater impact of the contextualization was found for the classic profile users.


679

Scenarios preferred by the user An emotional profile was elaborated for the six scenarios. The scenarios preferred by the users were: the modern scenario, the classic scenario and the design scenario (Figure 17 and Figure 18).

Figure 17: The scenarios preferred by the user.

According to a characterization of the scenarios consistent with the emotional terms evaluated by the users, it can be determined which concepts are associated to each of the favorite scenarios. The modern scenario is associated to the term "elegant", the classic scenario to the term "simple", and the design scenario is perceived as innovative.

Discussion Research to date has shown that retail environments evoke emotional responses on consumers (Darden et al. 1994; Machleit et al. 2000), and that this affects their perception of the retail experience. One of the contributions of this study is the assessment of the influence that the retail environment has on the purchase intention of a habitat product. Also, it permitted the corroboration of the fact that personal preferences play a key role on how a piece of furniture is perceived depending on how it is displayed. If the background matches consumer’s preferred (or liked) style, the perception of the product changes and it can modify the purchase intention, in the case of this particular study the increase detected was of a 29.5%. Therefore, the background of a piece of furniture can be designed according to the preferences of the specific public that buys in a store, as a way to enhance the user’s purchase experience.

680

HANDBOOK OF RESEARCH IN MASS CUSTOMIZATION & PERSONALIZATION User classification Clasificación usr

Would likegustaria to have this furnitureenatsu home? 95%you IC ¿Le tenerpiece este of ambiente casa?

2

Design Diseño

Clásico Classic

1,8

1,6

1,4

1,2

1

2

3

4

5

6

Escenario

Figure 18: The preference variable in relation to the scenario evaluated for both user profiles.

The experience showed that the process of adapting a retail environment to its client’s preferences could be accomplished by means of an emotional study enclosed in the specific clientele of a particular store. However, future lines of research must consider the consumer experience as a whole and as such, it must be studied taking in consideration all the different components of it.

Future Research The future lines of research are oriented to the promotion of this family of methodologies in which the user orientation is a constant through the different phases of the product development process. As the experience resulted to be confirmative of the influence that the emotional response to an environment has in the purchase intention, future researches will attempt to reach a deeper understanding of what personal preferences participate on the consumption experience. This experimentation design is considered to be a first step towards closing the gap between the concepts of a Living Lab focused on IT products, that involve the


681

user carrying the tested artefact, and a Living Lab that allows getting closer to the user of products that are not portable without modifying its natural behavior. The exploration of these methodologies, with the help of a Living Lab paradigm, is one of the next steps of the investigation, as this helpful tool will allow the study of the user in real life conditions.

References Barlow, J. and Maul, D. (2000). Emotional Value: Creating Strong Bonds with Your Customers. San Francisco, CA: Berrett-Koehler. Chaudhuri, A. (2006) Emotion and Reason in Consumer Behavior. Burlington, MA: Elsevier. Cohen, J. (1977). Statistical Power Analysis for the Behavioral Sciences. Academic Press, London. Cooper, R. G. (1982). New Product Success in Industrial Firms. Industrial Marketing Management. 11(2): 215–223. Darden, W. and Babin, B. (1994). Exploring the Concept of Affective Quality: Expanding the concept of Retail Personality. Journal of Business Research. 29(2): 101–109. Gobé, M. (2001). Emotional Branding. Allworth Press, New York. Herr, P., Krades, F. R., Kim, J. (1991). Effects of word-of-mouth and product attribute information on persuasion: An accessibility-diagnosticity perspective. Journal of Consumer Research. 14: 454–462 (March). Holbrook, M. and E. Hirschman (1982). The Experiential Aspects of Consumption: Consumer Fantasies, Feelings and Fun. Journal of Consumer Research. 9(2): 132–140. Levy, S. (1980). The Symbolic Analysis of Companies, Brands and Customers. Twelfth Albert Wesley Frey Lecture, Graduate School of Business, University of Pittsburgh, PA. Lihra, T., Graf, R. (2007). Multi-Channel communication and consumer choice in the household furniture buying process. Direct Marketing International Journal. 1(3): 146–160. Luce, M. F. (1998). Choosing to avoid: Coping with negatively emotion-laden consumer decisions. Journal of Consumer Research. 24(4): 409–433. Machleit, K. and Ergolu, S. (2000). Describing and Measuring Emotional Response to Shopping Experience. Journal of Business Research. 49: 101–111. Regueiro, R. and León, O. (2003). Estrés en desiciones cotidianas. Psicothema. 15(44): 533–538. Solomon, M. (2002). Consumer Behavior: Buying, having, being. Prentice Hall, Tornto. Such, M. J. (2004). La ingeniería Kansei como modelo de simulación de fenómenos de la percepción, aplicación en el sector del mobiliario de oficina. PhD Thesis. Univ. Politécnica de Valencia. Valencia.

Author Biographies Mss. Ximena Hernandez is a Product Designer, working as a Junior Researcher at the Biomechanics Institute of Valencia in the User Friendly Design Section, where she has participated in over 10 European and National Projects. She was an assistant professor at the Ergonomics Department at the Centre of Industrial Design (Montevideo, Uruguay) for

682


two years. Since her graduation thesis in 2005, her research has focused on the application of qualitative methodologies for the assessment of user experience. Contact: www.ibv.org | [email protected] Eng. Mr. Rubén Lahuerta Martínez is Telecommunication engineer by the Polytechnic University of Valencia. He is the coordinator of the User Centered Methodologies Group at the Biomechanics Institute of Valencia and leads a multidisciplinary group of professionals in the development of new technologies and methodologies focused on capturing user needs to help in the development of user-centric products and services. He has vast experience in diverse European and National Projects in which he has performed tasks as researcher, experiment designer and lead technologist. Contact: www.ibv.org | [email protected] Dra. Mss. Maria José Such in an Industrial Engineer with a PhD in the application of Kansei Engineering for the design of successful user oriented products. She is the director of the User Friendly Design Section at the Biomechanics Institute of Valencia. She has participated in over 70 European and National Projects, with a strong focus on the application of Kansei Engineering and other methodologies for the assessment of User´s perception of Design. Contact: www.ibv.org | [email protected] Eng. Mr. José Navarro Garcia is an Industrial Designer, with vast experience in the development of innovative products. He has participated in over 15 European and National projects as a design expert. He is currently working in the Industrial Design Group as part of the User Friendly Department at the Biomechanics Institute of Valencia, a group focused on the integration of innovative user research methodologies into the product development process. Contact: www.ibv.org | [email protected] Ms. Amparo Lopez is a Social Worker with great experience in the application and interpretation of qualitative methodologies framed in the User Centered Design body of knowledge. She is currently working in the section of User Friendly Design at the Biomechanics Institute of Valencia where she has participated in over 50 European and National Projects, for public and private clients. Currently she develops tasks involving the selection and application of innovative methodologies for the assessment of user experience with products and services. Contact: www.ibv.org | [email protected] Mr. Serafín Redondo Quesada is an Industrial Designer with vast experience in home and office furniture design. He has also developed a successful career as a Interior Designer with projects in main Spanish cities like Barcelona, Pamplona and Valencia. He is the manager of two highly successful stores, specialized in designer furniture. Contact: www.er-si.com | [email protected] Dr. Mr. Carlos Soler Gracia is the Technological Services and Applications Department Director of the Biomechanics Institute of Valencia. He has participated in over 150 European and National Projects, as lead researcher in new technologies, for which he beholds 5 patents for technological. Contact: www.ibv.org | [email protected]

1.5

One Size Fits All, Made-to-Measure, and Bespoke Tailoring: Challenges in Building an Attractive Service Portfolio Hans Björkman Unionen, Sweden

Customer Relationship Management (CRM) can be described as aiming at creating shared interests as a means for building loyalty between organizations and their customers. An important issue is hence whether mass customization strengthens or weakens such relations. On one hand, the element of customer co-design in mass customization tends to strengthen the relations. On the other hand, mass customization builds upon a fixed or restricted design space and limited personal interaction and may thus result in weaker relations than individually customized services. Unionen, a major white-collar trade union in Sweden, provides a broad and attractive service portfolio consisting of standardized, mass customized, and individually customized services. Each individual member creates an individual service portfolio through: (i) Personal choice of information, activities and services. (ii) Mass-customized/customized services. Linkages between standardized, mass customized, and individually customized services will be discussed and the issue of building relations with members/customers will also be treated. The chapter concludes with a description of the results of the chosen strategy and a discussion concerning the possibilities to generalize from the experiences gained.

Introduction The increasing economic importance of services and the specific nature of services have put a stronger focus on service management. Customer Relationship Management (CRM) has been a popular managerial concept since the 1990s. A broad definition of CRM is that it concerns the management of the whole relationship between a firm and its customers. Thus CRM is seen as more or less a synonym to relationship marketing (Grönroos 2000). This notion is used to position an emerging marketing perspective that is different from an earlier more dominant perspective in the marketing literature. This earlier dominant perspective, which may be described as transaction marketing, is based on the exchange of ready-made value for money. Relationship marketing, on the other hand, may be described as a perspective based on cooperation in order to facilitate a mutual creation of value (Grönroos 2000; Sheth and Parvatiyar 1995). Thus, customers 683

684


should be treated as relational customers – once a relationship has been established, customers should be treated as customers even when they do not make any purchases (Grönroos 2000, p.34). CRM can thus be described as aiming at creating shared interests as a means for building loyalty between organizations and their customers. An important issue is hence whether mass customization strengthens or weakens such relations. On the one hand, the element of customer co-design in mass customization tends to strengthen relations, at least in comparison with non-customized services. On the other hand, mass customization builds upon a fixed or restricted design space and limited personal interaction (such as face-to-face meetings between service provider and customer) and may thus result in weaker relations than individually customized services. This chapter is based upon experiences gained in Sif, a major white-collar trade union in Sweden. In 2008, Sif merged with HTF and took the new name Unionen. Hence, the chapter also serves as an account of an evolving process in the newly created organization. Most trade unions are financed by monthly or annual subscriptions from members, while services are provided for free to members in need. The trade union membership fee covers a portfolio of services (unemployment benefit schemes, negotiations on national and industry-wise agreements, individual negotiations, legal advice and representation for members in need, career advice services etc), but the members do not necessarily use all the offered services. For most trade unions, the service package consists of both collective and individual services, some of which are highly individualized. Not only service quality, but also member relations and cost-efficiency are of utmost importance, as decreasing union membership is a trend in the Western economies. Accordingly, it is not surprising that trade unions are increasingly interested in mass customized services. Concepts from the relationship marketing literature provide important insights into current developments in white-collar unionism in Sweden. In the organizational context of unions, however, certain services are delivered collectively and members also have a relationship to the union which is political in nature presupposing various forms of participation. Unions have traditionally strived for participation from members – as individuals or elected representatives – in service provision processes as well as in internal trade union decision-making. The empirical basis of this study is gained through my experience from different roles in the organization; student organizer, business controller, service developer, strategist and senior researcher. For several years, I have conducted action research as an insider in the organization (Björkman 2005; Björkman and Huzzard 2005). I have also used academic research on Sif, conducted by various


685

researchers (i.e., Bruhn 1999; Huzzard 2000), as an additional source. My research methodology is inspired by collaborative autobiography (Goldman 1993), auto-observation (Adler and Adler 1994), and, in particular, reflexive ethnography (Ellis and Bochner 1996). The empirical starting point is thus my familiarity with the setting.

Trade Union Services and Mass Customization Theory A quick glance at mass customization research tells us that production and customization of physical goods has been the main focus. However, services may be even better suited to mass customization. If so, why? The specific properties of a service have been described in an array of definitions. The specific nature of services compared to goods may be referred to as intangibility, perishability, simultaneity and heterogeneity of services (Edvardsson and Gustafsson 1999). That the production and consumption of the service takes place simultaneously makes the customer involved in the production and consumption process (Grönroos 1992; Johne and Storey 1998; Normann 2000).

Service characteristics Thus, as services (at least in many cases, with "collective services" described below as an exception) are created in interaction between the provider and the user, the provider/user interface can be used as an arena for user involvement and customization. In some service contexts, this interface consists of face-to-face meetings between service provider and user. In other contexts, the provider may be represented by an automated service, often with high technology content (ATMs, Internet based travel agencies, Internet banking, etc.). Cagan and Vogel (2002) state that "A service is an activity that enhances experience; it requires an array of products to deliver its core activity". This definition indicates the obvious interconnectedness between services and products. It is useful to describe trade union services from the perspective of the service users:

Individual services, targeting the individual member (in this chapter divided into "made-to-measure services" and "bespoke services").

Collective services, different from individual services as they target all members or groups of members (in this chapter also called "one size fits all services"). Even though the distinction between individual and collective services is hard to draw in practice, the notions individual and collective services are analytically

686


useful in the trade union context studied here. Trade union services can also be defined in terms of relieving and enabling as one dimension of user involvement during service delivery (Normann 2000). Other distinctions can relate to degrees of standardization/personalization. Other important perspectives include physical proximity between the service deliverer and the user (from "face-to-face" to the Internet). Services are often said to be intangible, and collective services may in a way be even less tangible than individual services. As an example, lobbying activities are collective services that are intangible and often distant from the individual member. Nevertheless, these kinds of activities can be described as services. However, the service delivery process is different from what is common in relation to individual services, as individuals in the case of collective services have a more distant role and seldom participate in the service delivery process. The individual benefits from collective services may also be more difficult to trace and evaluate, as they may be distant both in terms of geography and time. The totality of the trade union service portfolio can be expressed in a very simple model, which I have found to be very useful in discussions with trade unionists (Figure 1). In this model, "participation" is treated as a distinct factor (or service), as members' participation is essential for all democratic trade unions. In other settings, this dimension could be replaced with "user involvement", "community building" or other similar notions. Participation

Collective services

Individual services

Figure 1: The trade union service portfolio.

The terms used in the model have been described in detail by Björkman and Huzzard (2005). A major advantage of the model, besides its simplicity, is that it does not just provide basic terms that can be useful for defining the union service


687

portfolio. The model can also be used to define the strategic position of the union, as this position consists of how the union acts and is perceived (and is perceived to differ from other organizations) by members, potential members, other trade unions, employers and governmental organizations – and this can be expressed on the basis of the three basic terms. As most trade unions are financed by monthly or annual subscriptions from members, and services are provided for free to members in need, pricing mechanisms do not help unions assess how members (or potential members) perceive the value of the different items in the service portfolio. However, pricing mechanisms (the level of the fee) play a role for evaluation of the offered service package as a whole made by members or potential members, compared to what is offered by other unions or other service alternatives (Björkman 2005). Hence, mass customization and personalization carry some specific challenges for trade unions (and maybe also for some other service providers, such as banks, insurance companies etc) connected to non-existent pricing mechanisms on a specific service level. This is, however, not an obstacle for mass customization and personalization. On the contrary, personalization and mass customization provide opportunities for gaining knowledge about users, letting users participate in the service creation process and delivering individual services of high quality to increasingly demanding users – and this can be done cost-efficiently. In this chapter, three metaphors will be used to describe the degree of customization of the services offered: "one size fits all" (standardized services, limited interaction between service provider and user), "made-to-measure" (mass customized services, a higher degree of interaction between service provider and user), and "bespoke" (fully customized services, well developed interaction between service provider and user). These will be further defined later.

Mass customization and personalization How do the trade union services studied fit the theoretical models provided by mass customization researchers? Piller (2005) has defined mass customization as a "Customer co-design process of products and services, which meet the needs of each individual customer with regard to certain product features. All operations are performed within a fixed solution space, characterized by stable but still flexible and responsive processes. As a result, the costs associated with customization allow for a price level that does not imply a switch in an upper market segment." The "made-to-measure" services provided by Unionen fall without problems into the definition given above, but have, as will be discussed later, some interesting

688


implications concerning costs and price. Is there a difference between mass customization and personalization? It has been stated that mass customization is about changing, assembling and modifying product and service components to fit the user, while personalization is about intense communication and interaction between supplier and customer (Piller 2007). From this it follows that personalization could be an important element of a mass customization strategy, implying fluent communication and interaction between supplier and user during the mass customization process. Personalization could, from this perspective, be perceived as a means for meeting two challenges connected to mass customization (Zipkin 2001):

Eliciting the customer’s wants and needs

Eliciting the customer’s willingness to pay for customization In the setting studied, personalization is the core strategy for the development of new individual offerings of information and communication. From the user’s (member’s) perspective, personalization hence means that:

The member indicates his/her areas of interest

The member gains access to specific information, activities, communities, etc due to the indicated interest areas

The member can improve the usefulness of information and services offered through interaction, assessments, and idea generation. From the organization’s perspective, personalization is:

A way of guiding members towards suitable services, in respect of both service area (i.e. insurances, advice on salaries, career profiles, etc) and in respect of service type: ("one size fits all", "made-to-measure", or "bespoke").

A means for collection of information to be used for further development of information, communication, and services.

Unionen: The Organization and Its Strategy Unionen is the leading white-collar trade union in Sweden. The union is independent in party-political and religious terms and organizes white-collar employees in the technical and knowledge-based sectors of the labor market. The members include managing directors, middle managers, engineers, accountants, business graduates, technicians, and administrative staff in different industrial sectors, ranging from basic industries such as steel and forestry to IT. All whitecollar workers at a workplace can thus be members of Unionen, irrespective of the


689

level at which they work (Huzzard 2000). The total number of members is 485,000 (2008). Just under 50 per cent of members are women. Unionen organizes over 40 per cent of the organization’s potential members. The remaining employees in the organization’s serviceable job territory have chosen another union or are non-members. Thus, Unionen operates in a competitive market with a large recruitment potential. Unionen can muster considerable resources both in terms of a large membership and financial strength to meet both threats and opportunities in terms of membership recruitment and retention. Unionen provides services to its members in all the traditional union areas, such as advice on working conditions and pay, financial support in the event of unemployment and other forms of insurance in connection with illness and so on. Unionen also supports its members if they become involved in disputes with their employers. These traditional services are still very important in terms of resources utilized.

A new strategy for personalization and mass customization Internal studies have indicated that different members and groups have many trade union needs and expectations in common. Studies have also shown that different groups differ from each other. More than this, it is evident that Unionen consists of about 500,000 individuals, each of whom has his or her specific needs and interests. Thus, it has been decided that a strategy should involve collective as well as individual needs and services. The main arguments for further development of activities and services targeting individuals or specific groups have been:

Perceived needs to face up to competing organizations which have been successful in their recruitment of specific professional groups

The experiences from the union’s own efforts aimed at specific target groups have been evaluated as being mainly successful. The treatment of members as individuals should thus result in opportunities for members to select the kinds of trade union information and activities they are interested in. The further description focuses on personalized communication and on different modes of service creation and delivery.

My Profile – individual creation of the member interface There are no average members! Members and potential members position themselves on many dimensions: for example, as a Manchester United supporter, engineer, mother of two, new on the job, manager, employed by Volvo, and some of these dimensions could be recognized by the union. Some dimensions may then

690


"come for free": there are local Union branches at all Volvo premises in Sweden, while the member himself or herself selects other dimensions of interest. Such a strategy implies that all members can be given opportunities to affiliate to groups sharing their interests. Unionen’s strategy for communication with members can be described as personalization through:

multi-dimensional segmentation and

individual choice. Multi-dimensional segmentation means that specific information, activities, web based communities etc are offered according to

profession/education (clerk, manager, engineer, project manager etc),

industry sector (transport, services, media, ICT, chemicals & pharmaceuticals, etc)

interest (such as gender equality, health and fitness, career development). Initially, 32 different profile choices (9 professional/educational, 12 industry sectors, 11 interest based) have been developed and offered to the members. Individual choice means that all members are invited to create his or her specific combination of profile choices. Figure 2 provides an example of one individual member’s communication configuration.

Engineer

Service Industry

My profile

Career development

Figure 2: Personalized communication through multi-dimensional segmentation and individual choice.

691


Hence, individualization and personalization of the membership service portfolio follow a two-step model:

Personal choice of information, communication, activities and services.

Mass-customized and fully individualized services.

The service portfolio: a mixture of one–size-fits-all services, made-to-measure services, and bespoke services. The next issue is to look a bit more closely at the services offered. Table 1 is useful for analyzing mass customization in a trade union setting. Table 1: Three kinds of services. Service type

Description

Examples

One size fits all

Often collective services, no adaptation to specific individual needs

Collective agreements, lobbying for new legislation

Made-to-measure

Individual, mass customized, services, often provided through the Internet: specific limitations due to technology etc.

Internet based interactive career advice tools

Bespoke

Individual services, often provided through face-toface meetings or telephone contacts

Advice from trade union officers on salaries to individuals, trade union officers representing individuals in negotiation with employers, individual career coaching

As will be discussed below, it is essential that the union service portfolio consists of all three kinds of services. The service portfolio configuration will look different for each individual member.

"One size fits all" services: still important Collective bargaining, lobbying and many other activities of a collective character – aiming at strengthening specific member groups or the whole collective of members – are still essential activities for Unionen. Collective services differ from individual services as they target all members or groups of members. Hence, the service provision process is different, as it does not necessarily involve the individual member. Among the more important kinds of collective services, the following could be mentioned:

692


Negotiations and collective agreements at European, national, industry, or company levels.

Participation in co-determination procedures (which could, potentially, be an individual service, as such procedures may concern a single individual)

Participation on governmental boards and committees, lobbying and involvement in public debates, advertising on general issues to advance trade union arguments in the public domain.

In the Swedish union context, collective services have not declined in importance during recent years, even though they have changed in character. These changes can be described in terms of rationalization (such as the utilization of new technologies), reduction (of participation in governmental committees), extension (such as international work and lobbying), and decentralization (such as a decentralized responsibility for bargaining in conflict situations – from national to regional levels).

Made-to-measure: mass customized services as the bulk of individual services It is not new for trade unions to provide services to individual members. On the contrary, this has been a core activity since their inception. Thus, the provision of help and guidance to members in need – groups or individuals – has a long tradition, and has always been given considerable attention and resources. Individual services with a long tradition in Sif include advice on salaries and employment contracts, advice on insurance as well as support and representation when individuals come into conflict with their employers. These services are still very important. However, new individual services have emerged and the service provision model has been altered in three respects:

Problem-solving when problems have occurred has been combined with services aimed at being delivered before acute problems occur, such as career planning activities, thus indicating a switch from a reactive to proactive service mode;

The design of the new services indicates a shift in the trade union role, from reliever – taking care of the member’s problems – to enabler – providing the member with the tools for self-action (Normann 2000).

New services are seldom dependent on local union representatives as they are mainly provided on the Internet or by Sif staff. This change in service configuration has emerged over a substantial period and is to some extent a reaction to increasing levels of competition between unions for members. Sif has a long tradition of organizing activities and services for specific


693

professional groups, such as engineers or managers. Projects have targeted specific professional groups (i.e. managers) as well as members and potential members in one specific industry sector (i.e. IT). The latter has aimed at recruiting new members through a more attractive membership package, developing new individual (often Internet based) services, and enhancing the union presence in established companies and start-ups in the sector. Shifts in values among the members (Bruhn 1999) have led to a perceived need to strengthen the position of every individual at the workplace. Not only have values among members shifted, but the labor market has become more volatile. Swedish trade unions have seldom positioned themselves as defenders of "every job", and the current Unionen strategy holds a strong element of making members employable and attractive on the labor market. Thus, the goal that all members should have safe and well-paid jobs has not changed, but the means are totally different than before. A few examples of new services can be specifically mentioned:

A specific income insurance only available for Unionen members, providing additional unemployment benefits in proportion to one’s individual income loss;

"The Unionen Career Centre": Internet based self-assessment and career development tools, free of charge for members;

"The Career Coach": members can, free of charge, meet a professional career advisor for a few hours;

"The Job Chat": an Internet based platform for discussions within different interest groups in the organization, "virtual communities". As stated earlier, trade unions (and Unionen) are facing problems in attracting members. The challenge is to provide an attractive membership package, targeting the individual and his or her needs – and the attractiveness of the membership is strongly dependent on the membership fee level. Provision of mass customized services may be a viable strategy to keep the membership fee attractive.

Bespoke services: when there is a need The prioritization of made-to-measure services should thus give Sif opportunities to provide individual services of high quality when there is a need. Such need can either be related to the individual (some members may not be able to use the mass customized services) or to the situation. More complicated issues, such as individual consultation on complex issues, negotiations concerning an individual member’s situation or representation in court can hardly be mass customized.

694


Former Experiences and Early Results from Mass Customized Services It is too early to evaluate the results from the strategies presented for personalization and mass customization. However, some indicators can be presented.

Personalization seems to be successful strategy The new model for personalization, described above as My Profile, cannot be evaluated yet. Specific member groups have, however, been targeted for many years – and evaluation data concerning such groups will provide knowledge on results related to a more personalized approach. Self-employed members have been prioritized for almost a decade. A specific membership form with tailored services and activities has been offered. The results have mainly been positive. In 2007, 40% of these members were satisfied with their membership (7-10 on a ten-digit scale), be compared with 37% among all members. Similar results can be seen among managers, who have also been introduced to specific services and activities forseveral years. For both of these groups, the positive results are interesting, not least because these members are paying rather high union fees and they are by no means "traditional" union members.

Mass customization has been successful More than 11 000 members have used the web-based Career Coach, using the different tools to create more than 40 000 individual reports. About 75% of the users (2007) are satisfied with the tools and about two thirds have found them easy to use. These tools are typical examples of what has been described as enabling and pro-active trade union services: the utilization of the tools enables members to work more actively on their careers; and more than 75% of the users have a job or are students. Additional, individually customized services (each year, 1 500 – 2 000 individual job applications/résumés are assessed, and 500 – 600 members meet an individual career coach) have also been very well received. Interestingly enough, individual career coaching has reached only slightly higher satisfaction levels than web based career services! A problem for trade unions and other organizations providing a wide service portfolio for a set fee to its customers/members is to assess the usefulness of specific services offered. In this particular case, the impact of some specific services has been measured. Figures show that satisfaction with the union’s work on competence development among those who have used the offered career services is much higher (6.72 on a ten-digit scale) than among those who have not


695

used these (5.62). Moreover, overall satisfaction among the membership is much higher among those who have used career services (6.42) than among those who have not used them (5.63). Personalization has been described above as a means for meeting two challenges connected with mass customization (Zipkin 2001): Eliciting a customer’s wants and needs, and eliciting a customer’s willingness to pay for customization Our case has indicated that personalization and mass customization enable the organization to better fulfill members' needs and to create more satisfied members in the groups targeted. It will be an important challenge to further develop interactions with members in order to elicit wants, needs and willingness to pay for a customized service offering.

Conclusions The starting point of this chapter was the question of whether mass customized and personalized services may strengthen or weaken customer relations. We have looked into an organization providing three types of services, described as "one size fits all", "made-to-measure", and "bespoke". Other organizations provide goods of different types, and it is becoming more and more common that organizations provide solutions involving both physical goods and services. Hence, a wide range of product/service portfolio configurations exists. For example, one company may provide "one size fits all" goods, delivered and put together in a "made-to-measure" mode for most customers and more individualized solutions for others. Looking at goods and service types in the manner that has been described in this chapter may help organizations to find weak spots and to target interesting development areas. In many businesses, elaborate combinations of more and less individualized goods and services will be an important source of competitive advantage. What about CRM and the mass customization of services? Initially, I stated that the element of customer co-design in mass customization may strengthen relations, but, on the other hand, limited personal interaction may result in weaker relations than individually customized services. The organization studied strives to maintain close relations with members, and members' participation in the organization’s development is crucial for all democratic trade unions. Unionen combines the utilization of the democratic organization based upon representatives elected by the members with web-based surveys on strategy issues, service evaluations and the tracking of the utilization of services. Member integration is further enforced through member participation in the innovation and development

696


of new services and through Internet-based (and other) communities gathering members with common interests. Has mass customization of services and personalization of the member/union interface led to strengthened or weakened relations? From a short term perspective, the answer is obvious. It would have been impossible – too expensive and hard to organize – to offer bespoke services as a substitute for the made-tomeasure services. Thus, not offering made-to-measure services would have resulted in weaker relations with members. From a long term perspective, the answer is not as clear. Unions must realize not only that they have to meet individual needs, but they should also consider the democratic implications and opportunities inherent in a closer adaptation of activities and services to individual wishes. Listening closely to members and potential members – and acting on what is learnt – is a democratic process. One intricate challenge is to develop mass customized services stimulating members to participate actively in trade union work. Made-to-measure services may, or may not, enforce participation in trade union work. What can be learnt for other organizations? First of all, that mass customization can be an attractive strategy for more kinds of organizations than is usually expected. Secondly, that mass customization can be an attractive strategy for cost efficiency – mass customized services should in most cases be less expensive to provide than fully individualized services. Thirdly, one strategy to create a personalized interface with customers may be to combine multi-dimensional segmentation with individual choice. Finally, organizations need to manage an often complex portfolio of goods and services, in which different parts may be standardized, mass customized or fully individualized. The ability to manage such complex portfolios (and value chains/value networks) will be crucial for the competitive strength of many organizations.

References Adler, P.A. and Adler, P. (1994). Observational techniques. In N.K. Denzin and Y.S. Lincoln (Eds.) Handbook of qualitative research. Thousand Oaks, CA: Sage. Björkman, H. (2005). Learning from members. Tools for strategic positioning and service innovation in trade unions. Stockholm: Stockholm School of Economics, The Economic Research Institute. Björkman, H. and Huzzard, T. (2005). Membership Interface Unionism: A Swedish White-Collar Union in Transition. Economic and Industrial Democracy. 26(1). Bruhn, A. (1999). Individualiseringen och det fackliga kollektivet. En studie av tjänstemännens förhållningssätt till facket. Örebro: Örebro universitet.


697

Cagan, J. and Vogel, C. M. (2002). What drives new product development? Upper Saddle River, NJ: Prentice Hall. Edvardsson, B. and Gustafsson, A. (1999). Quality in the development of new products and services. In B. Edvardsson and A. Gustafsson (Eds.), The Nordic School of Quality Management. Lund, Sweden: Studentlitteratur. Ellis, C. E. and Bochner, A. P. (1996). Composing ethnography: Alternative forms of qualitative writing. Walnut Creek, CA: Alta-Mira. Goldman, A. (1993). "Is that what she said? The politics of collaborative autobiography", Cultural Critique: 177–204. Grönroos, C. (1992). Service management, Ledning, strategi och marknadsföring i Servicekonkurrens (Service management, leadership, strategy, and marketing in competition). Göteborg, Sweden: ISL Förlag. Grönroos, C. (2000). Service Management and Marketing: A Customer Relationship Management Approach. Chichester: John Wiley. Huzzard, T. (2000). Laboring to Learn: Union Renewal in Swedish Manufacturing. Umeå: Boréa. Johne, A. and Storey, C. (1998). New service development: a review of the literature and annotated biography. European Journal of Marketing. 32(3–4): 184–251. Normann, R. (2000). Service Management. Strategy and Leadership in Service Business. Chichester, UK: John Wiley & Sons. Piller, F. T. (2005). Mass Customization: Reflections on the State of the Concept. The International Journal of Flexible Manufacturing Systems. 16: 313–334. Piller, F. T. (2007). Mass Customization. In: C. Wankel (Ed.): 21st Century Management, A Reference Handbook. Tousands Oaks, CA: Sage Publications. Sheth, J. N. and Parvatiyar, A. (1995). Relationship Marketing in Consumer Markets. Antecedents and Consequences. Academy of Marketing Science. 23(4): 255–272. Zipkin, P. (2001). The Limits of Mass Customization. MIT Sloan Management Review. Spring 2001: 81– 87.

Author Biography Hans Björkman received his PhD in Business Administration at Stockholm School of Economics. In his role as senior researcher at Unionen (the major trade union in the Swedish private industry), he is responsible for innovation policies and entrepreneurship. His research interests concern organizational innovative capabilities and customer involvement in service innovations. Contact: [email protected]

1.6

Mass Customization for Individualized Life-long Learning: Needs, Design, and Implementation Hermann Klinger Festo AG & Co. KG, Germany Alexander Benz Ludwig Maximilian University Munich, Germany

We have reached Globalization 3.0. According to Tom Friedman’s book "The World is Flat", globalization has become a reality not only for countries and international corporations, but also for individuals. How can one single individual compete with millions of other people worldwide? How can one person cope with the exponentially growing resources of knowledge? The answer lies in learning and education – individualized, life-long, efficient, and effective. Our proposal is to replace the paradigm of traditional education with its idea of an "economy of scarcity" with an idea of an "economy of self-generation". Mass Customization (MC), understood as the roof for mass customization, customer co-creation, and open innovation provides the conceptual and operational framework for analy-zing needs and the status of individualized education. Consistent with MC a system for individualized lifelong learning from kindergarten and school to university and corporate level has been designed and implemented. Consequently, case studies, i.e. the outputs of education are used both for customization and individualization of curricula. Latest findings from neuroscience and systems theory are utilized as a basis for the argumentation. First results are reported and discussed.

Introduction For more than 100 years the concept of mass production has been at the core of industrial development. Advantages and disadvantages are obvious: low production costs, high quality of products, and amortization models for high development and marketing efforts, but also highly standardized products with low potential for individualization. Interestingly enough, the first concepts of mass customization (MC), aiming at balancing benefits from mass production and individualization appeared in 1899 in France in a picture by Jean Marc Cote (Figure 1).

698

699

VOLUME 2: APPLICATIONS AND CASES fast replenishment

modular products

stable processes

flexible manufacturing

dedicated people

system integration

configurator

Figure 1: Mass Customization as seen by Jean Marc Cote in France in 1899 (Piller 2006).

Piller (2006) thus defines mass customization: "Mass customization refers to a customer co-design process of products and services which meets the needs of each individual customer with regard to certain product features. All operations are performed within a fixed solution space, characterized by stable but still flexible and responsive processes. As a result, the costs associated with customization allow for a price level that does not imply a switch to an upper market segment." However, MC did not reach the masses until effortless and cheap communication and collaboration tools became available. Interestingly, it was not enough just to have e-mail, as in the mid 1990s. The key to success was sharing individual experiences, pictures, media, ideas, and emotions within a broad community. Keywords for this new quality of internet are WEB 2.0 and social computing. Websites like flickr.com, youTube.com, and del.icio.us are representatives for this new type of service. Social computing opened the doors for innovative partnerships between customers and suppliers. More or less Mass Customization, Mass Personalization, and Open Innovation merged into one field, which we call MC. Supported with configuration tools provided by the supplier, customers are configuring and customizing

700


products to meet their personal needs. Supplier and customer are co-creatively developing and/or enhancing individualized products and services within a framework called solution space. Product specifications, like drawings and lists of parts, are directly transferred into the production process without further costs for marketing and sales. Typical "MC products" are individualized snowboards, wrist watches, shoes, sporting goods, customized bicycles, clothes, and T-shirts. Today the concept of MC finds growing acceptance not only in B2C markets, but also in B2B markets. Examples from Festo at www.Festo.com, supplier of industrial automation components, system, and services, are the so-called valve terminals.

Figure 2: A Festo valve terminal: mass customization in B2B markets.

Valve terminals are technical systems consisting of pneumatic and electronic components with some 100,000 possible combinations and a broad range of lengths from a few centimeters to several meters. The customer configures his specific solution with an electronic configuration tool provided by Festo and transfers the data online to Festo. The standardized assembly processes allows for a delivery time of only 48 hours. This includes complete testing of all components and all functions of the system. The chance to producing a specific system more than once is close to zero. Speaking with Chris Anderson’s "The Long Tail" (Anderson 2006), the tail in B2B is sometimes long, very long. Further interesting concepts for other B2B branches, e.g. the plastic industry, can be found in a series of articles on strategic innovations (Plastics Industry 2008).

Growing Need for Education As globalization has reached the individual the connectivity between education and work has become both an individual and a global issue: Closely linked with reduced costs and transportation risks, communication and knowledge exchange,


701

Tom Friedman (Friedman 2005) in his bestseller "The world is flat" identified three areas of globalization over time:

Globalization 1.0, driven by countries,

Globalization 2.0, driven by multinational companies,

Globalization 3.0, driven by individuals, collaboration, and global competition.

Figure 3: Changes in globalization.

In the ongoing discussion on education in Germany, OECD Pisa coordinator Andreas Schleicher recommends raising the quotes for students to go to university from some 30% in Germany to 95% over the next 5 to 10 years, like in Finland (Schleicher 2008). Nobody knows, however, what kind of education will be needed then. Therefore, education for individualized, lifelong learning must at least in parts become configurable by the learner. To keep up, the new process needs to ensure connectivity or, even better, to provide a bidirectional interface between education and work. Reflecting these changes, the aim of personalizing learning is becoming increasingly prominent in scientific and policy discussions on the future of education. It is a natural component of the OECD’s CERI program on "Schooling for Tomorrow" (Centre for Educational Research and Innovation 2006). "Personalizing education springs from the awareness that "one-size-fits-all" approaches to school learning and organization are ill-adapted both to individuals' needs and to the knowledge society at large. …."personalization" can mean many things and raises profound questions about the purposes and possibilities for education."

702


In the report’s chapter on "The Future of Public Services: Personalized Learning" Charles Leadbeater (Leadbeater 2006) covers mass customization and masspersonalization. He concludes: "A mass personalized learning service would be revolutionary. By giving learners a growing voice, their aspirations and ambitions would become central to the way services were organized. At the moment the heart of the system are its institutions and professions –schools and teachers – that lay down what education is and how it should proceed. Studies of performance management across a wide range of organizational fields show that productivity invariably rises when people have a role in setting and thus owning their targets. The same is true for learning." This implies far-reaching changes in the role of professionals and schools. Schools would become solution-assemblers, helping children to get access to the mix and range of learning resources they need, both virtual and face-to-face. Schools would have to form networks and federations which share resources and centres of excellence. An individual school in the network would become a gateway to these shared resources…" Faced with the diversity-efficiency dilemma, private companies apply mass customization strategies to add diversity without adding costs. As schools are urged to become more personalized and customer oriented, they also face a diversity-efficiency dilemma. Sietske Waslander, Rijksuniversiteit, Groningen in her report on "Mass customization in schools: strategies Dutch secondary schools pursue to cope with the diversity-efficiency dilemma" (Waslander 2007) asks how Dutch secondary schools cope with this dilemma and to what extent they apply mass customization strategies. A careful selection procedure aimed at creating a maximum variety of school practices resulted in seventeen schools in which case studies were conducted. Data collection included written material, observations, and interviews. Analysis of the combined data indicated six dimensions along which schools differ in their educational programs:

The first dimension refers to the goals, ranging from school practices, where the same goals apply to all students within a given track, to practices where different goals apply for every single student.

A second dimension refers to content, ranging from schools where all students take exactly the same courses to schools where students are free to choose what they want to learn, irrespective of the track or program they follow.

The third dimension has to do with the pace of learning, ranging from schools where all students need to complete tasks within a given timeframe to schools where students can work entirely at their own pace, irrespective of their agegroup or academic year.


703

The use of learning materials is the fourth dimension, ranging from schools where all students work with the same material to schools where different materials are used for different students, for example a book for student A and a computer for student B, or one textbook for student C and another for student D.

The fifth dimension of diversification refers to learning activities. In some schools all students are engaged in the same learning activities, whereas in other schools different students are involved in different activities. In this last case, some students may work alone, while others work together in small groups, while still others attend lectures in large groups.

The last dimension refers to timetables, ranging from strictly regulated, obligatory daily activities to allowing students great freedom to choose when they want to start, finish, or have a break.

Based on emerging patterns of differentiation, four categories of schools were distinguished. At the lowest end of this continuum is the Guards, offering hardly any diversity. At the other extreme they find the Radical Customizers, offering by far the most diversity. In the middle are two categories of schools that do not greatly differ in the amount of diversity they offer or in the way they offer it. Differentiators capitalize on differentiation of content and pace, while Economizers try to offer differentiation by diversifying learning materials and learning activities. These categories appear to be closely linked to organizational strategies pursued by schools. The main strategy adopted by Guards is to reduce heterogeneity in the student body. By all appearances, this strategy may require a school to have considerable control over its intake, be it overt or covert. If reputation and market position are indeed reasons for pursuing a strategy of reducing heterogeneity, this strategy will only be a viable option for a selected number of schools. Radical Customizers try to escape the diversity-efficiency dilemma by adding resources. The two Radical Customizers in this study offer their students fully customized education. Both schools are deliberately small and operate at the margins of the educational system. In a sense, these schools reflect the severity of the diversity-efficiency dilemma. Even small schools need substantial additional resources to customize education, making it highly unlikely that their strategies and practices could be adopted by larger schools. Few schools will thus be able to fulfill the necessary conditions to escape from the diversity-efficiency dilemma, leaving many schools to face the dilemma in its most severe form.

704


Figure 4: Patterns of diversification in Dutch schools.

Differentiators attempt to postpone the decoupling point and a strategy following almost naturally from it: collaborations and combinations. Schools applying these strategies decompose the curriculum into standardized modules, leaving it mainly up to the students to construct their own learning pathways by mixing and matching modules. While modules themselves are nothing new in education, the modularization that these schools have employed crosses the usual organizational boundaries between tracks and year groups, and that is certainly new. By modularizing all courses in all tracks for all year groups, schools seek to maximize the number of possible combinations that are available to students. Some schools go one step further and collaborate with institutions that are further along in the education chain. Hall and Thomas (Hall 2004) have reported similar developments in the UK. If any conditions are vital for this set of strategies, school size and the number of tracks offered are the likely candidates. Economizers apply the strategy of enlarging the unit of organization. This educational version of exploiting economies of scale follows a basic economic principle to achieve efficiency gains. The three schools adopting this strategy most rigorously share important contextual factors, indicating that both economies of scale and economies of scope are relevant issues. These schools were part of a large school board with a correspondingly large budget, making appropriate new housing possible. These boards also utilize economies of scope. The boards act as regional monopolies, aiming at diversifying their services in an attempt to cater to all educational tastes. Loss of clients is hardly a threat for these boards. Students


705

not attending one school are most likely to attend another school of the same board. At a general level, the two other schools in this category share these characteristics. These schools are also secluded from local competition among schools and receive additional funding for new buildings. It might be that protection from competition and access to appropriate accommodation are necessary conditions to adopt this strategy. If this turns out to be the case, few schools will be able to meet these conditions at short notice. All in all, many – if not most – schools will turn into Differentiators, since the strategy of modularization requires the least strict conditions and is, therefore, a viable option in many situations. " The paper concludes that practices adopted by schools to cope with the diversity-efficiency dilemma strongly resemble mass customization strategies applied by companies producing tangible goods.

Design Elements for a System of Individualized Lifelong Learning According to the findings mentioned above the challenge of future education is to deal with increasing dynamics and complexity and to simultaneously handle the diversity-efficiency dilemma. It appears to be a promising approach to focus on the core processes of value creation. There are four relevant sources for innovative ideas, which ultimately have to be synergetically integrated: Neuroscience, Systems theory, Logistics, and Collaboration and communication.

Neuroscience Neuroscience clearly indicates that efficient and sustainable learning is a highly individual process, depending on the individual’s background, the lessons learned, interests, emotions, attitudes, motivation and more. However there is a framework of communalities we humans share in our brain functions. It is the interconnected functional structure consisting of the 4 function blocks for perception, memory, evaluation and activation. As we all know it is not very efficient for example just to try memorizing something. The first hurdle already is to overcome the "evaluation barrier" and to reach the memory at all. The evaluation function of the brain decides mostly unconsciously whether a signal from the outside world, it may be pictures, noises, speech or words, is relevant or not. If not, the signal is not accepted for further processing. There are no dedicated areas for each function in the brain. Each neuron is linked with about 10.000 other neurons; they inhibit or fire, influenced by each other. They are forming ensembles of neurons in a most flexible and dynamic way.

706


Figure 5: The functional and hierarchical time structure of the brain.

Timing is a very important strategy to structure and coordinate processes (Pöppel 2007). Time windows in the brain can be observed from outside looking into the electromagnetic spectrum transmitted continuously from the brain. Induced emission of radiation or changes in the blood chemistry show other time windows. The shortest time window (Figure 5) is a window of 30 milliseconds duration. As signals from different sensors like the ear or the eye have different pre-processing times, 30 ms is the time window in which the brain assumes that all signals in that time window belong to the same event and are simultaneous. The next longer time window is about 2 to 3 seconds long. This is the temporal platform for conscious activities. As the hierarchical model in Figure 4 shows, there are time windows of longer duration as well. As Pöppel points out, the automatic temporal integration of successive events is provided on the next higher level. To make this happen there must be an anticipatory control mechanism on that higher level. The underlying mechanism is well known as a generalized reafference principle (Holst 1950). An essential feature of Pöppel’s model is "that optimal learning must be embedded in a structure allowing goal orientation or anticipation of what could and should be attained by learning. Without the definition of a goal, the knowledge seeker would be treated as a passive learner neglecting the possibilities of intrinsic motivation provided by the goal, which is the driving force of successful learning and the creation of knowledge. Thus, a learning episode is embedded in time…The time required to reach this goal can sometimes be several years, although it takes just seconds or even milliseconds." From neuroscientific findings we know that the human brain is organized for effortless learning. The mechanisms to set learning goals, to define successive steps, and to evaluate and to correct outcomes are essential parts of human learning.

707


Systems theory Heinz von Förster (1988) calls systems which show deterministic input-output relations trivial machines. Non-trivial machines are systems with internal stati, depending, for example, on their learning history. As opposed to trivial machines, non-trivial machines do not display the same output for the same input. Trivial Machines Synthetically determined

Non-trivial Machines Synthetically determined

Analytically determinable

Analytically un-determinable

Independent from history

Depend on individual history

Predictable behavior

Unpredictable behavior

In this sense the human brain is clearly a type of non-trivial machine, and it should be treated as such. We take the human brain as the "black box" in the center of Figure 6 and add input and outcome. In an educational scenario, inputs are all providers of education – the organization, individuals, or the media used. As we consider the brain as an undeterminable system, the outcome from the brain will only be determined by the brain itself. The outcome can only be influenced directly through the input itself, getting feedback from the facilitator feedback loop. The learner’s feedback loop is set through "self evaluation". Through the "pull line" the learner requests a supply of appropriate knowledge from the educational source, an obvious argument for a guided self-organization of successful learning scenarios.

Figure 6: The human brain represented as a non-trivial machine.

708


Logistics Experience from complex material flow and logistics, e.g. the production of cars (Womack 1990), clearly indicate that the complexity of delivering components just in time can only be handled with a "pull type" control mechanism. As opposed to "push type" control mechanism, "pull type" means, station "n" in a delivery line asks for delivery from station "n-1" just in time and on demand. "Push type" control implies that the flow of material is planned from the past into the future as carefully as possible, and the material flow will follow in reality as planned. Due to the likelihood of unforeseeable events, this type of regime only works as efficiently as anticipated if conditions are sufficiently stable over time. push control media

pull control media

Figure 7: Push-and-pull control structures.

Communities and collaboration Obviously, the past has shown that human learning is never stable, neither for individuals nor for groups. Social computing, WEB 2.0, open innovation, and crowd sourcing are neologisms based on "cooperation and resonance" as an innate human behavior (Bauer 2007). Best practices, like Wikipedia, Linux, and numerous blogs, are known and used worldwide. To build sustainable communities, all individuals must share common visions and goals, trust in the co-creation of value, have a sharing attitude, and are motivated to achieve. To handle the diversity-efficiency dilemma for individualized education, we suggest to adapt the above-mentioned MC definitions worked out by Piller (Piller 2006) to the field of education and to further differentiate it using the 4 fields of innovation mentioned above. Mass Customization for Education (MC4Ed) refers to learner co-designing educational products and services to meet their own individual needs with regard to the 3-dimensional solution space consisting of learning content, learning context, and time and place to learn. Within the solution space offered (Figure 8,


709

building in a conceptualization by Piller 2006), learners are empowered by the providers of education to find their best solutions, configure it in the best way possible, and get it "delivered on demand" and "just in time". The pull type learning process is seen as a co-creative and co-operative process between providers of education and learners. It comprises both the initial configuration phase, as well as the co-creation of the individual solution as a continuous improvement process for education.

Figure 8: Solutions space for clothes in comparison with education.

A Learning System for Individualized Lifelong Learning iL3 The following practice report is intended to shed some light on possibilities for applying MC strategies and experiences to an educational framework. The pilot project for individualized lifelong learning iL3 at Festo is set to cover five dimensions:

Ages from 5 to 50+

Education from kindergarten to school to university

Bridging scenarios of formal education and informal/non-formal education

Contents covering technology, sciences, and management

Context in leisure time, vocational education, professional areas Following the MC4Ed strategy developed in the previous sections, all iL3 activities are based on the same structural building blocks, processes, and tools. Up-to-date information on iL3 can be found at: www.applied-knowing.org. In the

710


remaining of this section, we will describe the building blocks of the program in larger detail. Program: The program consists of various modules to cover a curriculum or parts of it. Realized examples and the programs in preparation include programs for vocational education, in-house company training, and university-level programs (Figs. 9 and 10). Both programs are master degree programs with nine mandatory modules. The duration per module is 8 weeks on the job. The sequence of modules is configurable by the learner. The minimum time required to complete the Master’s Program, including a master’s thesis, is 2 years. Modules: As illustrated in Figure 11, learning takes place in a project-type style with four defined phases (Figure 11).

Figure 9: University Master of Science "Applied Knowing".

Cases: Cases are co-creatively developed from a variety of case assignments strongly referring to the relevant context of learners. The typical context is the working environment for most of the in-house company training, the assumed environment in the student’s aspired career, or just the area of individual interests. There is a standard structure for case assignments with a description of a starting point and a task to solve the case problem. Adaptations and fine tuning of the case


711

assignments both through the "teacher" and the learner occur regularly over the course of the module. Changes are documented and are part of the reflection process in phase four (Figure 11).

Figure 10: Master of Science in Mechatronics.

Figure 11: Standardized learning process for iL3 modules.

712


Roles: Consistent with findings from neuroscience and systems theory, the learner has to be the process owner of his individual learning process. Teachers are regarded as facilitators. They enable students to select and configure appropriate cases for a specific module. They assist in the teamwork in their role as an expert or coach. They deliver knowledge content "just in time" and "on demand". Configurator: As in traditional education, the top-down configuration starts with the selection of a program and/or the modules attached to it. At the module level, however, the learner can select among different case assignments. The learner cocreatively adapts and shapes the chosen case assignment with his/her team of students and the facilitator. For special interest groups, we also offer bottom-up configuration from selecting cases first, to bundling modules to individualized programs, and linking them with knowledge maps. This approach is interesting for leisure-time activities, e.g., fans of robotics or corporate training (Figure 13).

Figure 12: Standardized structure of configurable case assignments.

Development: Since the learning process is a co-creative process between the learner and the teacher, developing new material and ideas for individualized lifelong learning is an ongoing process (Figure 14).


713

Figure 13: Age dependent iL3 online configurator.

Figure 14: The co-creative and continuous development process.

An internet platform called case-factory was created to enable open innovation. The case factory permits learners at all levels to share their ideas with others and document them in a standardized form as case assignments. More information can be found at www.applied-knowing.org.

714


Lessons Learned and Recommendations One of the main challenges for education in the future is to solve the diversityefficiency dilemma of individualized learning. We used MC concepts and operations to set up a pilot project for individualized lifelong learning iL3. Feedback from learners of all ages, all educational levels, and various industries and organizations has been very positive. Here are a few selected quotes:

A.K. "Preparing a solution for my own problem and at the same time learning something new was an unexpected and indeed authentic learning experience."

M. L. "At first I was appalled when they told me that we had to learn this topic in self-organizing teams, but in the end I was amazed at how much and how fast we were able to learn in and with the team and, at the same time, we even applied what we learned."

M. K. "It was amazing how much knowledge there was in the course and in my team and how much creativity the intense discussions were able to bring forward."

S. Q. "Having the opportunity to learn self-organized at my own pace, but still knowing that in case I am stuck there is someone I can talk to, gives security and helped me to create a new self-confidence for future tasks."

H. K. "The applied knowing method was a great way to bring knowledge into application."

Figure 15: The case factory as a platform for open innovation in education.


715

Consistent with the initial assumptions from neuroscience and systems theory, it has clearly been proven that learners are willing and able to select and configure individualized cases, even though the content areas are new to them. Case assignments are specifications for individualized learning goals and can be used as means for diversification. To actively define them creates motivation and structure for the learner to self-organize individual learning steps, milestones, and evaluation criteria and to synchronize the individualized learning process in a highly customized way with teams. Teams have to be as heterogeneous as the members consider helpful to reach the common goals. The role of teachers has to change from the person in front of the class pushing the content to a person willing and being able to facilitate self-organized learning. In all scenarios we critically evaluated our approach to see if requesting learners to configure and customize prepared case assignments provided enough diversity for the learner to accept the learning situation as personalized learning. Standardization of the internal structure of programs, modules, and cases is sufficient to make iL3 work efficiently, both for students and for teachers. Sometimes students "complained" of having spent much more time than they anticipated; nonetheless, they enjoyed doing so. According to their new role as facilitator and coach, teachers will, after a phase of change, spend more time on the core process of learning with the individuals and the teams. In addition to that, a lot of knowledge transfer is shifted to "just in time" and "on demand" processes through peer to peer communication. The average time teachers spent in iL3 was about the same as before. To customize education, we consequently used the innovative paradigm of "output orientation". This is in direct conflict with the more mechanistic philosophy of education – first pre-defined bits of knowledge have to be delivered like components in a parts list to be assembled later for any indeterminate application by the learner himself. Our results clearly indicate that learning and applying knowledge are two sides of the same coin. One of the issues of a further project planned in educational systems in 5 EU countries must, therefore, be to change mindsets and paradigms. We are, of course, using the MC approach to achieve these goals!

References Anderson, C. (2006). The Long Tail. New York: Hyperion. Bauer, J. (2007). Prinzip Menschlichkeit. Hamburg: Hoffmann und Kampe. Förster, H. von (1988). Abbau und Aufbau. In: Lebende Systeme: Wirklichkeitskonstruktionen in der systemischen Therapie. Simon F. B. Frankfurt: Suhrkamp, 32–51. Friedman, T. (2005). The World is Flat. London: Penguin Books.

716


Hall, D. and Thomas, H. (2004). Teaching university modules in sixth forms: the shifting boundaries of post-compulsory education? Journal of Education Policy. 19: 179–193. Holst, E. von and Mittelstaedt H. (1950). Das Reafferenzprinzip. Die Naturwissenschaften. 37(20): 464– 476. Centre for Educational Research and Innovation (2006). Schooling for Tomorrow: Personalizing Education. OECD Publishing. Leadbeater, C. (2006). The Future of Public Services: Personalized Learning. Centre for Educational Research and Innovation. (2006): 101–113. Piller, F. T. (2006). Documentation of the Workshop "Mass Customization" as part of the 8. German Mass-Customization-Meeting, Salzburg, 15 Feb. 2007. Plastics Industry (2008). The Premier Journal for the European Plastics Industry. 1: 20–29. Pöppel, E. (2006). Knowing: Different Forms and Unifying Principles. In: Reports on Applied Knowing I, 20–48. Schleicher, A. (2008). Deutschland braucht mehr Abiturienten. AP March 12. Retrieved from www.nettribune.de/article/120308-152.php. Waslander, S. (2007) Mass customization in schools: Strategies Dutch secondary schools pursue to cope with the diversity-efficiency dilemma. Journal of Education Policy. 22(4): 363–382. Womack, J. P., Jones, D. and Roos, D. (1990). The Machine that Changed the World. New York: Rawson Associates.

Author Biographies Dr. Hermann Klinger is head of the department “Business Development Knowledge” at Festo AG & Co. KG in Esslingen. He also is director of the Festo Program for Applied Knowing and member of the Human Science Center at the Ludwig Maximilian University Munich. Applied Knowing is focusing on economic and social value creation – knowledge is the key production factor to achieve this. His objectives are to design humane, organizationally efficient and technically feasible processes of learning. Based on theses ideas a Master of Science Program for Leadership in Knowledge Driven Organizations has been developed at the LMU Munich, a Master of Engineering Mechatronics Program at the University of Bremen. Contact: www.Festo.com | [email protected] Alexander Benz is the Scientific Coordinator of the Festo Program for Applied Knowing at the Human Science Center of the Ludwig Maximilian University Munich, Germany. His research focuses on innovative ideas and concepts of education combining neuroscientific insights with systems theory ideas and matching them with experiences and best practices in business administration. By combining Mass Customization ideas with a neuroscientifically grounded push-2-pull learning approach the goal of personalized education can be achieved and not only an authentic learning experience but also a winwin situation with integrated value creation for all stakeholders in the learning process can be realized. Contact: www.applied-knowing.org | [email protected]

1.7

A Mass of Customizers: The WordPress Software Ecosystem Andrew Watson Northeastern University, United States

While mass customization is usually associated with tangible goods, it is also relevant to less tangible goods, such as software and services. This chapter focuses on WordPress, which is both software and service. To be specific, the chapter presents WordPress as: blogging software; widely distributed and deeply customizable; a family of products built on a common platform; the focus of a vibrant community; the keystone of a thriving ecosystem; a for-profit business for Automattic, the company founded by its lead developer. It concludes with implications for WordPress itself, and for mass-customized software more generally.

Introduction Software is vital to mass customization. It is widely used in the configuration of mass customized goods, and in the delivery of mass customized services. Conversely, software itself may be mass-customized. In fact, Meyer and Webb (2005) describe the mass customization of software as "extreme" and "pervasive." This article describes the blogging software WordPress in terms of mass customization. The case of WordPress is particularly instructive, due to the number of ways in which WordPress can be and has been customized, and to the importance of customization in the WordPress architecture and in the WordPress ecosystem. As we will see, the diverse members of this ecosystem customize the software in many different ways, and the WordPress platform is architected in order to enable the ecosystem to thrive as it customizes. Mass customization involves "low-cost, volume production of great variety, and even… individually customized goods and services" (Pine: 7). WordPress is software that is coded rather than a tangible object that is produced. Much of the code has been and is being written by people currently employed by a privately-held firm: Automattic. The base software—the mass in mass customization—is architected and developed in such a way as to encourage its customization by others. A fuller discussion of Automattic’s strategy will be possible after we have examined the mass, the customization, and some of the others who do much of the customizing.

717

718


It is useful to regard Automattic and these other organizations and individuals as an ecosystem. Iansiti and Levien (2004) conceptualize an ecosystem as comprising those organizations that "affect, and are affected by, the creation and delivery of a company’s own offerings." The organizations fall in to a number of domains, which may be shared with other ecosystems. One of the domains in Automattic’s ecosystem consists of firms providing web hosting. For example, the main WordPress web site provides a link to BlueHost. BlueHost offers, among other things, automatic installation of WordPress software. It also offers tools for eCommerce sites. Hence BlueHost, and the firms with which it shares a domain, are part of eCommerce-related ecosystems, as well as of the WordPress ecosystem. We will consider the WordPress ecosystem toward the end of this chapter. First, we must establish some relevant key terms. Then we will consider the mass aspects of WordPress. After that come the customization aspects, from which discussion of the ecosystem flows.

Wordpress as Open Source Blogging Software WordPress is a blogging software. It is free/open source software, covered by the GNU General Public License (GPL). Developers of such software often refer to themselves as hackers. WordPress is offered, by Automattic and others, as software as a service (SaaS). The purpose of this section is to define the italicized terms. Hence readers familiar with the terms may wish to skip to the next section ("Code Bases"). First, the purpose of WordPress is to enable people to write weblogs, or blogs. Robert Scoble and Shel Israel define blogs as follows: A blog is really quite simple. It’s nothing more than a personal web site with content displayed in reverse-chronological order. New posts are placed at the top of the page... making it easy to see what has changed. In most cases, site visitors can identify the author and leave comments for others to see. Blogs are loosely joined to each other through hyperlinks (Scoble and Israel 2006, p.26). Although they stress the "personal" aspect of blogging, these authors go on to describe the power of blogging for firms. Similarly, a Business Week cover story insisted that "Blogs will change your business" (Baker and Green 2005). WordPress is free/open source software. This means that anyone is free to take the WordPress code, run it, examine it, modify it, and to distribute the derivative code resulting from the modification. To be more specific, WordPress is licensed under


719

the GPL, which means that if the derivative code is distributed, it must itself be distributed under the GPL. The remainder of this chapter describes WordPress as open source software rather than as free software. It would be cumbersome to keep combining the two terms or to retell the history of the arguments between their respective advocates; Moody (2001) and others have already told the story well. It is, however, worth pointing out that advocates of both terms favor the word hacker to describe someone who contributes to software such as WordPress. The current chapter uses hacker in this sense, rather than to describe one who makes unauthorized intrusions into systems. The final point of this section is that the WordPress software is available as a service. You can easily create and post to a blog at WordPress.com without having to install the software yourself. WordPress.com is run by Automattic. It is part of a wider trend toward software as a service (SaaS), the most prominent example of which is SalesForce.com.

Code Bases The "mass" aspect of WordPress is best described in terms of three code bases. We will refer to the original WordPress code base as "WordPress Classic." It was based on, but split off from, earlier blogging software called b2/cafelog. The possibility of such "forks" of software code is a feature, not a bug, of the GPL. WordPress Classic inherited a simple architecture, wherein each instance of the software supports one and only one blog. Hence a multiple-blog system required that WordPress be installed multiple times. A multiple-blog WordPress Classic installation still requires this. This helps to keep WordPress Classic a relatively small, simple software product. However, the one-to-one relationship between blog and blogging software installation hinders the scalability of WordPress Classic. It is expensive in terms of system resources: each instance of the software consumes disk space and, when running, memory. It is also expensive in terms of system management resources: each instance must be not only installed prior to blog creation, but also upgraded in order to take advantage of features and fixes introduced in new versions of WordPress. Hence several projects sprang up to customize WordPress to turn it in to a more scalable multi-blog system. Such customization is of course permitted and indeed encouraged by the GPL. One of those who customized the code was Donncha O Caoimh. His fork of WordPress became WordPress Multi-User (WPMU), and he currently works for Automattic as lead developer of WPMU.

720


WPMU is the second of the three code bases. A WPMU site administrator can in effect offer WordPress as a service. For example, Edublogs.org offers WordPress as a service to academics, providing the hardware, upgrading the software when necessary, and so on. The third code base, WordPress.com, is more of a particular WPMU site than a very different code base. It is of particular relevance for at least two reasons. First, it is by far the largest WPMU site, with over two million blogs as of the start of 2008. Second, it is one of the means by which Automattic earns a return on its investment in the WordPress software (Watson 2008). For example, many of the blogs hosted at WordPress.com carry advertisements, with the revenue going to Automattic, rather than to the bloggers. Table 1 summarizes the differences between the three code bases. Figure 1 shows a WordPress.com blog that has been lightly edited following creation. A WordPress Classic or WPMU blog would look very similar since, as Table 1 shows, much of the difference between the three code bases comes in terms of system management and the back end. Much of the remaining difference comes in terms of customization, and it is to customization that we now turn.

Figure 1: Sample blog with default theme ("Kubrik").

721

VOLUME 2: APPLICATIONS AND CASES Table 1: WordPress code bases. Code Base

Source

Software as a service?

Blogs per code base

Potential for customization

WordPress Classic

Wordpress.org

No

One

Extensive

WordPress Multi-User

Mu.wordpress.org

Software used to provide service

Many

Depends on system administrator

WordPress.com

Not available

Yes

Hundreds of thousands

Within limits

Customization Table 2 shows the ways in which a blog can be customized by the blogger and site administrator. For each of these ways, the table identifies the customization and the means by which it is implemented and the way in which the customization currently applies to each of the three code bases. The current (March 2008) versions of WordPress Classic and WPMU are 2.3 and 1.3 respectively. WordPress.com does not have a public version number: software upgrades are not performed by users, since timely upgrading is part of the service aspect of SaaS. Table 2: WordPress customization. Customization

Mechanism

WordPress Classic v2.2

WPMU v1.2

WordPress.com

Free/open source

GPL v2

Yes

Yes, for site administrator

No

Adding function

Plugins

Yes

Admin provides menu, blogger selects from it

No

Changing page structure

Themes

Yes



Changing page style

CSS

Yes

Admin may enable

Yes, as premium service

Changing sidebar content

Widgets

Yes (requires plugin)



Localization

Translation files

Yes

Yes, although not well-documented

Yes, language selected by blogger

722


WordPress Classic is the most customizable of the three code bases. This is to some extent because the source code is open, and so any hacker is at liberty to modify WordPress. This is not to say that one must be an accomplished hacker in order to customize one’s WordPress Classic blog. For example, a blogger wishing to extend the function of WordPress Classic may well be able to find and "plug in" code already written and made available by another person. Hence the blogger provides administration, as well as content, for her WordPress Classic blog. When it comes to WPMU, customization is possible at two levels. At the "higher" level, the administrator of the WPMU system in effect sets up a series of menus for the whole site. Each blogger then customizes her blog by selecting from the menu. WordPress.com is to a large extent a WPMU site, with site administration provided by Automattic. The first row of Table 2 refers to the fact that WordPress is open source software, under the GPL. Given the above-discussed implications of the GPL, it may seem curious that anything other than "Yes" appears in any cell of this row. However, WordPress.com offers the software as a service. It does not, strictly speaking, distribute the software. Hence Automattic is at liberty to modify the WPMU code for use at WordPress.com without having to release the modifications under the GPL. This is a point about the GPL, rather than about WordPress in particular. It is sometimes described as a loophole in the GPL, and one that will become increasingly important if the trend toward SaaS continues. The next means of customization is well introduced in the WordPress documentation. Plugins are tools to extend the functionality of WordPress. The core of WordPress is designed to be lean, to maximize flexibility and minimize code bloat. Plugins offer custom functions and features so that each user can tailor their site to their specific needs. (WordPress 2008) As an example, one of the most widely-used plugins is "Subscribe to Comments." When visitors to a blog leave comments, they may wish to be informed of further comments, rather than having to revisit the posts on which they commented in order to check for subsequent comments. If the blogger has used "Subscribe to Comments," visitors leaving comments will be invited to check a box requesting that subsequent comments be emailed to them. That is just one example. There are hundreds of plugins, varying along dimensions such as complexity, popularity, and visibility to visitors. Plugins have many advantages in addition to those explicitly identified in the above quote from the documentation. First, they allow function to be added in a modular fashion. Second, plugins allow anyone to extend the function of their WordPress blog, and to make that extension available to the WordPress community by contributing it to


723

the WordPress plugins directory. This may mean that the same function is offered to the community by means of multiple different plugins. This allows a blogger to select the implementation of the function best suited to her needs. On the other hand, plugins have their drawbacks as well as their advantages. While the existence of multiple plugins with similar function allows the blogger choice and flexibility, it may also create confusion for the blogger—and for a reader who finds the same function provided in different ways at different WordPress blogs. Plugins can also make upgrading to a new version of WordPress more complex, in that new versions may "break" plugins, and plugin writers do not always provide timely upgrades. Plugins fit differently into each of the three code bases. A blogger running WordPress Classic finds the plugin she needs, uploads it to the appropriate directory of her blog site, and activates it. A blogger using WPMU may activate any plugin uploaded by the site administrator. She may not upload plugins herself, although she may, depending on the WPMU site and its administrator, be able to lobby for a particular plugin to be available. A blogger using WordPress.com never sees the plugins administration menu. We can say, at the risk of oversimplifying, that Automatic, as WordPress.com site administrator, selects, uploads, and activates plugins on a site-wide basis. The third row of Table 2 describes themes, which determine the layout of the web pages comprising a blog. The contrast between Figure 1 and Figure 2 illustrates the different that themes make. Figure 1 shows the demo blog with the default theme: Kubrik. Figure 2 shows the same blog with a different theme: Benevolence. Although the content has not changed, almost everything about the blog looks different: the header; the sidebar, which has moved from right to left; etc.. The demo blog is at WordPress.com, where the number of themes available is currently in the dozens. The length of the theme menu for WPMU differs from site to site, under the control of the site administrator. A blogger using WordPress Classic can choose from over a thousand themes that have been developed and made available by members of the WordPress community. As with plugins, the blogger uploads the code she needs to her blog site. What if the blogger likes some aspects of a theme, but not others? One answer is that, if she is using WordPress Classic and has some proficiency in PHP, she can modify the theme, or even write her own theme. There are other answers that do not require programming, and that work at WordPress.com. Some themes allow replacement of the header image. For example, our blogger could retain the Benevolence theme, but replace the grassy header image with an image of her own choosing, provided it had the same dimensions.

724


Figure 2: Sample blog with different theme ("Benevolence").

The fourth and fifth rows of the table describe two other ways in which the blogger can customize her preferred theme. One is by using CSS: Cascading Style Sheets. The emphasis belongs on the word style because CSS is a means of describing the style of a web page separately from its structure. Both the structure and the style of the demo blog in Figure 2 are determined by the Benevolence theme. Structure refers to things such as the placement beneath the header of the post of the date and time on which the post was made. Note from Figure 1 that the Kubrik theme defines a different structure; for example, it does not include the time of the post. Style refers to things such as the font. Hence our blogger could use CSS in order to change the size of the type for the post headings. She could also use it to change the color of link text from red to a more restful blue. She could not use it to change the structure of the web pages comprising the blog; for example, if she wanted the date and time of posts to appear below the post content, rather than above it, she would have to switch to a theme other than Benevolence. CSS is important to WordPress in several ways. First, it is an example of modularity and of sound web design. If our blogger wants to change the size of the type, or any other style-related aspect of the theme, she has to change only the CSS file for the theme, rather than having to look through all the PHP files for points at which style is specified. Moreover, when she changes the style by editing the CSS, she can be confident that she is not at the same time making unintended


725

changes to the structure of the page. Second, CSS is important to WPMU and to WordPress.com in terms of security. Allowing bloggers to edit PHP and HTML presents a security risk to other bloggers and to the site in general. Allowing bloggers to edit CSS presents no such risk. The third way in which CSS is important to WordPress follows from each of the first two. Due to the modularity and security implications of allowing CSS editing, it is possible to allow bloggers to edit CSS, and to charge them a premium for this. Indeed, Custom CSS is one of the for-pay addons available at WordPress.com. The availability of Custom CSS at a WPMU site is a decision for the site administrator. The last mechanism for customizing WordPress themes is sidebar widgets. Again, this mechanism is best illustrated by a "before and after" contrast. This time the relevant contrast is the one between Figures 2 and 3. Figure 2 shows the blog with the standard Benevolence sidebar. In Figure 3, the standard sidebar has been replaced by three sidebar widgets: a calendar, a set of links, and a more freeform "text widget." Many available widgets are not shown: these include a search box, a selection of photos from the blogger’s Flickr.com account, and many more.

Figure 3: Sample blog with sidebar widgets.

726


The Ecosystem Automattic, as the firm that defines WordPress, occupies the central position in the WordPress ecosystem, and is its own single-firm domain. A second domain in the WordPress ecosystem was introduced early in the current chapter; that domain comprises web hosts. Then there are the bloggers themselves, who comprise multiple domains. Before identifying and describing these domains, it bears remarking that almost every member of the ecosystem is a blogger. Matt Mullenweg, for example, was blogging (atPhotoMatt.net) before WordPress, let alone Automattic, existed. Nevertheless, Mullenweg and his colleagues are more accurately and usefully placed in the Automattic domain than in a more general domain of WordPress bloggers. A similar point applies to another category of WordPress blogger. This category comprises what von Hippel (2005) describes as "lead users." These include the bloggers who customize WordPress, and make the customization available to others. For example, only a small percentage of the WordPress themes come from Automattic. Many are developed by individuals, sometimes in response to a theme design competition. WordPress.com is designed for customization. In fact, Table 2 is not a bad illustration of the very architecture of WordPress. The effect of this is to encourage and expand the lead user domain. We can now turn from the architecture of WordPress to the strategy of Automattic. It is what Iansiti and Levien (2004) describe as a keystone strategy. This contrasts with the strategy of dominating the ecosystem at the expense of other members. The keystone strategy aims at a central, although not dominant, position within a thriving ecosystem. Iansiti and Levien argue that it is preferable to the strategy of ecosystem dominance in complex and turbulent environments; the web is certainly such an environment.

Conclusions Although Automattic is privately held, and thus does not release financials, recent events suggest that its keystone strategy is working well. In January 2008, CEO Toni Schneider blogged that Automattic closed a $29.5M round of financing. Over the last two years, Automattic’s business has been expanding at a rapid rate. Our most prominent service, WordPress.com has grown to over 2 million bloggers. Their blogs are read by an astounding 114 million unique visitors from all over the world… Revenues have been growing as well, we've been profitable as a business, and we've accomplished all of this with a fantastic team of fewer than 20 people.


727

Late last year we sat down to figure out how we'd like to expand our business in 2008 and beyond. Since things are working well, we didn't want to make any major changes. However, we did set a couple of new goals. One was to put enough money in the bank to have financial security for years to come. Another was to invest more aggressively into our "other" products and services. Of particular relevant to the keystone strategy is the intention to grow, not only the WordPress business, but other businesses. These other businesses are related to WordPress. Akismet provides a good example. Most bloggers allow readers of their blogs to add comments. Unfortunately, "spammers" have discovered this, and so many comments are in fact generated by software rather than written by humans, and are attempts to drive traffic to commercial sites. Akismet is a software service to detect and quarantine spam comments. It is implemented as a plugin for WordPress, but can be used to fight spam for software other than WordPress. Akismet is free of charge for individual use, but there is a charge for its commercial use. Two comments linking Akismet back to WordPress will complete this account of Automattic’s keystone strategy. First, the decision to branch out from WordPress itself reflects an intention to grow Automattic without trying to dominate the WordPress ecosystem. Second, Akismet is not the only spam-fighting plugin. It faces competition from other members of the WordPress ecosystem. Such competition is one of the things that makes the ecosystem thrive.

References Baker, S. and Green, H. (2005). Blogs will change your business. Business Week, May 2. www.businessweek.com/magazine/content/05_18/b3931001_mz001.htm. Ianstiti, M. and Levien, R. (2004). Strategy as Ecology. Harvard Business Review. 82(3). Langer, M. and Jordan, M. (2006). Visual Quickstart Guide: WordPress 2, Berkeley, CA: Peachpit. Meyer, M. and Webb, P. (2005). Modular, Layered Architecture: The Necessary Foundation for Effective Mass Customization in Software. International Journal of Mass Customization. 1(1): 14–36. Moody, G. (2001). Rebel code: Linux and the open source revolution, New York: Basic Books. Mullenweg, M. (2003). The blogging software dilemma. Photo Matt. Jan 24. photomatt.net/ 2003/01/24/the-blogging-software-dilemma/. Pine, B. J. (1992). Mass customization. Cambridge, MA: Harvard Business School Press. Schneider, T. (2008). Automattic fundraising. Jan 22. toni.schneidersf.com/2008/01/22/automatticfundraising/. Scoble, R. and Israel, S. (2006). Naked conversations: How blogs are changing the way businesses talk with customers. Wiley. Von Hippel (2005). Democratizing Innovation, Cambridge, MA: MIT Press.

728


Watson, A. (2008). Making Money From WordPress.com. March 12. changingway.org/2008/03/12/ making-money-from-wordpresscom/ WordPress (2008). Plugins. codex.wordpress.org/Plugins

Author Biography Dr Andrew Watson is currently an independent writer and consultant, focusing on social media and strategy. His particular focus within social media has been on the WordPress blogging/publishing platform. More recently he has turned his attention to web widgets: modules that can be shared and reused across multiple web pages. Dr Watson was previously on the faculty of Northeastern University’s College of Business. He was based there while preparing the earlier versions of his account of the WordPress ecosystem. He remains Boston-based. Contact: changingway.org/ & widgetstrategist.com/ | [email protected]

2

Beyond Bespoke Tailoring: Mass Customization in the Apparel Industry

729


2.1

Virtual Fit of Apparel on the Internet: Current Technology and Future Needs Susan Ashdown Cornell University Emily Calhoun Cornell University Lindsay Lyman-Clarke Cornell University

The growth of online retailing of apparel is limited by lack of information on clothing fit. The development of fit visualization and size selection technologies can help provide this information, and can contribute to making the online shopping experience an easy and fun experience for the consumer. Three technologies were studied in this pilot test; 1) My Virtual Model avatar creation and virtual try-on was studied for ease of use, appeal of the technology, and satisfaction with the avatar and style image, 2) H&M size selection technology was tested for effectiveness of the technology and consumer acceptance, and 3) three-dimensional body scan technology was studied as a source of a more realistic and dimensionally accurate avatar and the acceptance of this avatar by the consumer. Twenty women aged 20-23 with a self-identified pant size of 0-12 were recruited to be participants in the study. Participants used the My Virtual Model (MVM) website to create their avatar and view a virtual try-on of a style of H&M jeans. They also completed the size selection process for the jeans, and physically tried on the recommended size of the jeans. Finally participants were scanned and they viewed their body scan along with the MVM avatar to judge which format they preferred. Most participants enjoyed the MVM avatar creation process and the H&M size selection process, and about 2/3 of them said that it was a useful tool for estimating and visualizing garment fit. However, only four of the 20 participants actually liked the fit of the jeans selected with the size selection technology, though manufacturing issues resulting in unreliable garment specifications may have contributed to fit problems. Most participants felt the garment in the suggested size had an error in at least one area with regard to fit. Participants were divided on their preference for the MVM avatar or the body scan, but most said that they would prefer a more realistic avatar. Overall these technologies can improve consumer confidence in purchasing, and ultimately boost sales and reduce returns.

731

732


Background and Literature Review Technological developments are changing the fashion industry rapidly. The ability to view and purchase apparel on the internet is one developing technology with many implications for the industry. Use of the internet has begun to have a great impact on fashion retailing. Websites that compete with or compliment bricksand-mortar stores are growing in popularity on a global basis, perhaps due to the boundary-free nature of the internet (which allows international access to sites and products) as well as the convenience of shopping online. Significant gains were posted in "e-tailing" between 2003 and 2004, indicating the increasing strength of the online market (US Department of Commerce 2004). However, the growth of this direct marketing avenue is limited by the consumer’s need for information about the sizing and fit of apparel. As sizing is not standardized and size labeling is often inconsistent and uninformative, many consumers frequently experience difficulty finding clothing that fits them well. Manufacturing variability and the wide variation of body dimensions and proportions of populations complicate this situation further. This problem and the high number of returns from apparel purchases made over the internet is a major limitation. New technologies for visualizing virtual garments on an avatar and automatic size selection from body measurements are under development. However, these technologies are at an early stage of development and have not yet reached their full potential. The apparel industry has taken a keen interest in these technologies for several reasons. Retailers who sell apparel on the web find that their products are returned frequently due to poor fit. As the consumer cannot try on clothing they purchase online, they have little information on how the clothing will fit, unless they have tried that retailer’s clothing in a brick-and-mortar store already. Two issues, that the sizing systems for women in the US do not use size labels based on body measurements, and that many companies engage in vanity sizing (labeling their garments with increasingly lower size numbers) confuse the issue even further. Returns can cost a business $10 to $15 per garment return, which can be a serious problem if the company’s sizing is atypical and they face a large quantity of returns (Speer 2002). My Virtual Model (MVM) provides a website which allows the user to create a digital model or avatar of themselves. A range of aesthetic features are available, such as hairstyle and skin tone, to create an avatar that bears a resemblance to the user’s appearance and ethnicity. The parametric avatar is created using a computer algorithm to construct the model using simple measurements such as height and weight, and the user’s concept of their body shape. This avatar can be viewed


733

from several angles. Apparel retailers including Lands' End, Sears, Speedo, and H&M provide digital garments through the MVM website that can be "dressed" on the avatar to give the buyer a virtual representation of how they might look in the outfit. Another technology-based aid to locating well-fitting clothing is automated size selection from body measurements. Several online sites provide size selection based on self-measured body measurements (zafu.com, truejeans.com, sizemeuponline.com). The scan company Intellifit in the US places body scanners in shopping malls in which a customer can be scanned; they can then print out a list of apparel products from various vendors in the mall, and appropriate sizes that have been selected based on their body measurements. H&M on the MVM website offers size selection from self measured body measurements. Yet another developing technology with the potential to contribute to virtual fit is three dimensional body scanning. The leader in the development of this technology for the apparel industry in the US is the Textile Clothing Technology Corporation ([TC]2) (www.tc2.com). [TC]2 has pursued the development of a body scanner for many years, and their scanner was first made commercially available in 1998. Their initial focus was acquiring scan measurements for custom clothing, an initiative that has been successfully implemented by Brooks Brothers, Lori Coulter, and Benchmark Clothiers. [TC]2 also conducted the SizeUSA project, a much needed anthropometric survey of the civilian population of the United States (www.sizeusa.com). Data from SizeUSA has been used by many companies to improve their sizing systems in order to provide effective and consistent sizing and fit of their products. Recent work by [TC] 2 is beginning to focus on the use of the scanner for other purposes including the use of a body scan in the creation of an avatar for fit testing of virtual garments (imagetwin.com). Avatars for virtual fit can be created two different ways, by morphing an artist’s rendition of a standard body based on linear body measurements (the parametric model), or using a body scanner to create a 3D model of the customer. Most conventional 3D surface scanners use light sources (eye-safe lasers or white lights) and a number of cameras (2 to 16) to capture coordinate data to model the surface of the body. Some scanners collect information about the color and surface texture of the body in addition to the measurement data, but the basic data consist of an X,Y,Z coordinate dataset that can be used to reproduce the surface of the body on a computer screen. The image can be rotated and viewed from any angle, and zoomed for a closer view. Scan images can achieve a relatively high resolution of 2 to 4mm. The person to be scanned dons a close fitting scanning outfit, and must stand immobile for the time that it takes to capture the scan data

734


(generally about 12 seconds). An avatar from a 3D scan is extremely detailed and personal compared to a parametric avatar, and can capture exact body proportions, posture and even facial expressions clearly. The 3D scan offers the person scanned a unique view of their own body. There is no other medium, except sculpture, that gives an accurate 3D view of a person. Even a three-way mirror only shows 2D versions of the back and/or side of the viewer. A body scan provides a full 3D view that can be rotated and is a radical departure from a reflected mirror image, a photograph or a video. Though many people find viewing their scan an exciting and positive experience, this image can also be disconcerting to a person viewing themselves in 3D for the first time. This effect may be less evident for scans that incorporate color and texture information providing a more familiar view of the body. Studies show that people are generally very willing to be scanned for research studies or for custom-fitted garments. In research studies 203 women, aged 35 to 54 years, were scanned in both a Lycra scan suit and research test pants and 35 undergraduate female students, aged 19 to 22 years, were scanned in either a Lycra scan suit or street clothing. Participants in both age groups generally found the body scan process acceptable: 88% of the older sample and 77% of the students indicated they were comfortable to very comfortable with the process (Loker, Cowie, Ashdown and Lewis 2004). Interest in technologies related to apparel purchases was also high among these research participants. Both samples were questioned about potential clothing applications of body scan data. Both groups gave size selection the highest rating indicating interest in this technology. The older women rated virtual try-on second highest, and the younger group rated custom fit second highest and virtual try-on third highest. These technologies allow the consumer a chance to feel that their garments are matched to them, and that the fit will be exactly what they desire. Study participants indicated that they would be more inclined to make internet purchases if technology such as size selection and virtual try-on were more widely available (Loker, Ashdown, Cowie and Schoenfelder 2004). For apparel businesses involved in e-commerce, incorporation of a technology that is effective and engaging for the consumer has the potential to boost sales.

Study Questions This technology is only in its infancy and may not be as effective as the consumer would like. The visualizations may not show a believable representation of the user’s body. Also, the interface may not be optimized for the functions provided. It would be of use to the program designers to know how the average user manipulates the technology. In this study we investigate overall satisfaction of the


735

participants with virtual try-on and size selection technologies, as well as the participant’s preference of avatars, a parametric avatar or a 3D scan of themselves. We also investigate the garment fit and the predictive validity of the technology interface, and the consumer reaction to this technology. We test the size selection process with both self-measurement and with measurements taken by the researcher, and we look at possible sources of problems with the fit of the selected jeans.

Methods For this research project, 20 female participants, who represent self-reported sizes 0-12 and ages 20-23, used the website as the typical consumer would in order to create a virtual avatar (www.myvirtualmodel.com). These participants used the H&M online section of the website, because this retailer offers relatively inexpensive clothing that targets this particular age range. H&M also offers a size suggestion along with the avatar, based on body measurements. Most of the participants were familiar with the H&M brand due to the high exposure in fashion magazines and publications. H&M also features celebritydesigned collections, such as Madonna and Karl Lagerfeld, resonating with a younger consumer interested in fashion-conscious designs. This particular age range also identifies well with the selected brand because of the affordability of the clothing. Items within the store are typically under $100 dollars and represent "fast fashion"; the quick changing of styles to track current trends closely. Garments in this price range allow customers to follow trends, as the buyer is not locked into an expensive investment piece that is intended to be worn for a long time, ideal for the age of the consumers in the study. The selection of study participants in this age group considers their disposable income levels, as well as their response to influences from the fashion media. The H&M garment chosen for this study was a low rise, boot cut jean called the "STAR" style. The fiber content is predominately 99% cotton, 1% Spandex. This particular garment was selected because its silhouette would be least restrictive for fit and the style is favored by the selected target market. A straight leg, tapered jean, for instance, would appeal to fewer participants because it would have a more exacting and possibly unforgiving fit. The standard boot cut with slight flair offers the greatest versatility for the participant group. A full size set of 10 pair of these jeans in one inseam length was purchased, ranging from waist size 25 to 34. Because H&M does not offer their apparel online the jeans were purchased from several retail stores in the northeast.

736


The study procedure was as follows: each participant took her own measurements, following the instructions offered on the MVM website, and input the data (without the researcher’s assistance). Once the participant input her measurements and selected and adjusted the body type she then chose basic features such as eye shape, hair color etc. to customize the avatar. The participant selected the H&M Star jeans from the options in her virtual dressing room and tried them on her newly created avatar. The jeans were displayed on the avatar and the selected size for the participant based on her measurements was indicated. The participant then tried on the actual jeans in the size selected by the size selection interface to see how well they would fit. The participant was scanned in a body scanner to create three-dimensional visual documentation of the fit of the jeans, and also a 3D image to view as a personal avatar. The participant then viewed her body scan in comparison to her MVM avatar in the jeans. Finally the participant completed a questionnaire to gather data on user opinions and reactions to the technologies, perceived garment fit, and locations of misfit in the jeans. The participant was then measured by the researcher for the second stage of the research. To further test the size selection process to determine whether any inaccuracies that may be introduced by errors in self-measurement would affect the success of the size selection process, the researcher input new measurement data for each participant in the MVM site substituting the expert measurements for the measurements taken by the participant. This should limit the error due to measurement variability and ensure greater reliability. The avatar created by the researcher using the expert measures was then compared with the avatar created by the participant, and the size selection process was repeated to see if a new size would be recommended based on the new measurements. To further the findings, a second try-on of the H&M jeans was conducted for those participants for whom a new size was indicated based on the new measurements. These participants then answered selected questions from the initial survey regarding how they perceived the garment fit and the type and location of any misfit for the new jeans size.

Results Demographics. The 20 participants were college students who represented a variety of different interests and geographic origins as they came from a range of academic majors and were originally from different parts of the country. Participants were selected based on self-reported size measurements, including what size the participant would typically buy in a retail environment, and what size they perceived themselves to be in the H&M brand. To create a balanced participant


737

pool, participants were selected to represent a range of body sizes based on selfreported sizes. Of the 20 participants, the two smallest had waist measurements of 26" and 27", three had a waist of 28", four had a waist of 29", two had a waist of 30", four had a waist of 31", and three had a waist of 32". The remaining two subjects had waist measurements of 35" and 36". The largest subject was eliminated from the size selection portion of the study as she did not fit into any of the jeans in the range. User interface and reaction to avatar. Participants generally enjoyed the process of creating their avatar on the MVM website, and found the process easy to do (only one participant found the process "moderately difficult"). Twelve of the participants found the MVM site to be a useful tool for estimating and visualizing garment fit (65%) and an equal number said that they would recommend the website to a friend. Initial reactions to the MVM avatar varied, with 11 respondents reacting positively (55%), seven indifferent responses (35%) and two negative responses (10%). On the other hand, when asked how "representative of how you currently think you look" the image was only one participant felt the MVM avatar was "very accurate" and eight participants felt that it was "accurate" (45%). Five participants felt the avatar was neither accurate nor inaccurate (25%), and six participants found it to be inaccurate (30%). Comments from the participants raged from "I look better virtually than in real life" to "very accurate/ realistic: wish it was a bit more flattering." One participant commented "It was fun to create and I felt it was as good as any virtual model could be." Seven of the participants expressed dissatisfaction with the limited number of body shape options, for example "Weight distribution inaccurate; only 3 adjustments for changing build; too many facial adjustments." Many participants commented that the number of aesthetic controls was excessive, and they would prefer a greater focus on the body. For instance, correcting the size of the lips or the style of hair was of minimal interest to the participants as they created their avatar. Many of the participants suggested that using a scanned photograph for facial representation would have been more time efficient, as well as aesthetically accurate (this feature is now being implemented on the MVM website, but was not available when the study was conducted). The actual avatars created were highly dependent on the participant’s self perception, which influenced the choices they made for the input parameters. For example, two of the participants who have similar proportions (as can be seen from their body scan) and whose measurements (from the expert measurements) were similar created avatars that were visually very different in waist to hip ratio (Figure 1a and b). Length proportions are not well represented in the MVM avatars. The two subjects in Figure 1c and 1d have very different torso lengths.

738


Figure 1c is shorter from neck to crotch proportionally than Figure 1d, but their MVM avatars are identical (Figure 1c-d).

a

b

c

d

Figure 1: MVM avatars compared to images from the participant’s 3D scans. Note. a and b – Avatar’s differences in waist-to-hip proportions reflect the participant’s self perceptions, not actual body measurements. c and d – Measured torso lengths of the participants are different, but these differences are not reflected in their avatar.

Fit ratings. The fit of the jeans predicted by the size selection option provided by H&M, using the measurements input by the participants, was perceived as excellent or good by half of the participants (10 participants). The other half of the participants rated the fit as mediocre, poor, or very poor (Table 1). One participant

739


was outside the range of sizes provided by H&M. 12 participants reported that the size suggested was different from their usual size purchased (60%). Table 1: Recommended sizes from participant’s measurements and from expert measurements, and participant’s judgment of the fit of the pants in the recommended sizes (participants listed from smallest to largest size).

Subject #

Participant Measurement Size

Perceived Fit

Expert Measurement Size

Perceived Fit

19

25

good

27

good

2

25

excellent

27

poor

7

28

good

same - 28

---

14

28

good

29

mediocre

13

27

good

29

mediocre

10

28

mediocre

29

mediocre

5

25

good

29

good

17

26

mediocre

29

good

11

29

mediocre

30

good

9

31

excellent

same - 31

---

3

29

very poor

31

mediocre

12

30

good

31

good

16

30

good

31

mediocre

15

30

poor

31

poor

1

31

poor

same - 31

---

4

30

mediocre

32

poor

8

29

good

33

good

20

33

mediocre

same -33

---

18

31

poor

34

mediocre

6

outside range

---

outside range

---

The size selection process was then tested using expert measurements in place of the self-measured data. The self-measurement process, even with the directions

740


provided from the website, proved to be a problem. Many of the participants had difficulty manipulating the tape measure in a way that allowed for precise and reliable measurements. The expert measurements, when compared to those taken by the participants were different by as much as three inches, with different areas of the body affected. All of the participants had at least one category of measurement that was different from the expert measurement. The average absolute value of the differences in inches between the expert and self measurement data was 0.875" for the waist, 2.25" for the hip, 0.9" for the thigh, 0.75" for the inseam, and 1.15" for the torso. The expert measurements taken by the researcher generated a different size selection in 15 of the 20 cases. In all cases the size recommended from the expert measurements was larger than that generated by the participant measurements, by as much as 3 to 4 inches ( waist size increments) in some cases. The average difference in size was 1.58 inches (garment sizes) between the self-created and researcher-created size selection. The perception of the fit of the jeans in the new size recommended from the expert measurements was equally distributed: five participants judged them to fit worse than the first pair of jeans, six participants judged them to fit the same, and 4 participants judged them to fit better. Participant responses were divided into two groups for comparison between the smaller participants and the larger participants; the results of the fit perceptions of the two pair of jeans were similar for the two groups. Overall, when participants judged the fit to be different, the fit of jeans of a larger size was perceived as better fit than that of a smaller size four times, and the smaller size was perceived as better fit five times. Of the four participants whose jean sizes exhibited the most extreme shift in sizes (3"-4"), two judged the fit of both pair of jeans as the same ("good"), and the other two rated the fit of the pair from the expert measurements as better than those from their own measurement process. "Good" to "mediocre" fit was the most frequent assessment given to the second set of jeans. The extreme responses of "very poor" and "excellent" fit were not represented in the second fitting. The areas of misfit identified by the participants for the first set of pants (from participant measurements) were greatest at the waist (11 participants, 55%), at the hip and buttocks (6 participants each, 30%) and the thighs (5 participants, 25%). For the second set of pants in sizes generated by the expert measurements, the waist again had a high frequency of reported fit problems (7 of the 15 participants). Although no formal fit analysis of the jeans was conducted by experts, the scan images were reviewed, and the fit at the waist was often visibly incorrect.

741


Either excess flesh from a tight fit, or a large gap between the body and the waistband indicating a loose fit were visible in many of the scans. As the results from the size selection and fit testing were so inconsistent, the researcher began to question the dimensions of the jeans. Analysis of the jeans used in the study revealed many problems with the sizes and fabrics that could affect the study. Inter-size intervals at the waist and hip were inconsistent, ranging from .5" to 2.35" at the waist, and from .5" to 2.3" at the hip (Table 2). One size (size 33) was consistently smaller in all dimensions than the previous size, suggesting the possibility that this pair of jeans was actually a smaller size with an incorrect size label. It is also notable that the fabric content varied among the pants, with three pair having a higher Spandex content (2% instead of 1%), and therefore more stretch, and one pair having a totally different composition – 68% cotton, and 32% Polyester. The Cotton/Polyester jeans would have little or no stretch. Although the variations introduced in the manufacturing process could certainly contribute to the wide range of responses to the question of fit preference of the jeans, analysis of the data without the two pair of jeans with the greatest variation in measurements did not result in any changes in the overall results. Table 2: Pant specifications (all measurements in inches).

Size

Fabric Content

Waist:

Intersize interval

Hip:

Intersize interval

Front Crotch to Top of Waistband

Back Crotch to Top of Waistband

Right Thigh

25

99% C., 1% S.

29

0.5

33

1

7.25

12

19.75

26

68% C., 32% P.

29.5

0.5

34

0.5

7.5

12.5

20.75

27

98% C., 2% S.

30

1.5

34

-0.5

8

13.25

21.25

28

99% C., 1% S.

31.5

0.5

34

1

8.25

12.75

21.25

29

98% C., 2% S.

32

1.6

35

1.1

8

12.75

20.75

30

99% C., 1% S.

33.625

1.15

36.125

1.15

8.125

13.313

21.375

31

99% C., 1% S.

34.75

1.35

37.625

1.5

8.438

13.438

22.625

32

99% C., 1% S.

36.125

-0.85

39.125

-2

8.5

13.875

23.125

33

99% C., 1% S.

35.25

2.35

37.125

2.3

7.875

13.25

22.375

34

98% C., 2% S.

37.625

8.875

14.313

23.25

C. = Cotton, S. = Spandex, P. = Polyester

39.375

742


Avatar preference. Ten of the participants, when asked "Which image do you prefer?" (50%) preferred their MVM image to their 3D body scan image, seven participants (35%) preferred their scan image, one participant liked both images equally, and two participants did not like either image. Participants were asked "How did your MVM virtual model compare to your scanned image in the garment?" Half of the participants felt that they were completely dissimilar, seven participants (35%) felt they were alike, and three participants felt they were identical. One participant commented, referring to the MVM avatar "Not effective tool; smoothes over faults to create unrealistic image. I don't think it is an accurate system." In this study we did not distinguish between preference of an avatar image due to how participants liked seeing themselves (i.e., they might like the less realistic image because it looks more like the cultural norm), or preference of an image because it was an image that they trusted to provide useful information about fit.

Discussion and Conclusions One area of the study that revealed the greatest user influence and variation was the self-measuring process. The website offers guidelines for sizing based on measurements made by the user. A tape measure appears on the virtual model to give visual directions on how to make these measurements, but many of the participants disregarded this information. Studies have shown that the process of self-measurement poses problems (Spencer, Roddam and Key 2004), and one of the chief issues observed in this study was the incorrect placement of the tape for the measurements. The hips were most often misrepresented in the measurements, with an average difference between the participant’s reported measurement and the researcher’s expert measurement of 2.25 inches. Participants generally measured their hips too high and the waist too low, compared to the recommendation in the measurement guidelines provided by H&M. Another problem with measurements that resulted in erroneous data was the nature of the measurements required. Some measurements, particularly the inseam and full torso length, were awkward or impossible for the participant take unassisted. Such problems could lead to misreported measurements that would affect the final size selection. All 20 participants viewed the process of creating an avatar positively. All reported a moderate or easier level of difficulty in interacting with the website to create their avatar. A majority also responded positively to the image presented of their virtual selves on the screen. One recurring comment made by many participants was that they felt the avatar was primarily designed to be aesthetically pleasing, resulting in an unrealistic


743

image. Participants felt the body was too smooth and that it looked more like a model than a real person. On the other hand many participants felt that their body scan was too detailed and realistic. It may take some time to get accustomed to this new, three dimensional view of ourselves. Ways of presenting a scan that do not change the essential dimensions but that smooth or modify the scan to present it in a more visually acceptable manner may be an important part of the process to ease the shock of seeing one’s 3D image. Adding color and texture to the image may also make their scan more acceptable to participants in this new technology. Many participants also had difficulty selecting a body shape from the limited choices available on the MVM website: hourglass, triangle (shoulder and upperbody prominent), and inverted triangle (hip and lower-body prominent). Several commented that more body shape options should be included, such as a rectangular shape. Many participants noted that some of the choices that they expected to impact waist shape did not result in visible changes in the avatar. Overall one of the greatest frustrations experienced by the participants of the study was the avatar’s lack of variation in appearance related to the input of body measurements. Manipulations of the aesthetics of the avatar were visible, but differences resulting from measurement changes were often not visible to the user. As shown in Figure 2 the differences in the avatar with different measurements are undetectable. The participant whose avatar is shown in Figure 2 had significant differences in measurements between self-measurement and expertmeasurement. Theoretically there should be a noticeable change in the avatar. The recommended size for the participant when she input her own measurements was 4 sizes (size 29 to 33) different from the recommended size for the participant using the expert measurements. The dramatically different measurements and size difference is substantial enough that it should be obvious in the two different virtual avatars created. This may not be a functionality intended by the designers of the website (that is, measurements may be used only for size selection, and not intended to provide a visible interactive virtual fit process). If this is true then the expectation of the users must be taken in consideration. Ultimately, neither the MVM avatar nor the highly detailed body scan avatar satisfied every participant, possibly due to the conflict between aesthetic issues in how people may prefer to see themselves in an avatar (whether a smoother figure with fewer body variations or a visibly accurate shape) and functional issues of how accurate and therefore how useful the avatar is for showing the true fit and appearance of a garment. An image that does not show every curve of the body, but that does preserve the body proportions and posture of the customer may be the most acceptable avatar. Another option would be to provide a choice to the

744


user, with a variety of options in how realistic their avatar appears on the screen. Further testing in this area would be needed to identify the best choice. One issue noted related to the H&M section of the MVM website, which was designed for the age group studied, is that online purchasing is not an option on this website. The study participants are members of a technologically savvy generation who are generally used to instant gratification. A lack of internet retail options from this particular brand may lower the appeal of using My Virtual Model for H&M fashions by this target market.

Figure created from participant measurements

Figure created from expert measurements

Figure 2: MVM avatars created from self-measurements and expert measurements do not reflect the differences (Note: participant measures – waist 30", hip 38", 8" drop; expert measurements – waist 31", hip 43", 12" drop).

A major complication this study faced was the discovery of the differences in fabric type and garment specifications within the same style of jean. One pair had a heavier-weight fabric with no stretch, and some pairs offered a slightly greater percentage of stretch than others, possibly resulting in variation in fit among the different sizes. The test garments also were not graded correctly according to typical proportional industry grading. The intervals between the sizes were not evenly distributed, which could cause problems in size selection if it is an indication of unreliable garment sizing. Size selection and body measurement based sizing depend heavily on the assumption that each pair of jeans in the style will have a predictable size and fit. However, this level of quality control is difficult given the inherent variability of fabrics and given the manual manufacturing process which can result in human error. The "fast fashion" product development process, in which different lots of the same style may be manufactured by


745

different producers (who may even source the denim from different vendors) will also introduce variation. A limitation of this study was that the size range available in the retail stores was limited. While the website size selection process specified a range of inseam lengths for the jeans, (30", 32", or 34"), only a 32" inseam was available in the stores. Retail space is always at a premium, so stores often will not stock every option in the range. Relying on only retail sales instead of internet sales, which can be drawn from larger size ranges in warehouses, will limit the functionality of a size selection process. The fact that most of the participants had problems with fit at the waist, and that there were both problems with the waist too small and the waist too large, indicates the essential difficulty of fitting the range of body proportions in the population with one style of proportionally graded jeans. A study that continues to experiment with the sizing available to identify whether a better size for the participant does exist within the size range may reveal that the size selection has actually identified the best option available, even though the jeans do not fit well. In this case the problem is with the size range offered, not the size selection process. This problem could be solved if the E-retailer can warehouse a broader range of sizes that provide different proportions between the waist and hip measurements. This choice in sizing is already the case with inseam lengths, for which three lengths are manufactured for each size. This study was conducted with a small participant pool, and can provide good pilot data to guide a larger study. However the participant pool of this study was from a limited demographic. The results from university undergraduate women may not be consistent with use of the website by consumers from other demographics. The participants' education level, ethnicity, age, and other factors may be different from the target market that My Virtual Model considers to be their average user. Studies using participant pools of greater diversity will provide more information about how the population as a whole would respond to this technology. Another limitation of this pilot test was that the fit testing of the two pair of jeans was not a blind test – that is, the second fitting held with a new size of jeans would raise the expectation in the participants that this new selection may fit better than the original selection. A better design of the study would be to generate the sizes from both sets of measurements at the initial meeting with the participant, and then to try on the jeans in a blind test, without revealing which pair was from the participant measurement and which was from the expert measurements.

746


The quality issues revealed in the measurement specifications of the jeans are not unusual, especially for a "fast fashion" vendor such as H&M. The differences in measurements of the different sizes, and even the differences in the fabric properties are not a problem in a retail setting where the customer can try on the jeans. However, in a virtual try-on system or a size selection system it is critical that the measurements and materials within a style are standardized and maintained throughout the sizing system. Without quality sizing and manufacturing, it is unlikely that the size selection can be successful, and consumers may disregard this technology as a worthwhile tool due to its inability to generate a fit representation that is true to life. Overall, My Virtual Model received positive feedback in at least one area from the study participants. It is currently viewed as a fun, enjoyable way to create an image of one’s self. This was enough to satisfy the study participants, who generally did not have high expectations for the technology. For example, one participant commented "Compared to what other internet purchasing options are currently available, I would recommend this service." Though the technology has room for modifications and improvements, it is certainly on track to become a highly valuable program for shoppers looking for convenience and personalized options on the internet.

Appendix: Questionnaire

Please rate the level of difficulty of interacting with My Virtual Model to create your avatar: (not difficult at all) (slightly difficult) (moderately difficult) (mostly difficult) (very difficult)

How did you react to the model of yourself created by My Virtual Model? (please circle one) (very positively) (positively) (indifferently) (negatively) (very negatively) Please describe your reaction

How representative was the virtual model of how you currently think you look? (very accurate) (accurate) (neutral) (inaccurate) (very inaccurate) Please describe

Was the size suggested to you by My Virtual Model the normal clothing size you would buy? Yes ____No ____ If "No" what size do you normally buy? _________

Was the clothing fit suggested on the model the way you would normally wear the garment? Yes ____No ____If no, please explain

How did you perceive the garment to fit? (excellent) (good) (mediocre) (poorly) (very poorly)


747

Please indicate the areas, if any, where you feel the fit is incorrect (check all that apply): Waist ____ Buttocks ____ Hips ____ Thighs _____ Length ____ Rise ____

How did your virtual model compare to your scanned image in the garment? (identical) (very similar) (alike) (a little alike) (looked nothing alike)

Which image do you prefer? My Virtual Model ____ Body Scan Image ____ Both ____ Neither _____

Rate your overall satisfaction with using My Virtual Model to estimate and visualize clothing fit: (excellent) (good) (somewhat satisfactory) (poor) (extremely poor)

Would you use My Virtual Model again? Yes ____ No ____ Please explain.

References Loker, S., Ashdown, S., Cowie, L. and Schoenfelder, K.A. (2004). Consumer interest in commercial applications of body scan data. Journal of Textile and Apparel, Technology and Management. 4(1). Loker, S., Cowie, L., Ashdown, S. and Lewis, V.D. (2004). Female consumers' reactions to body scanning. Clothing and Textiles Research Journal. 22(3): 1–8. Speer, J.K. (2002). Retail sales tools: The virtual world and beyond. Bobbin. 21. Spencer, E.A., Roddam, A.W. and Key, T.J. (2004). Accuracy of self-reported waist and hip measurements in 4492 EPIC–Oxford participants, Public Health Nutrition. 7(6): 723–727. US Department of Commerce (2004). Retail E-commerce Sales First Quarter $15.5 Billion, up 28.1 Percent from First Quarter 2003, Census Bureau Reports, US Department of Commerce News, Washington, DC, available at: www.census.gov/mrts/www/current/html (accessed July 30, 2007).

Author Biographies Susan Ashdown is the Helen G. Canoyer Professor in the Department of Fiber Science & Apparel Design, where she has taught and conducted research since 1991. She has a bachelor’s degree from Grinnell College (Theater: Costuming, 1971), a Master of Arts from Cornell University (Textiles; Apparel Design, 1989) and a Ph.D. from the University of Minnesota (Apparel, 1991). Her research is on the changes in the way that apparel is designed, produced, and distributed driven by new and developing technologies. Her research group addresses the design, sizing, fit, and function of functional clothing, apparel patternmaking, automated custom fit of apparel, apparel fit assessment in research and industry settings, virtual fit, mass customization, and interactions of materials and design, using a full body three-dimensional scanner. Professor Ashdown was awarded fellowships at the Hong Kong Polytechnic Institute (2005), the National Aeronautics and Space Administration (2005), and the Textile Clothing Technology Corporation (1998).

748


Emily Calhoun graduated from the Department of Fiber Science and Apparel Design at Cornell University in May of 2007. She is the Associate Product Manager of Men's Neckwear in the Macy's Merchandising Group. Lindsay Lyman-Clarke was the manager of the Body Scan Research Group at Cornell University from January 2006 through November 2007, where she coordinated diverse research in the area of sizing and fit. Her research interests are in the areas of anthropometric applications for three-dimensional laser body scanning, virtual fit technologies and end user studies on fit preferences and sizing communication. Prior to her work at the Cornell Body Scan Research Group she earned her BS in Apparel Design and MA in Design Research at Cornell. She is currently combining her sizing and fit research background with her love of design as she designs and develops athletic apparel for Isis, a women’s outdoor company. She lives and works in Burlington, VT. Contact: www.bodyscan.human.cornell.edu | [email protected]

2.2

RFID Diffusion in Apparel Retail: How Consumer Interest and Knowledge Lead to Acceptance Sanchit Tiwari Cornell University, United States Suzanne Loker Cornell University, United States

This section will begin with a general state of mass customization approaches using body scanning and RFID technologies. Then, consumer interaction with these technologies will be explored by discussing selected research studies. Finally, the methods and results from the empirical study we conducted about consumer perceptions of body scanning and RFID technologies and applications will be presented and discussed.

RFID Diffusion in Apparel Retail: How Consumer Interest and Knowledge Lead to Acceptance Benefits of Radio Frequency Identification (RFID) technology in the apparel industry have emphasized information transfer and improved product visibility and management across the supply chain (Turowski 1999). With growing consumer demands and the inability to address consumer needs using standardized products and traditional inventory management strategies, apparel firms have experimented with RFID and other information communication technologies, recognizing that "the key to success is no longer solely price competition but the ability to introduce sophisticated information links, forecasting abilities and management systems" (Abernathy, Dunlop, Hammond and Weil 1999, p.59). But RFID applications go beyond manufacturing and distribution logistics. A report on RFID deployment in the apparel supply chain by Kurt Salmon Associates (Bogart and Kay 2005) predicted maximum functionality of RFID to be item level RFID tags at the retail level. Tracking and replenishing orders from the warehouse can be quickly accomplished to maintain the desired amount of store inventory. Product quantity assessment at the store can take place using RFID readers through cartons without individually unpacking and counting every single garment and produce labor cost savings. RFID tags and readers at checkout points 749

750


minimize inconvenience of manual check of consumer shopping bags at the time of store departure to help prevent or reduce shoplifting losses. Customers can even check out by themselves using RFID readers that automatically detect the passive tags and ascertain the quantity being purchased. The purpose of this research was to investigate consumer perception of the diffusion of RFID technology and its potential benefits that attract consumers to this technology’s applications at the retail level. Since retail acts as a link between consumers and businesses, it serves as the ideal level of the supply chain to study the consumer perception of RFID technology required for mass diffusion. Using an e-mail questionnaire method, this research explores possible consumer responses to the deployment of new applications of RFID technologies at a retail store and asks the question: Can RFID increase the consumer likelihood of a return visit to a store and overall purchase at the retail outlets where it has been deployed?

Background: RFID Technology RFID tags are essentially semiconductor microchips used to store and transmit a unique serial number or extensive information about a product. An RFID tag is a redundant technology without the presence of an RF reader in the vicinity of the tag. An RF reader operates at a defined frequency and protocol and has the capacity to read and write information to and from an RFID tag. RFID readers can read or write data from and to RFID tags and serve as the interactive tool between the mainframe computers and the RFID tags, transferring information for real time asset tracking. They are connected to the company PC or mainframe computer through interfacing ports that monitor information collected from the RFID tag (Zhang and Tseng 2005). The roots of the modern day Radio Frequency Identification (RFID) device can be traced back to the invention of radar by Sir Robert Alexander Watson-Wyatt in 1935 and was actively used during the Second World War to warn of approaching airplanes and distinguish between allied and axis military aircrafts ("History of RFID" 2005). Early academic research exploring the potential of a device capable of reflecting information was conducted by Harry Stockman, in his paper "Communication by Means of Reflected Power" in 1948. Stockman stated that "…considerable research and development work has to be done before the remaining basic problems in reflected-power communication are solved and before the field of useful applications is explored" (Stockman 1948). As the potential of RFID was recognized by scientists, RFID applications were developed for factory automation and animal or vehicle tracking between 1960


751

and 1980 (Landt 2001). Multiple tests were conducted using RFID tags for toll collection in the US in the 1980s (Foote 1981) and IBM engineers led RFID research in the early 1990s developing and patenting an ultra-high frequency (UHF) RFID system capable of read distances up to 20 feet and faster information transfer than ever before. UHF RFID was developed further by a consortium setup by Uniform Code Council, EAN International, Procter & Gamble and Gillette to set up the Auto-ID center at the Massachusetts Institute of Technology (MIT) in 1999 (History of RFID, 2005). Government agencies explored the potential role of RFID in toll collection resulting in the establishment of the E-Z Pass Interagency Group that set up a unified toll collection system in 1990 for seven regional tolling agencies in Northeastern US. Building on this, the Harris County Toll Road Authority launched the world’s first toll collection and traffic management system in 1992 (Landt 2001). Over the past two decades, academic institutions, government agencies and large conglomerates have tested RFID for a wide array of applications ranging from asset tracking and identification to payment systems. Since RFID systems can work on various frequencies such as low frequency (124 kHz, 125 kHz or 135 kHz), high frequency (13.6 MHz) or ultra-high frequency (860-960 MHZ) and use different encoding techniques for varied applications, it is extremely important to develop industry standards for adoption across geographical regions and industries. The United States and Europe have historically had regional differences in adopting RFID standards. The US market tended to prefer the use of passive RFID tags that do not require an external power source and receive power through magnetic coupling with the reader. On the other hand, the European market preferred active tags as passive tags cannot generate enough power from readers due to the European Union maximum power limit of 0.5W. The US market standard has been 902-928 MHz band for Ultra-High Frequency (UHF) RFID compared to 420-460MHz in Europe where the 900MHz band has been allocated to GSM cellular networks (Purvis 2001). ISO (International Organization for Standards) and EPC (Electronic Product Code) standards have been created for the purpose of introducing new standards for collaborative RFID ventures across the world. The ISO standards include ISO 15961:2004 application interface and ISO 15961-3 RFID data constructs among others (www.iso.org). The EPC standards include EPC Tag Data Standard, EPC Tag Data Translation standard, Reader Protocol (RP) Standard, Object Naming Service (ONS) Standard among others (www.epcglobalinc.org/ standards/). These standards are now being adopted by businesses around the world to prevent any geographical differences and enable the use of RFID across borders.

752


RFID tags allow real-time tracking facility and the RF reader can read an entire pallet or truckload of materials within seconds as compared to barcodes that allow information transfer only when individually scanned (Booth-Thomas 2005). Moreover, it can transmit information buried under concrete or even under water (Arner 2003) without being in the line of sight. Multiple RFID tags can be read at the same time and the information can be overwritten several times eliminating the chance of human error (Juban and Wyld 2004). Barcodes store simple information such as product SKU and size. RFID can be used in combination with barcodes or be an effective replacement for barcodes as it can store extensive information like product manufacturing date, manufacturer, date of delivery, washing instructions, among other things. Unlike barcodes, RFID can transmit and write new data at every level of the supply chain (Friedman 2003). The world’s largest retailer Wal-Mart ("Wal-Mart Draws Line in the Sand" 2003) and the US Department of Defense served as change agents for carton-level RFID deployment by mandating all of their suppliers to be RFID compliant by January 2005. Manufacturers and distributors were required to incorporate RFID technology within the supply chain in order to retain their contracts.

RFID and the Apparel Industry With most apparel production occurring far from its design and distribution locations and few vertically integrated firms remaining in the industry, coordination between different levels of the supply chain has become extremely challenging and important. Problems such as lack of or excess inventory, long lead times and security threats often lead to dissatisfied consumers and losses for retailers and suppliers (Kilduff 2000). Radio Frequency Identification can play a key role in the coming years in addressing these areas of concern for the industry (Bogart and Kay 2005). RFID tags can be deployed within the apparel supply chain at two levels:

Batch/Pallet level RFID: The containers carrying manufactured products use RFID tags for identification.

Item level RFID: Each unit/SKU uses its own RFID tag for identification. With the cost of each silicon-based RFID tag close to 15 cents ("Cut-price tag" March 2008) and UHF RF readers around $2000 (Merritt 2007), firms like WalMart ("Wal-Mart Draws Line in the Sand" 2003) with smaller margins on each product have adopted Batch level passive RFID tags. RFID tags are attached to cartons, cases or even pallets, depending on the importance of real time information and level of security required for the product. It has been estimated that batch level RFID can reduce the cost of labor at distribution and retail by as much as


753

20% (Laubacher, Kothari, Malone and Subirana 2006) and increase accuracy by replacing human actions with computerization. With item level RFID, inventory information and lead times are available in real time for production planning and distribution activities that can maximize profits. This information can also be shared with distributors and retailers in the supply chain to help them track the manufacturing status of individual items and entire orders and plan their activities accordingly (Zhang and Tseng 2005). Advantages for consumers in apparel retail stores are not always visible to them. Tracking individual items quickly and efficiently and fast check outs are not always attributed to RFID technology. To provide additional product information to consumers while also showcasing the advantages of RFID technology through greater visibility, some retailers such as Prada, British Marks & Spencer, and New Balance have piloted new services. For example, Prada installed large RFID closets in its SoHo store in New York instead of small scanners. These closets read the information from the RFID tags attached to the garment and used monitors in fitting rooms to display materials, garment construction, information as well as garments shown on runway models (Ideo Prada Case Study 2003) for customers to watch as they tried on and considered purchasing apparel items. Unfortunately, the Prada pilot was pulled from the store shortly after it was initiated due to privacy concerns expressed by Prada consumers (Learning from Prada 2002). There has been minimal study of consumer interest, knowledge, and acceptance of RFID technology. A CapGemini report by Vethman (2005) studying the perception of RFID among European consumers reported that only 18% of consumers had ever heard of RFID and only 52% of those had a favorable opinion of the technology. Approximately 41% of consumers felt that RFID would lead to an increase in overall cost of goods, while only 11% of consumers believed that it would lower costs. Security to prevent car theft, shoplifting and counterfeiting was identified by consumers as the most important benefit of RFID and faster checkout, consumer savings, improved access to product information and in-store product suggestions were other important benefits. Consumer tracking through purchase records, use of shopping data by third parties and health issues were perceived by consumers as the most important consumer risks with this technology. Another study conducted with US adults (Roberti 2005; Collins 2005) found that 65% of the consumer respondents had overall privacy and health concerns with the deployment of RFID and consumers having knowledge of RFID had fewer concerns about RFID than those unaware of the technology.

754


Recognition of the need for consumer research has grown among businesses as the number of RFID consumer activist groups questioning the privacy risks involved with RFID technology has increased, such as Electronic Frontier Foundation and Privacy Rights Clearinghouse. CASPIAN (Consumers Against Supermarket Privacy Invasion and Numbering) is dedicated to fighting RFID enabled loyalty cards and refers to RFID tags as “spychips” claiming that RFID technology "will compromise privacy, security and freedom" (spychips.com). One example is the consumer outcry led by the CASPIAN founder Katherine Albrecht that forced Benetton to abort its plan to attach RFID tags in its Sisley line of women’s clothing. Another is accusation by CASPIAN in an April 2006 press release that Levi’s was secretly testing RFID tags at an undisclosed location in the US ("Spychipped Levi’s" 2006). These public media criticisms of RFID have raised concerns in consumers' minds but there has been limited research conducted to study these concerns and their effect on acceptance of RFID technology applications in apparel retail. The results of these studies suggest a relative lack of consumer knowledge of RFID benefits and an accurate picture of its risks in apparel retail. In addition, the research that has been reported has not connected consumer acceptance of RFID to any theoretical framework about diffusion of technology innovation to explain its findings in a broader context.

Theoretical Framework Most new technologies face consumer related roadblocks in their nascent stages and the ones that successfully convince consumers that its benefits outweigh risks are more likely to succeed. Rogers (1995) identified five attributes of innovative technologies that influenced rate of adoption or diffusion within a social system: relative advantage, compatibility, simplicity, trialability and observability. In addition, perceived risks such as time, effort and price (Bauer 1960) as well as privacy and health risks may prevent consumers from trying a new technology. Bauer purports (1960, p.390), "Consumer behavior involves risks in a sense that any action of a consumer will produce consequences which he cannot anticipate with anything approximating certainty, and some of which are at least likely to be unpleasant." Based on Rogers' diffusion of innovation theory, an innovative technology that has a measurable or relative advantage over an incumbent technology, is compatible with the consumer’s needs and currently existing shopping experience, and is simple, that is, can be explained in one sentence and easily understood, tend to be adopted faster than others. RFID offers advantages, compatibility, and simplicity with garment tracking, fast POS checkout, customized shopping


755

through loyalty cards and interactive trial rooms. Innovative technologies that can be trialed in a store visit or during a store purchase without a huge investment in training or infrastructure tend to be adopted faster than those more expensive or logistically difficult. Observable use of a technology tends to increase its consumer acceptance and adoption over technologies that are not visible. Tracking consumer purchases and even behavior through RFID tags attached to clothing has been identified as one of the biggest risks of RFID technology (www.spychips.org). The diffusion of innovation variables based on Rogers’ and Bauer’s work provided the foundation for our study as we investigated the importance consumers place on each of the variables as they considered RFID technology applications in apparel retail. Retailers can consider these results as they search for ways to educate, inform, and interest consumers in the benefits of retail applications of RFID.

Methodology A sample of US consumers, men and women between 20-60 years of age, was purchased from Connecticut-based Survey Sampling International (SSI). The U.S. panel for SSI contains over 2.5 million households and over 7 million household members (all of them over 18 years of age), thus allowing SSI to support surveys based on specific age, gender, ethnicity and geography requirements. Since most of SSI’s surveys are Internet based, panel members recruited through permissionbased techniques are assumed to have access to a computer and Internet. The panel members are not considered a part of a database, but updated regularly based on their response rate for previous surveys, validity of their email addresses and record duplication. Members also have the option of automatically removing themselves from the panel with every survey correspondence and survey invitation. SSI uses its proprietary software capabilities to ensure a representative sample and to prevent respondents from completing a survey multiple times, entering surveys that they are not invited for, or forwarding surveys to other respondents. A website was developed using WebSurveyor (now EFM Feedback), an online survey building tool providing free service to the Cornell University community for instructional and research purposes. A Cornell University server domain dedicated to research projects powered by WebSurveyor hosted the site. Approximately 2900 emails with a brief description of the study, the survey URL and description of an incentive offered by SSI to its respondents were sent to a Response Balanced Survey Sample of 1700 males and 1200 females between the ages of 20-60 years within all income groups. A Response Balanced Survey

756


Sample allows for different response levels across demographics to enable the completed survey to fall in proportion to the census population on age, gender and income. 183 completed surveys were obtained in 5 days (at a response rate of approximately 6%) in March 2006. The questionnaire was developed in three parts. First, respondents were asked for demographic information (age, sex and household income); behavioral information (shopping frequency, likeliness to shop, and technology use); and awareness and knowledge of the technology. Then, a treatment in the form of a brief description and illustration of RFID was developed to present basic information about the technologies in order to assure that all respondents had the same basic information about RFID technology. Finally, the five variables of innovation diffusion as set forth by Rogers (1995) and perceived risk by Bauer (1960) were operationalized using scenarios about RFID use in retail apparel stores. The scenarios presented RFID technology enabled applications such as garment tracking, faster point-of-sale (POS) checkout, customized promotions through loyalty cards and hi-tech trial rooms. Observability was tested by a scenario that addressed consumers' ability to recognize the use of RFID in a store, while trialability was tested by questions on consumer willingness to visit and purchase from stores using RFID. Questions on consumer willingness to use loyalty cards within stores and pay a price premium for the use of RFID and the need for retailers to disclose the use of RFID within a store were used to understand risks perceived by consumers (See Appendix). In most cases, responses were evaluated on a 5-point Likert scale, 5 being highest (most positive) and 1 being lowest (most negative). Two major limitations of this study were its small sample size and use of a random-access consumer panel for its subjects. The limited budget for this research restricted the sample size to 183 as the purchase was based on the number of participants. In addition, the random-access panel collects data from the first 183 participants. This creates a sample bias toward panel members who are quick to respond, and against members who wait more than a few days to respond, only respond to certain types of surveys, and perhaps other personal characteristics. Therefore, the results reported here should be considered with these limitations and future research should be conducted with larger samples that are recruited with more rigorous randomization methods.

Sample Description The sample included 45% (n=82) male and 54% (n=98) female respondents, with three missing values (Table 1). The sample was restricted to people between the

757


ages of 20 and 60 years and included 49% (n=90) respondents aged 40 years or under and 50% (n=91) above the age of 40 years with missing data for two respondents. There were 38% (n=69) respondents in households earning under $60,000 annually, 26% (n=47) between $60,000-$80,000 and 37% (n=67) above $80,000. The response sample had slightly fewer male and older respondents than in the general population and a much lower than expected income average. Therefore, the results can be generalized to a general population of adult male and female Internet users (as it was conducted by email) but not based on income levels. Table 1: Demographic comparison between sample and population (N = 183). Sample frequency (%)

Population* in millions (%)

Male

82 (45)

138 (49)

Female

98 (54)

143 (51)

Under 40 years

90 (50))

81 (52)

Over 40 years

91 (50)

75 (48)

Under $60K

69 (38)

163 (61)

$60-$80K

47 (25)

50 (19)

$80K+

67 (37)

57 (21)

Demographic variables Sex1 (n=180)

Age2 (n=181)

Household Income (n=183)

* Source: US Census Bureau – US Census 2000 (www.census.gov/main/www/cen2000.html) 1. Three missing values for sex; 2. Two missing values for age.

The frequencies for technology ownership and shopping behavior among consumer respondents are presented in Table 2. Cellular phones and DVD players were the most commonly owned media devices with 92% (n=168) ownership. Only 36% (n=65) respondents owned Ipods and 17% (n=31) owned a GPS device. Respondents were divided by how many media devices they owned with 43% (n=78) owning at least five or more media devices categorized as technology savvy and those owning less than five (57%) as less tech savvy. The tech savvy group was expected to have more positive responses toward the diffusion of innovation variables. The sample was fairly evenly divided by shopping frequency with the mode being shopping every one to three months. The respondents tended to like shopping with 64% of the sample saying they either liked shopping a little or liked it a lot. We

758


expected those who shopped more frequently and enjoyed shopping to be more likely to have high interest in RFID technology and its apparel retail applications. Table 2: Frequencies of Shopping Behavior and Technology Ownership (N=183). Behavioral variables

Technology ownership

Technology Savvy*

Shopping Frequency

Attitude towards shopping

Frequency (%) Laptop

104 (57)

Cellphone

168 (92)

Ipod

65 (36)

Digital camera

146 (80)

DVD player

168 (92)

Console games

76 (42)

GPS Device

31 (17)

Yes

78 (43)

No

105 (57)

Once or more than once a week

30 (16)

Every 1-2 weeks

25 (14)

Every 2-4 weeks

33 (18)

Every 1-3 months

41 (22)

Every 3-6 months

27 (15)

Every 6 months or less

26 (15)

Dislike a lot

16 (9)

Dislike a little

17 (9)

Neutral

32 (18)

Like a little

57 (31)

Like a lot

61 (33)

* Calculated value – respondents owning five or more devices categorized as Technology Savvy.

759


After respondents read the treatment statement about RFID technology, they indicated their prior awareness of RFID as displayed on Table 3. 39% (n=72) of the respondents had seen an RFID tag in their clothing, 61% (n=112) indicated that RFID was used for real-time garment tracking, 81% (n=149) that it was used for inventory management, and 34% (n=63) that it could be used to improve customer service. Only 25% (n=46) of respondents identified all three benefits of RFID – real-time tracking, inventory management and customer service – correctly and were considered to have prior knowledge of RFID. We expected those with prior knowledge of RFID applications to be more likely to have interest in RFID applications. Table 3: Sample distribution based on awareness and knowledge of RFID (N=183). RFID Awareness / knowledge RFID Awareness

RFID Benefit Identification

RFID Knowledge*

Frequency (%)

Yes

72 (39)

No

111 (61)

Real time tracking

112 (61)

Inventory management

149 (81)

Customer service

63 (34)

Yes

46 (25)

No

137 (75)

* Calculated value – respondents identifying all three benefits considered knowledgeable.

Diffusion of RFID in Apparel Retail Respondents were given four scenarios on RFID applications in apparel retail to read based on Rogers' five variables on diffusion of innovation variables and Bauer’s perceived risk (See Appendix). For example, the scenario for an application to be used as customers' try-on clothing in a fitting room is displayed in Figure 1. Their responses, on a five-point Likert scale with five being the highest and most positive response, evaluated the RFID retail applications for usefulness, ease of understanding, likelihood of use, and price and security risks and these are displayed in Table 4. The overall mean for the single question regarding observability of RFID over other technology used for garment tracking, smart cards, POS information, and try-on rooms was the highest of all five diffusion of innovation variables at 4.11.

760


The other variables were measured by combined means of RFID applications and were well over the neutral score of three, with ease of use (4.07) and relative advantage (3.84) the next highest, and compatibility with one’s shopping behavior and ease of trial without great cost in money and time with means of 3.54 and 3.55 respectively. This suggests that all of five variables have importance for consumers in the adoption of RFID technology and could be successfully used to educate and engage consumers in these applications, though the simplicity variable might be targeted for use first as the most important.

RFID - Scenario 4: You are at an upscale clothing store. You take a shirt into a fitting room. The fitting room is enabled with an RFID reader linked to an interactive screen inside the fitting room. The reader reads the RFID tag attached to the shirt and the screen shows product information (e.g. a picture of a runway model wearing the shirt, or photos of pants or jackets that might look good with that shirt). The screen even provides an option that allows you to electronically request to have those complementary items (in your requested sizes) brought to your fitting room by a sales associate. Figure 1: Sample scenario for an RFID enabled interactive fitting room

On the other hand, the overall mean for perceived risk was 2.28 and indicated some apprehension about RFID applications in apparel retail. Mean scores for the three individual perceived risk questions were 2.99 for using loyalty cards, 1.61 for disclosure of RFID use, and 2.24 for price. The use of loyalty cards to track shopping behavior for personalized promotions might be seen as an invasion of privacy and most respondents thought retailers should disclose when RFID is being used within a store, even if it is for internal tracking purposes. Consumer respondents were not willing to pay a price premium for RFID so explaining how RFID actually lowers product cost is an important step in gaining consumer acceptance. These results confirm that retailers need to address the risks perceived by consumers with consumers, even though respondents had positive and some very positive opinions toward RFID applications within a store. Respondents who were able to identify all three applications of RFID in apparel retail and were considered to have prior knowledge of RFID had significantly higher means for relative advantage, simplicity, compatibility, trialalability, and risk variables than the rest of the sample (Table 5). These results reveal the importance of knowledge about RFID technology and its application in gaining consumer acceptance. Promotions and marketing that educate consumers about the benefits of RFID are integral to its acceptance. In addition, the fact that knowledgeable respondents in our sample had significantly higher means for the

761


risk variables along with the overall low mean for these questions, suggest that addressing the privacy and price risks are very important to gaining consumer acceptance. Retailers need to aggressively attend to these risk concerns. Table 4. RFID Consumer Ratings for Diffusion of Innovation Variables (Based on a five-point Likert scale where 5=most positive, 3=neutral, 1-most negative). Innovation Diffusion Variables

Relative Advantage

Compatibility

Simplicity

Trialability Observability

Risk

RFID Application

Individual mean

Individual StDev

Garment tracking

4.07

1.07

Smart card

3.52

1.23

POS checkout

3.97

1.16

Trial room

3.85

1.23

Garment tracking

3.76

1..14

Smart card

3.19

1.38

POS checkout

3.74

1.25

Trial room

3.54

1.29

Garment tracking

4.27

0.94

Smart card

4.20

0.99

POS checkout

4.29

0.94

Trial room

4.32

1.02

Store visit

3.60

1.24

Store purchase

3.46

1.13

Recognition

4.11

0.87

Loyalty cards

2.99

1.35

Retailer nondisclosure

1.61

0.73

Price premium

2.24

1.23

Combined mean

Combined StDev

3.84

0.08

3.55

0.10

4.27

0.04

3.54

0.08

4.11

0.87

2.28

1.27

762


Table 5: Consumer responses to RFID scenarios by prior knowledge of RFID (a,b indicate statistically significant difference in mean with p≤0.05). Prior RFID Knowledge Innovation Diffusion Variables

No (n=137)

Yes (n=46)

F

Relative Advantage

3.64a

4.44b

24.95

Simplicity

4.16a

4.57b

7.39

Compatibility

3.31a

4.26b

30.69

Observability

4.00

4.11

0.41

Trialability

3.28a

4.26b

29.22

Risk

2.16a

2.67b

14.94

There were no significant differences in mean levels of Rogers' five diffusion of innovation variables or the perceived risk variable as related to RFID for respondents by gender or age (Table 6). This indicates that men and women, and older and younger consumers can be addressed similarly to encourage interest and adoption of RFID technology applications at apparel retail. But higher income respondents showed higher mean responses than respondents with household incomes less than $60,000 for the ease of use and trialability variables that could encourage diffusion of RFID technology. This suggests that the benefits of RFID related to ease of use and trialability should be emphasized in activities targeting higher income consumers to increase their acceptance of RFID technology applications in apparel retail, but may be less effective with lower income consumers. Respondents with positive shopping attitudes expressed significantly higher interest than those with negative shopping attitudes in four of the five diffusion of innovation variables and trialability responses were in the same direction through not significant (Table 7). This indicates that retailers may have an easier time getting consumers with positive shopping attitudes to accept RFID technology and these consumers could be targeted to be early adopters. Conversely, retailers may have to work harder to convince consumers with negative shopping attitudes to accept RFID technology applications at apparel retail.

763


Table 6: Consumer responses to RFID scenarios by sex, age and household income (a,b indicate statistically significant difference in mean with p≤0.05). Sex

Age

Household Income

F

40 years or under (n=90)

Over 40 years (n=91)

F

Under 60K (n=69)

6080K (n=47)

80K+ (n=67)

F

3.84

0.28

3.97

3.71

3.33

3.70

4.04

3.85

1.67

4.28

4.25

0.04

4.36

4.17

2.01

4.00a

4.38b

4.45b

5.21

Compatibility

3.51

3.59

0.20

3.68

3.42

2.75

3.38

3.68

3.64

1.41

Observability

4.11

3.97

0.64

4.02

4.03

0.01

3.87

4.00

4.27

2.00

Trialability

3.56

3.50

0.11

3.68

3.37

3.43

3.26a

3.63

3.74b

3.29

Risk

2.28

2.30

0.02

2.35

2.23

0.93

2.28

2.23

2.35

0.32

Innovation Diffusion Variables

Male (n=82)

Female (n=98)

Relative Advantage

3.86

Simplicity

Table 7: Consumer responses for RFID scenarios by shopping attitude and frequency (a,b indicate statistically significant difference in mean with p≤0.05). Shopping attitude

Shopping Frequency

F

Every 2 weeks or more (n=55)

2-12 weeks (n=74)

Less than every 3 months (n=53)

F

3.97b

5.15

3.51a

3.88

4.11b

3.89

4.39b

4.35b

5.82

4.05

4.31

4.41

1.91

3.12a

3.46

3.69b

3.84

3.14a

3.59b

3.89b

5.33

Observability

3.39a

4.31b

4.13b

9.23

3.87

3.99

4.23

1.5

Trialability

3.17

3.59

3.61

1.98

3.15a

3.62b

3.76b

3.58

Risk

2.19

2.24

2.33

0.43

2.06a

2.32b

2.45b

3.29


Negative (n=33)

Neutral (n=32)

Positive (n=118)

Relative Advantage

3.35a

3.84

Simplicity

3.8a

Compatibility

764


Frequent shoppers (shopping every 2 weeks or more) were significantly less positive than infrequent shoppers (shopping less than every three months) about RFID technology’s relative advantage and compatibility with current shopping behaviors, as well as its use in trials during store purchases and its risk. It is not surprising that frequent shoppers may be unaffected by the presence of RFID because they will shop under any conditions. The more important result here is the low negative mean (2.06) for risks of RIFD applications that call for educational and marketing programs that promote its benefits and address risk concerns so that these frequent shoppers are accurately informed. Technology savvy respondents were significantly more likely to consider RFID technology applications as a relative advantage and simple to use than less tech savvy respondents. This implies that apparel retailers could choose to focus on tech savvy consumers as the ones who would lead the acceptance of RFID technology application in apparel retail. Retailers could emphasize the advantages and simplicity of RFID technology and even its observability which was rated high by all respondents in educational and promotional activities. Table 8: Consumer responses to RFID scenarios by technology savviness (a,b indicate statistically significant difference in mean with p≤0.05). Tech Savvy No (n=105)

Yes (n=78)

F

Relative Advantage

3.68a

4.05b

6.38

Simplicity

4.12a

4.46b

7.13

Compatibility

3.42

3.72

3.52

Observability

3.93

4.15

2.16

Trialability

3.41

3.68

2.69

Risk

2.32

2.25

0.38


Recommendations and Implications Rogers' Diffusion of Innovation Theory explains that the diffusion of an innovation in a social system is dependent on two factors: consumer acceptance and business deployment. Numerous studies have empirically confirmed the businesscentric advantages of RFID deployment such as asset tracking, inventory


765

management, reduced labor costs and anti-counterfeiting leading to increased interest in deployment among retailers. This study explored diffusion of RFID technology in apparel retail with respect to consumer acceptance, identifying what consumers think about RFID technology applications and what will facilitate its acceptance in apparel retail. Although survey respondents had positive responses towards RFID applications, lack of knowledge has resulted in consumer concerns regarding RFID applications of privacy and price, leading to lower interest. This current state of RFID diffusion is indicated in the Technology Diffusion Matrix (Figure 2) (Tiwari 2007), which underscores the need to enhance customer acceptance of RFID to facilitate diffusion to its target level.

Figure 2: Technology Diffusion Matrix for RFID.

Survey results suggest that consumer concerns are alleviated by increasing their level of RFID knowledge. Apparel retailers have an opportunity and responsibility to educate their customers about RIFD technology and to be transparent about its use. Increased understanding of the direct consumer and retailer benefits of RFID applications as well as accurate information about its risks can increase consumer acceptance of RFID. Innovative consumer-centric applications of RFID such as garment tracking, faster POS checkout, customized promotions and interactive trial rooms offer an improved shopping experience to customers that could build their relationship with the store while it profits from the stream-lined and efficient inventory management. But the Prada case should serve as a warning to retailers that perceived risks of consumers are real and need to be acknowledged and addressed. Understanding and addressing consumer requirements and concerns

766


prior to a large-scale investment in RFID applications by conducting consumer research and pilot tests could greatly improve success as well as protect brand image and prevent financial losses. Specifically, health concerns of radio frequencies, privacy issues such as consumer tracking, and added costs for RFID-enabled services have been identified in the literature and by consumer activist groups as potential risks to consumers. Although radio frequency technology is used by consumers in a number of familiar products such as EZ passes, when the tags are either attached or embedded in clothing that is going to be worn next to the body, the perceived risk may rise. Accurate information about how the technology works, how can be deactivated, and that it has no negative effects on the body needs to be provided. In addition, even though almost 40% of our sample indicated that they had seen RFID tags in their clothing, most tags are being used at the carton level, not at the clothing article level and it is not always obvious to consumers when RFID technology is being used in a particular retail store. This has raised the concern that consumers will not know when the technology is being used to collect information about their shopping habits and following them whenever they are wearing that clothing if tag is not deactivated. While RFID technology provides many attractive enhancements to retail inventory management and customer store experiences, there are still questions about its observable and transparent use in retail stores. Retailers should be open and transparent about RFID use, by posting and talking about its use, the limits of its information gathering, how tags are deactivated at time of purchase, and the benefits to their store operations that are passed on the customers through lower prices.

Conclusions While information communication technologies such as RFID enable mass customization and improve overall efficiency creating wide appeal and acceptance among businesses, consumer acceptance is vital to enhance the diffusion of innovative technologies from pilot to mass deployment. Although funding limited our study in terms of sample size, its results are clear. There is a relationship between consumer knowledge and RFID acceptance and it can be conceptualized as relative advantage, ease of use, compatibility with current behaviors, visibility and easy trial of the technology, and perceived risk. More research is needed to establish a causal relationship between these variables and acceptance of RFID applications in apparel retail. This will help retailers to focus their efforts on RFID enabled services that consumers want and need while addressing the real or perceived risks of RIFD technology. This study demonstrates the overall appeal of


767

RFID applications among apparel retail consumers and highlights the need to improve consumer awareness and knowledge to increase acceptance as well as to alleviate price, privacy and health concerns that deter the mass acceptance of RFID in apparel retail.

References Abernathy, F., Dunlop, J., Hammond, J. and Weil, D. (1999). A Stitch in Time: Lean Retailing and the Transformation of Manufacturing: Lessons From the Apparel and Textile Industries. New York: Oxford University Press. Arner, F. (2003). Mighty smart labels [Electronic Version]. Business Week Online. Retrieved April 2006, from www.businessweek.com/magazine/content/03_39/b3851617.htm. Bauer, R. A. (1960). Consumer behavior as risk taking. Paper presented at the 43rd National Conference of the American Marketing Association. In Cox, Donald F. (ed.). Risk taking and information handling in consumer behavior. Boston: Harvard University. 1967: 389–398. Bogart, S. and Kay, M. (2005). Moving forward with item-level Radio Frequency Identification in apparel and footwear. Retrieved September, 2005, from www.apparelandfootwear.org/UserFiles/ File/RFID/White_Paper-VICS_AAFA_RFID_v11.pdf Booth-Thomas, C. (2005). The see-it-all chip: Radio Frequency Identification. Time, 162, 12-17. Collins, J. (2005). Consumers more RFID-aware, still wary. RFID Journal Retrieved February, 2007, from www.rfidjournal.com/article/articleview/1491/1/1/ Cut-price tag. (2008). The Engineer. Retrieved March, 2008, from http://tinyurl.com/mt6wrp Friedman, D. (2003). Say goodbye to barcodes. Supply House Times. 46(2): 32–36. Foote, R. S. (1981). Prospects for Non-stop Toll Collection using Automatic Vehicle Identification. Traffic Quarterly. 35(3): 445–460. History of RFID. (2005). RFID Journal. Retrieved March, 2008 from http://tinyurl.com/25dvfn Ideo Prada Case Study. (2003). Retrieved November 2006 from www.ideo.com/case_studies/ Juban, R. and Wyld, D. (2004). Would you like chips with that? Consumer perspectives of RFID. Management Research News. 27(11/12): 29. Kilduff, P. (2000). Evolving strategies, structures and relationships in complex and turbulent busines environment: The textile and apparel industries of the millennium. Journal of Textile and Apparel, Technology and Management. 1(1). Landt, J. (2001). Shrouds of time: The history of RFID, Aimglobal, Retrieved March, 2008, from www.aimglobal.org/technologies/rfid/resources/shrouds_of_time.pdf Laubacher, R., Kothari, S., Malone, T. and Subirana, B. (2006). What is RFID worth to your company? Measuring performance at the activity level. MIT Sloan Research Paper No. 4601-06. Learning from Prada (2002). RFID Journal Retrieved November, 2006, from rfidjournal.com/article/view/425 Merritt, R. (2007, March 09). [Electronic Version]. Intel unrolls low-cost UHF RFID Reader chip. EE Times Asia. Retrieved February, 2008 from www.eetasia.com/ ART_8800455865_499488_NP_f455c39d.HTM Purvis, G. (2001). Radio-frequency identification benefits from global alliances. Wireless Web. Retrieved May, 2006, from wireless.iop.org/articles/feature/2/2/4/1

768


Roberti, M. (2005). Consumer awareness of RFID grows [Electronic Version]. RFID Journal. Retrieved November, 2006 from www.rfidjournal.com/article/articleview/1202/1/1/. Rogers, E. (1995). Diffusion of Innovations. New York: The Free Press. Spychipped Levi’s brand jeans hit the US. (2006). RFID Nineteen Eighty-Four Retrieved April, 2007, from www.spychips.com/press-releases/levis-secret-testing.html Stockman, H. (1948). Communication by means of reflected power, Proceedings of the IRE, October 1948. 36(10): 1196–1204 Tiwari, S. (2007). Diffusion of RFID and 3D Body Scanning in Apparel Retail for Mass Customization: A Consumer Study. Master’s Thesis, Cornell University, Ithaca. Turowski, K. (1999). A virtual electronic call center solution for mass customization. Paper presented at the 32nd Annual International Conference on Systems Sciences, Hawaii. Vethman, A. (2005). RFID and consumer. CapGemini Consulting. Wal-Mart draws line in the sand. (2003). RFID Journal. Retrieved April, 2006, from www.rfidjournal.com/article/view/462/1/1/ Zhang, Q. L. and Tseng, M. M. (2005). The role of RFID in mass customization supply chain. Paper presented at the World Congress on Mass Customization and Personalization, Hong Kong University of Science and Technology, Hong Kong.

Appendix: Questionnaire Scenarios for Consumer Perception of RFID RFID Scenario 1 You can't find a particular garment in a store, as it has been incorrectly shelved or it is out of stock. You ask a sales associate for help who has the ability to immediately confirm the product’s availability and its "last known location" using the store database. If the item still cannot be found, the associate can even use an RFID-enabled handheld reader to hunt for the garment. The device beeps more loudly whenever the desired item comes into closer proximity. a) Do you consider RFID-enabled inventory tracking to be useful? Very useful Somewhat useful Neutral Barely useful

Not useful at all

b) Do you consider the notion of RFID-enabled inventory tracking easy to understand? Very easy Somewhat easy Neutral Very complex Somewhat complex c) Would you be likely to shop at a store that uses RFID to track its on-hand inventory? Very likely Somewhat likely Neutral Very unlikely Somewhat unlikely RFID Scenario 2 Your apparel retailer offers optional an RFID-enabled loyalty card to enhance your shopping experience. The RFID-enabled card can alert the retailer of your presence when you enter the store and offers personalized promotions or fashion suggestions based on your prior shopping preferences stored in the retailer’s database. a) Do you consider the optional RFID-enabled loyalty cards to be useful? Very useful Somewhat useful Neutral Barely useful

Not useful at all

b) Do you consider the notion of optional RFID enabled loyalty cards easy to understand? Very easy Somewhat easy Neutral Very complex Somewhat complex

769


c) Would you be likely to request and use the RFID-enabled loyalty Very likely Somewhat likely Neutral Very unlikely Somewhat unlikely RFID Scenario 3 An RFID reader can process multiple items at the same time without requiring a line of sight. To checkout, you can simply take your individual RFID tagged garments near the checkout area (with RFID readers installed) and pay the bill using your credit card without requiring a store employee at the checkout point. a) Do you consider the RFID-enabled checkout service to be useful? Very useful Somewhat useful Neutral Barely useful

Not useful at all

b) Do you consider the notion of RFID-enabled checkout service easy to understand? Very easy Somewhat easy Neutral Very complex Somewhat complex c) Would you be likely to shop at a store using RFID-enabled checkout service? Very likely Somewhat likely Neutral Very unlikely Somewhat unlikely RFID Scenario 4 You are at an upscale clothing store. You take a shirt into a fitting room. The fitting room is enabled with an RFID reader linked to an interactive screen inside the fitting room. The reader reads the RFID tag attached to the shirt and the screen shows product information (e.g. a picture of a runway model wearing the shirt, or photos of pants or jackets that might look good with that shirt). The screen even provides an option that allows you to electronically request to have those complementary items (in your requested sizes) brought to your fitting room by a sales associate. a) Do you consider these fitting room capabilities to be useful? Very useful Somewhat useful Neutral Barely useful

Not useful at all

b) Do you consider the notion of RFID-enabled fitting rooms easy to understand? Very easy Somewhat easy Neutral Very complex Somewhat complex c) Would you be likely to use these features yourself if you entered an RFID-enabled fitting room? Very likely Somewhat likely Neutral Very unlikely Somewhat unlikely Based on your understanding of the scenarios, do you think you can recognize RFID technology being used in a store? Definitely Likely Neutral Unlikely Not at all RFID enabled customer loyalty cards store your prior shopping information and preferences in the store database to provide customized deals and promotions. Do you have any privacy concerns about this RFID application? Neutral Not concerned at Not very Somewhat Extremely all concerned concerned concerned Do you believe it is necessary for the retailer to inform consumers that individual items contain an RFID tag and that RFID technology is being used within the store? Absolutely Maybe Neutral Maybe not Not at all

770


Would you be willing to pay a price premium for having the benefits of RFID technology as part of your shopping experience? Extremely willing Somewhat willing Neutral Somewhat Extremely unwilling unwilling Would you be interested in visiting a store that is using RFID technology? Very interested Neutral Somewhat Somewhat interested uninterested

Very uninterested

Would you be likely to purchase clothing from a store that is using RFID technology? Very likely Somewhat likely Neutral Very unlikely Somewhat unlikely

Author Biographies Sanchit Tiwari is a Senior Analyst at First Manhattan Consulting Group in New York. He has worked with several large financial institutions in the US and Australia on projects ranging from innovations in direct banking, payment strategies, deposit product management and mass affluent banking. He received a B.Tech from the Indian Institute of Technology (IIT) Delhi, and an M.A. from Cornell University. His Master’s research explored strategies for innovation diffusion for mass customization in apparel retail. Suzanne Loker Professor Suzanne Loker held faculty positions in the Department of Fiber Science and Apparel Design at Cornell University from 19998 to 2008 and previously at the University of Idaho, University of Vermont, and Kansas State University. Her research program focuses on innovative business strategies in the apparel industry, specifically those involving socially responsible practices, and body scanner and mass customization technologies. Contact: www.bodyscan.human.cornell.edu | [email protected]

2.3


Marie-Eve Faust Fashion Merchandising, School of Business Administration, Philadelphia University, USA Serge Carrier Ecole supérieure de mode de Montréal, Université du Québec a Montréal, Canada

This research presented in this paper focused on choice, complexity and simplicity of customized size label. It attempts to answer questions such as: What is the meaning of customization in the apparel industry and how can size labelling be put to use? Do people expect more information from a size label? How can a new size labelling system better support order givers, manufacturers, retailers and consumers? In this paper the authors discussed the application and configuration system and rules sets, i.e.: How to substitute the one size fits all label without going into a pure customization (i.e. cut and sew to fit one individual) by building an integrated sales system. The proposed size labelling system should be an efficient tool for mass customization in the apparel industry.

Introduction Consumer acceptance and production models of mass customization in the apparel industry have been the object of recent academic papers (Caldwell and Workman 1993; Fiore et al. 2001, 2002 and 2004; Anderson-Connell et al. 2002; Istook 2002; Ogawa and Piller 2006; Blecker and Abdelkafi 2006). Several apparel manufacturers, retailers and specialized apparel companies have moved to this new paradigm without difficulties. One needs only think of Levi’s, Nike with customize with nikeid, Lands' End, My Virtual Model, etc. On the other hand, in the last decades, an important segment of the apparel industry has adopted a global market perception: global competition among global companies for global customers (Kerin et al. 2007). This globalizing trend raised a number of questions on the ideal levels of product standardization and product customization. How do order givers determine the sizes they wish to offer? (The expression "order givers" is the translation of the French "donneur d'ordres" identifying the initiator of a garment production. Designers, manufacturers, wholesalers and retailers may all be "order givers" as each one may take the 771

772


responsibility to initiate the process.) Do they adhere to national size standards, and if so which ones? Do they use standard sizing codes even when they do not adhere to the standardized measurements? Do manufacturers produce according to other order givers' specifications, national standards or to their own standards? Do they inform the order givers of their decisions? Do retailers have the tools and knowledge to inform the consumers on the varying measurements between garments identified with identical size labels? [TC]2 (2004), on its Internet site, asked an interesting question: "How is that a 5'8", 150 lbs. woman, a 5'6", 135 lbs. woman and a 5'9", 125 lbs. woman all claim to wear a size 8?" It observed that the existing size labelling system does not fulfill the primary function that led to its creation: to assist consumers in selecting the best fitting garment (Chun-Yoon and Jasper 1996). We may paraphrase Gould-Decauville et al. (1998) and ask: is size labelling appreciated by consumers? This study addressed the question of apparel sizing and size labelling. Facts and assumptions are based on (1) the anthropometric data gathered in the last American national survey by [TC]2 which demonstrated that women come in different sizes and shapes, and (2) the results of our research showing that order givers in the apparel industry (a) do not adhere to national standards, and (b) focus on specific body shapes to mass customize their garments, yet (c) use the standard numerical labelling system regardless of the actual measurements of their garments, thereby (d) creating more confusion than information for the consumer. We therefore proposed a solution based on mass customized service marketing, i.e., we proposed a new customized size label. We concluded that the proposed customized size labelling system would better assist consumers. The apparel industrials by customizing their size labels, would mass customized a standard product at the marketing stage.

Literature Review To understand how sizing, in the garment industry, has evolved to what we know today, one must be aware of its evolution. Therefore, we started this research by looking at the evolution of the standards and sizes.

Measurements in a global world Measurement in the ancien régime referred to a physical standard, held in local hands and safeguarded by local officials. It differed from one place to another. It was the obligation of local officials (Seigneurs), in exchange for a small fee, to enforce the standards, ensuring fare exchanges in the marketplace (Alder 2003).


773

Nowadays standards can be de facto (generally agreed upon for convenience) or de jure (established through legal documents). Numerous organizations write them; some for internal usage only, others intended for use by a group of people or companies, or an industry. The definition of standards varies depending on the purpose for which they are used and on the school of thought that organizations adhere to. According to the oldest standards body institution in the world, the British Standards Institution (BSI), to which the International Standard Organization (ISO) refers for a definition, a standard is: "… a published specification that establishes a common language, and contains a technical specification or other precise criteria and is designed to be used consistently, as a rule, a guideline, or a definition." (www.standardsglossary.com) For the apparel industry, standard sizes, grading systems, measuring methods, and size labelling have been the object of discussion for a number of years and more specifically in relation to garment fit (O'Brien and Shelton 1941; Diffrient et al. 1974; ISO 1976; Workman 1991; Ashdown 1998; Winks 1997; McCulloch et al. 1998; Gould-Decauville 1998-99; Workman and Lentz 2000; Anderson et al. 2000-01; Yertutan 2001; Schofield and LaBat 2005). Glock and Kunz (2000, p.133) defined standards as: "a set of characteristics or procedures that provide a basis for resource and production decision". In the U.S.A. the first major apparel Commercial Standard known as "Body Measurements for the Sizing of Women’s Patterns and Apparel", commonly called the CS215-58, was available in 1958. Interestingly, the report states on its very first page that: "The adoption and use of a Commercial Standard is voluntary" (USDCOTS 1958, p.1) and one of its goals is to: "… provide the consumer with a means of identifying her body type and size from the wide range of body types covered, and enable her to be fitted properly by the same size regardless of the price, type of apparel, or manufacturer of the garment.". In 1970, the U.S.A. government updated the standards and renamed them "Voluntary Product Standard" (PS 42-70), (USDCNBS 1971). Those standards were reviewed in the last decade of the twentieth century by the American Society for Testing and Materials (ASTM in 1994 and after). A number of countries have since developed their own standards with more or less success. Apparel standards were never made mandatory in North America but were published solely to help commercial exchanges. Although this research focused on U.S.A. similar practices are used throughout Canada. The Canadian General

774


Standard Board (CGSB) replicated the American standard system yet specified that each size identification « code de taille » should, as much as possible, reflect commercial practice (Translation of « les tailles sont identifiées à l'aide de chiffres qui sont autant que possible conformes aux pratiques commerciales actuelles », CAN/CGSB-49.203-M87, page 1, 2nd paragraph.). The Canadian General Standards Board (CGSB 1997; flyleaf) state that a national standard is: "… a standard which has been approved by the Standards Council of Canada and one which reflects a reasonable agreement among the views of a number of capable individuals whose collective interests provide, to the greatest practicable extent, a balance of representation of producers, users, consumers and others with relevant interests, as may be appropriate to the subject in hand. It normally is a standard that is capable of making a significant and timely contribution to the national interest." The result now is that manufacturers and retailers’ adherence to national standards, in ready-to-wear, is almost non existent. Let’s now see where these standards come from.

Anthropometric surveys Anthropometry is defined as the measurement of lengths, widths, depths and circumferences of various parts of the human body (Chamberland et al. 1997-98). For O'Brien and Shelton (1941), a national survey based on anthropometric data was the solution of choice to arrive at standard sizes. The earliest population measurement studies of America started with the collection of measurement data on college women (Vassar College, New York in 1884, Stanford University, California in 1890 and Smith College, Massachusetts in 1903 (Kidwell and Christman 1974 read in Workman 1991)). They were soon followed by a more structured anthropometric survey of men and children. The first official U.S.A. national survey for women was the Women’s Measurements for Garment and Pattern Construction (WMGPC), conducted between July 1939 and June 1940. More than 10 000 women participated. The survey’s results were published by the United States Department of Agriculture in 1941. In 1945, the United States Department of Commerce recommended that it be used to establish standards for the garment industry. All U.S.A. apparel standards established in the twentieth century were directly or indirectly based on this survey. The arrival of technologies such as "three dimensional (3D) body scanners" provided the perfect opportunity for governments and private organizations to


775

conduct national surveys to either create or update their anthropometric database. Americans and Europeans initiated this 3D anthropometric age with the CAESAR (Civilian American and European Surface Anthropometry Resource) project which, although never completed, provided the initial experience to move on and succeed with Size UK, Eurotaille, and Size U.S.A. (A national survey was also done in Japan. Report is in Japanese.). Other countries such as Brazil, Mexico, and Thailand also initiated population measurements projects. No government has yet published new national standards based on these surveys. Noting the partial success of these surveys in arriving at standards, it is of interest to recall that Roebuck (1995) stated that anthropometry was first used in an attempt to distinguish races and ethnics groups before being extended to several industries. For Mellian et al. (1991) anthropometry was used to develop standard sizes in the design of military uniforms for the US Navy and US Army which is different from the ready-to-wear. Surprisingly, Schofield and LaBat (2005) showed that none of the 40 U.S. manufacturers' size charts they analyzed, produced between 1873 and 2000, were based on anthropometric research. Over and above the fact that existing standards are now obsolete because our body weights and fitness have evolved, authors believe that standards emanating from national surveys always were of limited use at best as women not only differ in body measurements but also in body shapes (Hamel and Salvas 1992; Rasband 1994; Yu et al. 2006). Glock and Kunz (2000) went further and proposed that, in order to ensure the best fit, it is manufacturers' and retailers' responsibility to consider its target customer when determining measurement specifications. Hence manufacturers often prefer customizing their garment sizing to their target market’s (Burns and Bryant 2002). The difficulty resides in communicating the information to consumer in order to help them in their purchase decision process?

Purchase decision process: size & fit issues According to Eckman et al. (1990) the apparel in-store decision process is divided in three phases. Phase I is "interest" where consumers are attracted by the color, pattern, style and fabric of the items presented. When interested, consumers move to phase II, trying the garment on to evaluate its fit. Lastly, phase III is the "buy or bye" phase. Fitting rooms are the culminating point in apparel retail stores (Reda 2000). Trying a garment on is essential to confirm fit, comfort, and image (LaBat and Delong 1990; Goldsberry et al. 1996; Feather et al. 1997; Horne et al. 1999; Otieno 2000; Anderson et al. 2001; Hart and Dewsnap 2001). It is also essential since size labels do not provide information on a garment’s measurements (Goldsberry et al. 1996; Ashdown 1998), underlying body shape (Rasband and Liechty 2006) or fit preferences (fitted, semi-fitted or loosely fitted) (Anderson

776


et al. 2000-01) even though it should be a tool to assist consumers in selecting the best fitting garment (Chun-Yoon and Jasper 1996). In an attempt to compensate for this deficiency of the size labelling system and better inform consumers, the industry has multiplied size codes (Table 1). Table 1: Summary of sizes offered in the USA. Sizes in USA Junior

1

3

5

7

9

11

13

15

5P

7P

9P

11P

13P

15P

42

6

8

10

12

14

16

18

4P2

6P2

8P

10P

12P

14P

16P

18P

10T

12T

14T

16T

Junior Petite Misses Misses Petite

2P2

Misses Tall Misses2 Women

S 36

38

40

42

20

22

18T

20T

22T

L 44

46

48

50

52

12½

14½

16½

18½

20½

22½

24½

26½

Women Large2

16W

18W

20W

22W

24W

26W

Women Large2

1X

2X

3X

Plus Sizes2

1X

2X

3X

HalfSizes

34

M

17

4X

Whereas some feel that this proves that the garment industry is moving toward more diversity, providing more choice for petite and plus sizes (Labat et Delong 1990), and developing size charts to satisfy their consumer’s needs (Schofield and LaBat 2005), others think that the industry is reacting too slowly to specific segments (Lennon 1992; Goldsberry et al. 1996; Yoo et al. 1999; Campbell and Horne 2001) or illogically using only numbers without meanings to the consumers. The sheer number of possible sizes to carry in stock oftentimes leads retailers to choose to distribute either one size fits all or a limited number of sizes (Kilney 2003; Faust et al. 2006) in an effort to customize their offer to what they believe is their target market.


777

Adoption of a limited number of size codes (1) allows manufacturers and retailers to limit the number of sizes offered hoping to "adequately" fit large population segments (Ashdown 1998) yet (2) it also opens the door to a phenomenon known as vanity sizing.

Vanity sizing Gruber (2005) pointed out that manufacturers have quietly increased their clothes measurements without changing the size label in a bid to keep pace with the changing American waistlines. "Apparel researchers and manufacturers are quite aware of the use of vanity sizing, downsizing the numerical size label to play to the consumer’s vanity or sensitivity about body size and the desire to be trim and slim" (Labat 2007). Vanity sizing is a marketing tool particularly used by companies selling at higher price points (Staples 1994 in Ashdown 2007, p.102) to flatter the egos of consumers who often feel better about buying smaller sizes (Chun 2007; Bougourd 2007). More than just a marketing tool, it can be a successful strategy to increase brand loyalty. Yet it also contributes to confusion about sizing, complicating the garment shopping experience, creating what some called a chaos.

The sizing chaos Faust et al. (2006) confirmed that manufacturers and retailers: (a) do not abide by national size standards and (b) they target women of specific sizes and silhouettes, either A, X or H shape (Faust 2007). The author also revealed that manufacturers do not always respect the specifications provided by order givers. Yet most still label their products using standard numerical codes such as 6, 8, 10 … 24 although their garment measurements diverged from what these codes are supposed to mean (Workman 1991). Identically size labelled garments have measurements that differ from one manufacturer to another (Faust et al. 2006). Hence Faust’s (2003) observation, that more than 50% of women bring at least two or more identical garments in different sizes to the fitting room, comes as no surprise. Furthermore, labels do not identify the body shape that garments are supposed to fit. From the consumer’s point of view, the result is confusion, dissatisfaction, and a waste of time. For retailers it means an increase of service costs for in-store sales and for online and catalogue sales it means a high numbers of merchandise returned. Kunick (1967) warned that "… if the clothing industry develops the practice of omitting the body measurements from the garment label we shall, inevitably, revert to the sizing chaos of the past" (in Ashdown 2007, p.115).

778


We illustrated this existing chaos according to our understanding of the actual size labelling system (Figure 1). Manufacturers on one side mass produce garments customized to fit their target markets’ measurements and shapes. As one can see, they do not adhere to national standards and sometime they do not respect retailers or other order givers specifications. Moreover they all use the same numerical sizing codes. Consumers on their side, with their own specific shape, have to find which garment suit them best with only a numerical code as for information. This creates more consumer confusion than satisfaction, or again a real chaos.

no

no no

Manufacturer

National Standards

Target

produces

A

A shape

? Manufacturer

Target

One size

produces

X

Small

X

Medium

shape

?

Large

?

Manufacturer produces H

no

no

H Order giver

shape

provides

no

Target

SPEC

Figure 1: Size labelling system.


779

Mass customization Mass customization was first coined by Stan Davis (1987) in his book Future Perfect. Not too long after Joseph II Pine (1993) made the expression "mass customization" fashionable. He describes mass customization as being "… a synthesis of the two long-competing systems of management: the mass production of individually customized goods and services" (Pine 1993, p.48) adding that new technologies would play a key role in a near future. The author brings up the idea that time may have come to discard the mass production paradigm and to move to mass customization. He states that a paradigm blindfolds us making it difficult to process contradicting new information until its preponderance becomes so great that it is almost too late. How can we recognize that the time has come to shift from mass production to mass customization? By searching for market turbulence which the author defines as: "… an imprecise term that denotes the amount of instability, uncertainty and lack of control within a firm’s marketplace" (Pine 1993, p.54). To understand where one stands in this respect, Pine (1993) proposes using Market Turbulence Mapping with such factors as: changing needs and wants, changing customers' demographics, saturation level of product within its marketplace, economic cycles and uncertainties that affect markets, and technological shock that overthrow current dominant design. The author proposes five possible avenues of mass customization:

Customize services around standardized products;

Create customizable products and services;

Provide delivery point customization;

Provide quick response throughout the value chain;

Modularize components to customize end products and services (Pine 1993, p.171). Nowadays there can be no doubt that the apparel industry has gone trough dramatic changes and "turbulence" in the last few decades. Production has moved to Asian countries forcing many Westerners to rethink their business models. New technologies have allowed pervasive modifications in production processes, supply chain management, and distribution.

DesMarteau (2000-04), Gray (1998), Fiore et al. (2001-04), Cardwell and Workman (1993), Robinette et al. (1999) Tae and Sung (2000), and many others proposed that new technology brought us to the threshold of a better garment fit. Istook et al. (2001-02-04) published numerous papers linking 3D body scanning and Computer Aided Design (CAD) to arrive at mass customization in the apparel

780


industry. Koontz and Gibson (2002) talked about a new mix of bricks and clicks. Anderson et al. (2002) published "A Costumer-driven model for mass customization in the apparel market". Lee et al. (2002) studied mass customization acceptance in relation to merchandising issues associated with product, process, and location. Loker et al. (2004) studied female consumers' reactions to body scanning for mass customization and sizing. Loker (2007) recently proposed possible trends for mass customization in the apparel industry: ready-to-wear size customizing, virtual try-on, and market-ofone mass customized garments. Yet she also stated that: "Although initial research has confirmed consumer interest in some applications for apparel sizing (Loker et al. 2004a), Western consumers are used to a shopping experience that includes a variety of immediately available choices, access to the tactile feeling of the fabric and try-on opportunities afforded by ready-to-wear clothing" (p. 259). As one can see, the meaning of customization in the apparel industry has been explore by many researchers, where most of them addressed the issue of customizing mass produced garments by adjusting the measurements or the color for one individual. Drawing on the Total Quality Management (TQM) literature the apparel industry could customize garments with an avenue suggested to be the easiest to start with according to Pine (1993) e.g. marketing customized services over standardized products. "Completely standardized products can be customized before being offered to customers by people of marketing", (Pine 1993, p.172). In short garments can be developed and mass produced by manufacturers as it used to be although before being out on the market, these garments could be labelled for a target group of customers before being delivered to the retailers.

Delivery

Marketing

Production

Development

Figure 2: Customizing services over standardized products (Pine 1993, p.173).

A homogeneous apparel size offer by manufacturers and retailers no longer addresses the heterogeneity of body shapes and sizes in a global world. Yet, as mentioned above, consumers still expect to find "ready-to-wear" on the shelf,


781

ready to be worn, when shopping for apparel. Businesses should therefore, as stated by Ogawa and Piller (2006), integrate customers into a priori innovation process before entering the final development and manufacturing stages, and mass customize at the service end in the marketing phase (Pine 1993) through the size labelling system.

Methodology, Research and Findings The American apparel industry is now in the chaos predicted by Kunick in 1967. To demonstrated that this is a period of turbulence (homogeneity versus heterogeneity in this case on the target market; subjects' shapes and sizes), we first need to understand the processes used in the past to arrive at national standards (homogeneity in sizes) as well as the literature pertaining to the determination of such standards. We then proceed to show the lack of agreement, among academics, about what constituted an acceptable procedure to arrive at size standards; regardless of the up-to-datedness of the database. We also demonstrated that nowadays sizes and shapes are definitely more heterogeneous than homogenous.Finally, we demonstrated, based on our analysis of the most recent U.S.A. anthropometric survey (The National Sizing Survey: "SizeUSA, Let’s size up America 2004), that one of the solution to adequately serve the female apparel consumer is to mass customized a standardized product at the service level offers, as suggested by Pine (1993). We concluded with a proposed new size and shape label.

Existing standards Three major anthropometric surveys have been conducted in the U.S.A. over the last 100 years:

End of the First World War, in 1921, data was gathered on more than 100,000 males during the demobilization.

End of the '30s, a series of Administrative State Work Projects led to publications two decades later by the United States Government Printing Office in Washington, D.C., endorsed by the U.S. Department of Agriculture, sponsored by the Bureau of Home Economics. The two best known were: (1) "Body measurements of American boys and girls for garment and pattern construction: a comprehensive report of measuring procedures and statistical analysis of data on 147,000 American children" (BMABGGPC), by the U.S. Dept. Agr. Misc.Pub. no. 366" and (2) "Women’s Measurements for Garment and Pattern Construction", (WMGPC) from the U.S. Dept. Agr. Misc. Pub. 545.

782


Early 21st Century, the latest American anthropometric campaign SizeUSA, Let’s size up America. Meanwhile, in 1945, the Mail Order Association of America recommended that the industry be provided with scientific body measurement data for the sizing of women’s apparel. The document was titled: Commercial Standards Report (CS151). This recommendation eventually led to the publication of apparel size standards such as the one in 1958, the "Body Measurements for the Sizing of Women’s Patterns and Apparel" commonly called the (CS215-58). A decade later, in 1970, again at the request of the Mail Order Association of America an updated version of the CS215-58 study was published: "Voluntary Products Standards PS42-70, Body Measurements for the Sizing of Women’s Patterns and Apparel", commonly named this time the "PS42-70". Then again, in 1994, an updated version was produced by the American Society for Testing and Materials (ASTM D5585-94), today being used as a reference even in Asia.

Before 1939-41 National Survey researchers' concern was with not having anthropometric data to work with (O'Brien and Shelton 1941). A few decades later they concerned themselves with the obsolescence of the data and the deficiency of its analysis from which size standards emanated. Half a century later, some stated that the biggest weakness in these anthropometric standards and charts was the lack of studies of size and shape distribution of the United States civilian population ([TC]2, 2004). A closer look at these anthropometric surveys helps us to understand how they served the apparel industry. In order to evaluate them, we more or less followed the general steps used in their presentation, summarizing on three important points: (1) purpose, (2) sample and socio-demographic variables, (3) key measurement points, statistical analysis, standards and sizes.

Purpose of study The 1939-1941 WMGPC research’s project was, according to O’Brien and Shelton (1941): "undertaken in order to provide measurements which could be used for improving the fit of women’s garment and patterns. No scientific study of body measurements used in the construction of women’s clothing has ever been reported….there are no standards for garment sizes and retailers and consumers are subjected to unnecessary expense and hazarded by the difficulties involved in obtaining properly fitting clothing." (WMGPC 1941). The 1958 CS215-58 standard was based on the WMGPC research. Its foreword mentions that:


783

"Commercial standards are developed by manufacturers, distributors, and users in cooperation with the Commodity Standards division of the Office of Technical Services and with the National Bureau of Standards. Their purpose is to establish quality criteria, and to provide uniform bases for fair competition." It also specifies (p. flyleaf) that: "The adoption and use of a Commercial Standard is voluntary." Its purpose was "…to provide standards classifications, sizes designations, and body measurements for consistent sizing of women’s ready-to-wear apparel (Misses, Womens, Juniors, etc.) for the guidance of those engaged in producing, or preparing specifications for, patterns and ready-to-wear garments……..to provide the consumer with a means of identifying her body type and size from the wide range of body types covered, and enable her to be fitted properly by the same size regardless of price, type of apparel, or manufacturer of the garment." (CS215-58 p.1) The following quote probably best summarizes the CS215-58’s reason for existence: "…the sizing system proposed provides the means for fitting the maximum number of women with good fitting clothes, without the need for repeated try-ons and expensive alterations." (CS215-58 p.33) As mentioned, an update to this study was published in 1970: the PS42-70. Its objective was to bring the existing standards up to date to better reflect the (then) current female population. The 1994 ASTM D5585-94 standards purpose was als to update sizes used at that period. It was an updated sizing based on the most commonly used sizes in the U.S. industry at that period. No anthropometric survey was conducted. In 2002 the SizeUSA, let’s size up America report states that (p. 7): "The objective of the SizeUSA National Sizing Survey is to measure the body dimensions of a representative sample of the U.S. population [which according to them has never been done]. Until now, the cost of conducting a statistically significant study has been too expensive. With the 3D body scanning technology developed at [TC]2, it became financially feasible." The stated objective of each one of these studies and revisions being to help the apparel industry by providing standardized measurements with, more specifically, ready-to-wear or mass production in mind.

784


Sample and socio-demographic variables In their project, O'Brien and Shelton (1941) stated that a sample may be called representative if it led to results not materially different from those that would have been obtained had the entire population been analyzed. The 1939-1941 WMGPC sample: The sample size of the WMGPC research project was of 14,698 women. All of them were from 7 states within the continental U.S. (Arkansas, California, Illinois, Maryland, New-Jersey, North Carolina and Pennsylvania). Most of the women who participated in the survey were white American women between the ages of 18 and 80. The WMGPC’s report indicated that a number of women were eliminated for different reasons, such as: not being Caucasian, an unidentified date of birth, being under 18 years of age, their measurements were incomplete or erroneous, a physical deformity was identified, or the information arrived in Washington after the deadlines. O'Brien and Shelton (1941) evaluate that this sample used in the WMGPC underrepresented the Middle West and Rocky Mountain areas and overrepresented single women. Numerous authors have since commented on the impact and benefits of this report; yet the original database has proven impossible to trace and thus to access. Since the forties, and until the end of the last Century, no national anthropometric campaign was done. We had to wait the early twenty-first century to get another anthropometric national campaign. The 2002 SizeUSA, let’s size up America sample: All together, more than 10,000 persons (men and women) were scanned. A total of 6310 women participated in the survey. Women were scanned in 13 different states within the continental U.S. (Cary, Columbia, Dallas, Miami, New York, Chattanooga, Los Angeles, San Francisco, Portland, Lawrence, Winston-Salem, Buford and Glendale). On contrary to the WMGPS national campaign, women who participated in the survey were classified according to their ethnicities: Non-Hispanic White, NonHispanic Black, Hispanic and other including Asian-America. Women were also classified according to age group (18-25, 26-35, 36-45, 46-55, 56-65, 66+). As it was the case in the WMGPC survey, some women were eliminated as their measurements were incomplete or erroneous. None were eliminated because of their ethnicity as it was the case in 1939-40. Our analysis showed that this sample did not cover the entire American female population (Tables 2 and 3) or again may not be 100% representative.

Young Non-Hispanic White women (less than 45 years old) were underrepresented while older ones were overrepresented; the opposite being true with the Non-Hispanic Black, Hispanic and Other women.

785


Cross-tabulations revealed that, in certain locations, some ethnicities were overrepresented whereas others were underrepresented:

Non-Hispanic White women were underrepresented on the West Coast (Glendale, CA; Los Angeles, CA) and overrepresented on the East Coast (Buford, GA; Cary, NC; New York, NY).

Other (including Asian-American) women were overrepresented on the West Coast (Los Angeles, CA and San Francisco, CA) and underrepresented in Cary, NC; Dallas, TX; Miami, FL; Chattanooga, TN or Winston-Salem, NC. Table 2: Scan location frequency. Location

Frequency

%

Valid %

Cumulative %

aL1: Loc. 1: Cary

836

13,7

13,7

13,7

aL2: Loc. 2: Columbia

598

9,8

9,8

23,5

aL3: Loc. 3: Dallas

1249

20,4

20,4

43,9

aL4: Loc. 4: Miami

69

1,1

1,1

45,1

aL5: Loc. 5: New York

303

5,0

5,0

50,0

aL6: Loc. 6: Chattanooga

282

4,6

4,6

54,7

aL7: Loc. 7: Los Angeles

304

5,0

5,0

59,7

aL8: Loc. 8: San Francisco

265

4,3

4,3

64,0

aL9: Loc. 9: Portland

263

4,3

4,3

68,3

aL10: Loc. 10: Lawrence

238

3,9

3,9

72,2

aL11: Loc. 11: Winston-Salem

108

1,8

1,8

74,0

aL12: Loc. 12: Buford

740

12,1

12,1

86,1

aL13: Loc. 13: Glendale

845

13,9

13,9

100,0

Total

6098

100,0

100,0

Besides these few minor under or over representation, we believe SizeUSA study was representative of the U.S.A. populations. The first standards published some fifty years ago were based on a relatively homogenous group of Caucasian White American. This sample obviously covered a homogenous population according to its ethnicity whereas the last national survey presents more diversified and heterogeneous American market at least in its origins.It becomes clear that the American market moved from a homogeneous market to a heterogeneous one.

786

HANDBOOK OF RESEARCH IN MASS CUSTOMIZATION & PERSONALIZATION Table 3: Cross-tabulations on ethnic origin, age, and scan location.

Age Group

18-25

26-35

36-45

46-55

56-65

66 and over

Loc. 1: Cary

Loc. 2: Columbia

Count

Non-Hispanic White

Non-Hispanic Black

Hispanic

Other (AsianAmerican incl.)

Count

678

280

148

137

Exp. Count

767.1

255.5

99.1

11.8

Adj. Res.

-5.6

1.9

5.8

2.4

Count

633

282

117

125

Exp. Count

726.9

242.1

93.9

109.7

Adj. Res.

-6.0

3.2

2.8

1.7

Count

676

265

75

104

Exp. Count

692.3

230.6

89.5

104.5

Adj. Res.

-1.1

2.8

-1.8

-.1

Count

666

190

55

80

Exp. Count

605.1

201.6

78.2

91.3

Adj. Res.

4.2

-1.0

-3.0

-1.4

Count

402

52

17

31

Exp. Count

313.9

104.6

40.6

47.4

Adj. Res.

8.0

-6.0

-4.0

-2.6

Count

172

6

5

10

Exp. Count

121.8

40.6

15.7

18.4

Adj. Res.

7.0

-6.1

-2.9

-2.1

Count

510

247

5

33

Exp. Count

464.0

154.6

60.0

70.0

Adj. Res.

3.5

8.9

-8.0

-5.0

Count

416

97

7

59

Exp. Count

334.8

111.5

43.3

50.5

Adj. Res.

7.1

-1.6

-6.0

1.3

787

VOLUME 2: APPLICATIONS AND CASES Table 3 (Continued ) Age Group

Loc. 3: Dallas

Loc. 4: Miami

Loc. 5: New York

Loc. 6: Chattanooga

Loc. 7: Los Angeles

Loc. 8: San Francisco

Loc. 9: Portland

Loc.10: Lawrence

Count

Non-Hispanic White

Non-Hispanic Black

Hispanic


Count

698

297

106

49

Exp. Count

681.0

226.9

88.0

102.8

Adj. Res.

1.1

5.8

2.2

-6.2

Count

37

14

1

0

Exp. Count

33.4

11.1

4.3

5.0

Adj. Res.

.9

1.0

-1.7

-2.4

Count

188

40

4

25

Exp. Count

162.0

54.0

20.9

24.5

Adj. Res.

3.2

-2.2

-3.9

.1

Count

131

124

5

4

Exp. Count

155.8

51.9

20.1

23.5

Adj. Res.

-3.1

11.4

-3.6

-4.3

Count

36

54

90

64

Exp. Count

153.5

51.1

19.8

23.2

Adj. Res.

-14.8

.5

16.8

9.1

Count

149

19

7

65

Exp. Count

142.8

47.6

18.4

21.5

Adj. Res.

.8

-4.7

-2.8

10

Count

77

6

14

21

Exp. Count

85.5

28.5

11.1

12.9

Adj. Res.

-1.4

-4.7

.9

2.4

Count

183

8

9

21

Exp. Count

132.6

44.2

17.1

20.0

Adj. Res.

6.8

-6.2

-2.1

.2

788

HANDBOOK OF RESEARCH IN MASS CUSTOMIZATION & PERSONALIZATION Table 3 (Continued )

Age Group

Loc. 11: WinstonSalem

Loc. 12: Buford

Loc. 13: Glendale

Count

Non-Hispanic White

Non-Hispanic Black

Hispanic


Count

86

18

0

0

Exp. Count

60.6

20.2

7.8

9.1

Adj. Res.

5.0

-.5

-2.9

-3.2

Count

504

127

6

50

Exp. Count

409.0

136.3

52.9

61.7

Adj. Res.

7.6

-.9

-7.2

-1.7

Count

212

24

163

96

Exp. Count

411.9

137.2

53.2

62.2

Adj. Res.

-16.0

-11.5

16.7

4.8

Key measurement points, statistical analysis and sizes In this section we presented the body measurements and statistical tools that were used to arrive at existing size standards along with a criticism of the statistical methods used. The 1939: The WMGPC (1941) final report presented the measuring procedure, the measurement points, and the distribution of the 59 measurements (weight and 58 body measurements points) for the 10,042 women retained (Table 4). Statistical tools used for data analysis were means, medians, modes, standard deviations, coefficient of variation and correlations. As there were no correlations between vertical and horizontal measurements both measurements were analyzed separately. The measurement system, a classification of women on six criteria (state, habitat, nativity, matrimonial status and number of children borne, occupation and family income), and a comparison of women’s measurements with those of girls 15, 16, 17 years old were also presented in the report. No standard sizes were provided, only statistical reports and guidelines.

789

VOLUME 2: APPLICATIONS AND CASES Table 4: List of measurements points for the WMGPC. Vertical measurements

Body measurements

Stature

Sitting height

Chest girth armscye

Elbow girth

Cervical height

Vertical trunk girth

Bust girth

Forearm girth

Bust height

Cervical to waist anterior

Waist girth

Waist height

Anterior waist length

Abdominalextension girth

Abdominal extension height

Shoulder to waist

Hip girth

Anterior chest width

Hip height

Neck to bust

Sitting-spread girth

Highest-bust level width

Sitting spread height

Posterior waist length

Maximum thigh girth

Posterior chest width

Crotch height

Scye depth

Midway thigh girth

Anterior bust arc

Tibiale height

Trunk line

Bent knee girth

Anterior waist arc

Ankle height

Arm length, shoulder to scye

Knee girth at tibiale

Abdominalextension arc

Total posterior arm length

Waist to hip

Maximum calf girth

Posterior hip arc

Upper posterior arm length

Total crotch length

Maximum leg girth

Bust girth over foundation garment

Anterior arm length

Anterior crotch length

Ankle girth

Waist girth over foundation garment

Neck-base girth

Abdominalextension girth over foundation garment

Wrist girth Shoulder length

Armscye girth

Hip girth over foundation garment

Upper-arm girth

Shoulder slope

The 1958: CS215-58: In the CS215-58 report besides weight, 47 body measurement points were retained. Table 5 presents the complete list of measurements for the CS215-58.

790


Table 5: List of measurement points for the CS215-58 and the PS42-70 (all measurements in inches). Girth measurements

Arc measurements

Width and length measurements

Bust

Bust front

Cross-back width

Waist

Waist front

Cross-chest width

Hip

Abdominal front (High hip)

Bust point to bust point

Neck base (Mid-neck)

Hip back

Neck to bust point

Armscye

Scye depth

Abdominal-extension (High hip)

Vertical measurements

Armscye to waist

Sitting spread

Stature (total height)

Waist to hips

Thigh maximum

Cervical height

Shoulder length

Thigh mid

Waist height

Shoulder slope (degrees)

Knee

Abdominal extension height (High hip)

Arm length shoulder to wrist

Calf

Hip height

Arm length shoulder to elbow

Ankle

Sitting spread height (deleted in PS42-70)

Underarm length

Upper arm

Crotch height

Crotch length total

Elbow

Knee height

Crotch length front (deleted in PS42-70)

Wrist

Ankle height

Cervical to center front at waist

Vertical trunk

Sitting height (deleted in PS42-70)

Waist length front Waist length back

In the CS215-58 report, women were first divided into four groups:

misses' sizes varied from size 8 to 22;

women’s sizes varied from size 30 to 42;

half-sizes sizes varied from size 8 ½ to 24 ½; and finally


791

junior sizes which were based on extrapolations of proportions of the Misses' classification and had odd number size designations that varied from size 7 to 19. Each group was divided into three heights: Tall (T), Regular (R) and Short (S), and each height group further divided in three bust-hip groups: Slender ("-" or minus), Average (without any symbol), and Full ("+" or plus). Size numbers and symbols were then combined to arrive at a complete size designation such as:

"14R" for size Misses, 14 bust, regular height and average hip type;

"14T-" for size 14 bust, height tall and slender hip type;

"14S+" for size 14 bust, height short and full hip type. A priori the classification in three heights: Tall (T), Regular (R) and Short (S) was a major point along with separating each height group in three bust-hip groups: Slender, Average, and Full. However, as one can observe from current apparel size labelling, these last information component was dropped and today’s labels only refer to a size (6, 8, 10….) and sometimes a length or height such as petite or tall. The 1970: PS42-70: PS42-70 definitions and measuring methods almost remained the same (Table 5). When the descriptions differ, the PS42-70 new descriptions are in parentheses, Span charts and grading guides were added, e.g.:

Women sizes vary from 34 to 52;

Half-sizes vary from 12 ½ to 26 ½;

Junior Petite sizes vary from 3, 5(P) … 15(P) even 17. Misses Petites and Tall sizes were identified as 6, 8(P), 10(P,T)… 18(P,T), 20(T), 22(T). Three measurements (sitting spread height, sitting height, and crotch length front) were removed. The 1994 ASTM D5585-94 standards: In 1994 ASTM D5585-94 new standards were developed. As Ashdown (1998) stated they were not derived from new anthropometric data but were compiled from designer experience and market observations. They reflect the sizing most commonly used by manufacturers and retail organizations in the U.S.A. at that period. The ASTM D5585-94 consists of ten sizes (2, 4, 6, … to 20) where each size is arrived at with 39 body measurements. Meanwhile north in Canada, the Canadian General Standards Board (1987) CAN/CGBS-49.203-M87, (p.1) states that: "The standard contains a selection from the complete system of sizes which is of greatest commercial interest" and "may be used as a guide in choosing the sizes of women’s wearing apparel."

792


"Although the intended primary application of this standard is stated in its scope, it is important to note that it remained the responsibility of the users of the standard to judge its suitability for their particular purpose." Table 6 shows how sizes and measurements are distributed. Table 6: Sizes and measurements distribution (women’s pants) of the CGSB. Circumferences

Women

Misses

Junior

Sizes CGSB

Waist height (low waist)

Waist height (high waist)

Bust

Waist

Hips

Short legs

Long legs

Short legs

Long legs

5

28 ¾

20 ¾

31 ½

37 ¾

40 ½

39

41 ¾

7

30 ¼

21

33 ½

37 ¾

40 ½

39

41 ¾

9

32

22

35 ½

37 ¾

40 ½

39

41 ¾

11

33 ½

24

37 ½

37 ¾

40 ½

39

41 ¾

13

35 ½

26

39 ¼

37 ¾

40 ½

39

41 ¾

15

37 ½

28

41 ¼

37 ¾

40 ½

39

41 ¾

17

39 ¼

30

43 ¼

37 ¾

40 ½

39

41 ¾

19

41 ¼

32 ¾

45 ¼

37 ¾

40 ½

39

41 ¾

6

30 ¾

22 ¾

31 ½

37 ¾

40 ½

39

41 ¾

8

32 ¼

23 ¾

33 ½

37 ¾

40 ½

39

41 ¾

10

33 ¾

24 ¾

35 ½

37 ¾

40 ½

39

41 ¾

12

35 ½

26 ¾

37 ½

37 ¾

40 ½

39

41 ¾

14

37 ½

28 ¾

39 ¼

37 ¾

40 ½

39

41 ¾

16

39 ¼

30 ¾

41 ¼

37 ¾

40 ½

39

41 ¾

18

41 ¼

32 ¾

43 ¼

37 ¾

40 ½

39

41 ¾

20

43 ¼

34 ¾

45 ¼

37 ¾

40 ½

39

41 ¾

10 ½

34 ¼

25 ¾

33 ½

37 ¾

40 ½

39

41 ¾

12 ½

35 ¾

26 ¾

35 ½

37 ¾

40 ½

39

41 ¾

14 ½

37 ½

28 ¾

37 ½

37 ¾

40 ½

39

41 ¾

16 ½

39 ¼

30 ¾

39 ¼

37 ¾

40 ½

39

41 ¾

18 ½

41 ¼

32 ¾

41 ¼

37 ¾

40 ½

39

41 ¾

793

VOLUME 2: APPLICATIONS AND CASES Table 6 (Continued ) Circumferences

Sizes CGSB

Waist height (low waist)

Waist height (high waist)

Bust

Waist

Hips

Short legs

Long legs

Short legs

Long legs

20 ½

43 ¼

34 ¾

43 ¼

37 ¾

40 ½

39

41 ¾

22 ½

45 ¼

36 ½

45 ¼

37 ¾

40 ½

39

41 ¾

24 ½

47 ¼

38 ½

47 ¼

37 ¾

40 ½

39

41 ¾

26 ½

49 ¼

40 ½

49 ¼

37 ¾

40 ½

39

41 ¾

28 ½

51 ¼

42 ½

51 ¼

37 ¾

40 ½

39

41 ¾

Figure 3: Women’s, ladies’, and junior’s CGSB bust, waist and hip measurements of different sizes, all in inches.

794


Bust, waist and hip graphic measurements distribution were done based on the CGSB’s three standard size classifications. Figure 3 shows how sizes and measurements are linearly distributed for these three groups: women, ladies and junior. Averages, percentiles and basic statistical methods were commonly used in national surveys’ reports and in studies that described how to determine sizing systems and charts from body measurements (Beazley 1997-98-99; Roebuck 1995). According to Beazley (1998) the development of an anthropometric sizing system requires four steps once the sample measurements have been taken (p. 264). An example extracted from Beazley’s paper is presented bellow.

Selection of the intervals for the key dimensions which will establish the sizing categories;

Development, for each size, of all other dimensional data which would be used in the design or sizing of the item;

Conversion of the summary data to an appropriate design value for the end item in terms of fit and function;

Establishment of estimates of the sizing tariff (the proportion of the population that falls within the limits of each size category), for manufacturing of the end item. Table 7: Excerpt from "Chart 5" presented by Beazley (1998; p. 266).

795


The author explains how to divide samples based on measurements. For example, apparel length (height of the wearer) is proportionally divided into short, medium and tall as is it was done in national standards. To do so, one needs to calculate average height and variance; one then retains only the values within two standard deviations covering 95.45% of the population, thus eliminating extreme measurements that can cause distortion. In the author’s example (Table 7), average height was 165cm and range was 30cm. Dividing the range by three she arrived at an interval of 10cm for each of the three groups: "short", "medium" and "tall". Beazley (1998) recommended using the same method to divide bust size, waistline and hip by groups. In these studies (Beazley 1997-98-99), the author compared the measurements obtained to those of commercial size charts. A bust size of 86cm corresponds to a commonly labelled size 12. Then she determined waist and hip measurements to match the same commercial size 12. Afterwards, a linear approach is used to increase or decrease sizes which resemble to figure 3. Table 8 presents a size chart developed by increasing or decreasing the measurement points with regular intervals of 4 centimetres at a time. Table 8: Excerpt from "Table 8" Presented by Beazley (1998; p. 275). Short 155 cm

Medium 165 cm

Tall 175 cm

"Size"

8

10

12

8

10

12

8

10

12

Bust (+4)

78

82

86

78

82

86

78

82

86

Waist (+4)

60

64

68

60

64

68

60

64

68

Hips (+4)

88

92

96

88

92

96

88

92

96

Beazley (1998) also noted that manufacturers who base their size charts on statistical data are rare. They generally prefer regular and consistent intervals. However, as she pointed out, the number of such intervals should be limited to four or five centimetre per size as they can rapidly create distortion from true body measurements. Some authors presented a different point of view. Melzer and Moffitt (1997) described the use of methods based on statistics (percentiles, average, etc.) as the impossible dream. They argued that percentiles are univariate statistics and that problems arise when one tries to use more than one variable at a time, as is the case with body measurements. Whitestone and Robinette (1997) showed that: the sum of percentiles does not equal the percentile of the sum and Zehner et al.

796


(1992) in Melzer and Moffitt (1997) corroborated the difficulty of using statistical approaches showing that the range of people accommodated with percentiles decreases as the number of variables increases. They demonstrated that the actual percentage of people who are within the 5th to the 95th percentile for one variable decreases to 57% when using 6 variables. Table 9: Example of successive reduction in accommodation for each application of 5th to 95th percentile values (Whitestone and Robinette, 1997; chap. 8: 3). Head Length

90% of population

Head Breadth

82% of population

Pupil to Vertex

78% of population

Face breadth

69% of population

Face Length

63% of population

Ear to Vertex

57% of population

Many researchers such as Istook (2002), Rasband and Liechty (2006) also disagreed with the use of a simple statistical approach to sizing, arguing that women have different body shapes. Rasband and Liechty (2006) defined seven figure types (figure 4) besides the "Ideal": Triangular, Inverted Triangular, Rectangular, Hourglass, Diamond, Tubular and Round. They described each figure and provided measurements or ratios for most of them. They stated, for example that a Rectangular figure type has a waist circumference that is 7inches or less smaller than the bust or hip, whereas the Hourglass figure has a waist circumference that is 11inches or more smaller than the bust and hips, and the Triangular figure has small waist with wider hip making it appear that the extra weight goes on the lower torso. Figure 4 give a good visualization of Rasband’s figures. They added that six characteristics need to be looked at to achieve a good fit:

height (petite under 5'4", average from 5'4" to 5'7" and tall over 5'7");

weight;

bone size or structure (referring to height and weight);

proportional body areas;

contour; and

posture.


797

Figure 4: Figure types from "Fabulous Fit Speed Fitting and Alteration" (Rasband and Liechty 2006; cover page).

Ashdown (1998) presented another and different method of using anthropometric data to create size sets. She demonstrated one way to better serve the population using distance templates with key points from anthropometric data allowing the grouping (clustering) of similar subjects. To illustrate this, Ashdown (1998) used the waist, hip, crotch height and length as measurement points. She then presented the comparative results of size measurements according to different grading methods (ASTM’s, linear method and distance template). She also indicated, for all these approaches, the number of people from the sample of 376 she used that would fall into each size category proving her point that distance templates allowed a more equal distribution of subjects. Melzer and Moffitt (1997) concurred with Ashdown stating that sophisticated anthropometric classification schemes such as statistical clustering (McCulloch et al. 1998) were now starting to be employed and that they were sufficient to arrive at well fitting sizes. In brief, the many surveys and reports done and published on the subject confirm apparel manufacturers and retailers' empirical knowledge that using a sizing chart adapted for a population segment or cluster is the best approach. U.S.A. and Canadian laws require that a label disclose what each garment is made of and where it was made; yet no law governs the relation between labelled size of a garment and its actual measurements. Manufacturers and retailers can therefore either adhere to national standards, target a specific market (in terms of silhouette) by adapting their measurements, or use a ploy such as vanity sizing ([TC]2 2004) yet all may use the same generally accepted numerical size labelling systems. This renders the determination of a standard size chart that would please manufacturers, retailers, and consumers quite complex and creates a real marketing challenge (Beazley 1998).

798


ASTM

Linear

Frequency

Waist circumference Inches (increase variably)

Hip circumference Inches (increase variably)

Size

Frequency

Waist circumference Inches (increase of 3.7%)

Hip circumference Inches (increase of 2.6%)

Size

Frequency

Difference inches

Waist circumference inches

Difference inches

Hip circumference inches

Size

Table 10: Data obtained combining Table III, Table IV, and Table IX (Ashdown; 1998).

Distance template

2

34.5

1

24

1

1

A

35.05

24.63

11

A

35.86

25.52

45

4

35.5

1

25

1

10

B

35.97

25.55

27

B

36.95

26.60

59

6

36.5

1

26

1

23

C

36.91

26.49

65

C

37.69

27.39

43

8

37.5

1

27

1

64

D

37.88

27.48

82

D

38.16

27.39

26

10

38.5

1.5

28

1.5

110

E

38.86

28.50

82

E

39.14

29.02

33

12

40.0

1.5

29.5

1.5

89

F

39.88

29.55

50

F

38.77

28.33

37

14

41.5

1.5

31

1.5

43

G

40.92

30.65

28

G

39.36

28.81

40

16

43

2

32.5

2

25

H

41.99

31.78

18

H

39.54

29.20

23

18

45

2

34.5

2

10

I

43.09

32.6

9

I

40.64

30.59

30

20

47

1

J

44.22

34.18

4

J

41.45

31.35

40

36.5

total

376

376

376

Research Results The SizeUSA database contained more than 10,000 subjects with over 200 variables. Our research focused on the 6310 women with their 200 measurements for each of them. Many measurements were considered more or less important for this research such as repetitious (left side, right side) or without significant importance since we focused only on the lower part of the body, therefore were eliminated at the very beginning. We reduced the database to alleger our calculus. We observed a number of correlations between measurements using statistical software such as SPSS, Statistica and Excel. Yet we also noted that none existed between height and circumference measurements, confirming the 1941 WMGPC’s findings; leading us (contrary to Ashdown 1998) to analyze these data


799

separately. The scatter plot in Figure 5 shows potential correlations between some height variables (variables identified with the first letter as being an "h") and clearly shows no correlation exist between height and girth (variables identified with first letter being a "t").

Figure 5: Scatter plot matrix: height and girth distribution (h: Total height; k: Weight in pounds; tB: Bust girth; hB: Bust height; tT: Waist girth; hT: Waist height).

Using a principal component analysis we validated that total height was correlated to most other height measurements (Pearson’s r > 80 %) and not correlated to weight; weight on the other hand, was correlated to circumference measurements (Pearson’s r > 0,85 in most cases). Table 11 provides Pearson’s r coefficient. It has often being said in popular journals that American population is taller. We therefore started by looking at the height variable. O'Brien and Shelton (1941), Rasband and Liechty (2006), as well many retailers separated women into three groups of height. These three height groups were often being named as petite, regular and tall whereas:

petite refers to women smaller than 5'4",

regular identifies women between 5'4" to 5'7" and

tall covers women taller than 5'7".

800

HANDBOOK OF RESEARCH IN MASS CUSTOMIZATION & PERSONALIZATION Table 11: Pearson’s correlation coefficient: height, weight and circumferences correlation.

When separated as mentioned, according to our results, nearly 50% of the U.S.A. female sample is petite (Figure 6). Tall 11%

Regular 41%

Petite 48%

Figure 6: SizeUSA height distribution, Percentage of petite, regular, and tall.

Figure 6 shows that nearly 50% of [TC]2’s sample falls in the petite category while 11% are taller than 5'7". Looking at it this way it obviously seems in contradiction with the popular opinion. On one hand people are saying that the American population is taller and on the other side to have almost 50% of the female population as being petite. Actually, women of youngster generation are probably taller than the previous generation, as our results showed that the highest percentage of tall women falls in the 18 to 45 age group while the highest percentage of petite can be found in the older women group. Almost 50% being petite can be explained by the fact that American population is also getting older. New generations are taller but less numbered than their elders. Our results also showed that ethnicity plays a key role in height measurements. Almost 70% of the Hispanic and other Asian-American women are shorter than

801


5'4" regardless of their age group. Although a relationship exists between total height and inseam, one cannot suggest any generalization to the effect that a short, regular or a tall (as categorized by overall heights) would need an inseam of such or such a length. Table 12 also shows for what can be defined as a regular height (between 5'4" and 5'7") that majority of Non-Hispanic White, Non-Hispanic Black, Hispanic and other Asian-American women have an inseam between 28.51 inches and 31.5 inches. More than 15% of the Non-Hispanic White, Hispanic and other American women in the regular size category have a shorter inseam varying between 27.01 inches and 28.5 inches compared to only 3.26% for Non-Hispanic Black. On the other hand close to 15% of Non-Hispanic Black regular height women have an inseam of between 31.51 inches and 33 inches compared to 3% for Non-Hispanic White, Hispanic American and Asian-American and other women. These results also showed that young regular women’s inseams are generally longer than the women inseams of 55 years old of regular size. Similar results were found for petite and tall. Table 12a: Summary of inseam measurements for regular by ethnicity. Ethnicity Non Hispanic White

Non Hispanic Black

Hispanic

AsianAmerican/Others

x ≤ 24

0.00

0.00

0.00

0.00

24 < x ≤ 25½

0.06

0.00

0.47

0.35

25½ < x ≤ 27

0.57

0.77

1.90

2.09

27 < x ≤ 28½

16.89

3.26

18.01

18.47

28½ < x ≤ 30

49.34

34.17

50.24

50.17

30 < x ≤ 31½

30.11

47.79

26.07

25.78

31½ < x ≤ 33

2.97

13.24

3.32

2.79

33 < x ≤ 34½

0.00

0.77

0.00

0.00

34½ < x ≤ 36

0.06

0.00

0.00

0.00

36 < x

0.00

0.00

0.00

0.35

Total

100

100

100

100

Inseam measurements

802

HANDBOOK OF RESEARCH IN MASS CUSTOMIZATION & PERSONALIZATION Table 12b: Summary of inseam measurements for regular height by age groups. Age groups

Inseam measurements

18-25

26-35

36-45

46-55

56-65

66 and over

x ≤ 24

0.00

0.00

0.00

0.00

0.00

0.00

24 < x ≤ 25½

0.00

0.00

0.36

0.00

0.45

0.00

25½ < x ≤ 27

0.45

0.63

1.46

1.28

0.00

3.08

27 < x ≤ 28½

8.86

11.71

15.88

17.70

26.82

20.00

28½ < x ≤ 30

41.59

47.31

47.81

49.68

49.09

44.62

30 < x ≤ 31½

39.94

35.60

30.84

26.87

21.82

30.77

31½ < x ≤ 33

8.56

4.59

3.65

4.26

1.82

1.54

33 < x ≤ 34½

0.45

0.00

0.00

0.21

0.00

0.00

34½ < x ≤ 36

0.15

0.00

0.00

0.00

0.00

0.00

36 < x

0.00

0.16

0.00

0.00

0.00

0.00

Total

100

100

100

100

100

100

Clearly our results showed that women formed a heterogeneous population on height and inseam measurements but some similarities can be found when grouping them according to certain demographic data. Girth data analysis demonstrated that women’s body shapes also varied according to their ethnicity, geographic location of residence and age. Figure 7 shows the data cloud of the sample population on two these girth variables: waist and hip measurements. One may notice that waist and hip were not correlated. Moreover, hip measurement may fluctuate by as much as 15 inches for a given waist measurement. Subjects with measurements within the generally commercialized sizes (i.e. sizes 4 to 24; waist between 27.5 and 45.5 inches and hips between 36.5 and 53.5 inches) which represent 87.7% of the original sample or 5615 individuals fall within the highlighted rectangle. Further analysis of the data cloud enabled us to cluster women not only according to size, measurements but also to their shapes (Figure 8). The following figure shows that women not only are of different sizes (looking at the data cloud from left to right or horizontally) but also have different body silhouettes or shapes (looking at the data cloud vertically) as expected. Women with a large differential between hip and waist measurements (generally referred to as A shaped) can be found on the upper diagonal while women with a small hip-waist differential (known as H shaped) are found on the lower diagonal.


Figure 7: Sample population data cloud.

Figure 8: Data cloud and clusters of equal amplitude (KA).

803

804


In order to help order givers, manufacturers, retailers, and consumers to identify their target or belonging group, as for size and shape class, a mathematical model based on equal amplitude distribution, or equal effective distribution, was developed. Table 13 provides an example of the actual results of our model using the waist and hip measurements of one subject. Table 13: Size and shape grouping (all measurements are in inches). Rotation

Direct measurements

Rotation

Transf. measurements

axe1

0.707107

0.707107

tT

0.707107

0.707107

axe2

0.707107

-0.707107

tH

0.707107

-0.707107

Average

34.357900

43.117900

Average

0

0

Standard deviation

5.359220

4.985790

Standard deviation

1.3789090

0.3140158

min P

27

37

min P

-1.83849

-0.52678

max P

44

54

max P

2.81555

0.56720

min T

23.642

32.0072

min T

-2.78944

-0.94040

max T

61.8504

71.0173

max T

7.58424

1.32650

Waist

Hip

Axe 1

Axe 2

31.00

40.00

-0.885243

-0.000854

▬▬►

▬▬►

Class KA

Class KE

15

15

In this example we chose one subject. The subject had a waist measurement of 31 inches and a hip measurement of 40 inches. When treated in our mathematical table, our results showed that clustering either with the equal amplitude (KA) method, or with equal effective (KE) method, she (the subject) would fall in what we identified as class fifteen. Since we had eleven classes per diagonal (Figure 8), class fifteen refers as the fourth class from the middle diagonal line. This particular class the fourth size group of our sample represents subjects with what we call an X body shape. Furthermore, according to our understanding of literature review and to our research, women usually search for one specific type of fit, Therefore if, as an example, they like a tight fit they would either choose garment size from the same cluster as they were classified to or from a smaller one (same shape "X" but one size smaller). On contrary women who prefer a loose fit search for a garment size located in the superior class (same shape `X` although one size bigger). In both cases, women searched for garment shaped according to their silhouettes. If they were consider as an "X" shape they searched for an X shaped garment; if they were more as an "H" shape they searched for an "H" shaped garment, finally if their silhouette was more as an "A" shape they searched for garments that suited an "A" shape.


805

In the same figure (Figure 8), we drew out some dots. Dots S1 to S12 identify the specifications provided by one of the biggest Canadian retailers who sales private brand and who act as an order givers in this situation. The representation clearly shows that, as stated above, and demonstrated by Faust et al. (2006) based on data analysis of more than 800 pairs of pants, this retailer focused on one and only one specific shape. It produces for one target market it has in mind. This analysis approach enabled us to match women’s sizes and shapes to manufacturers' and retailers' product measurements and shapes. Furthermore, we can state that it seems that manufacturers and retailers focus on one specific market of their own. Doing so, they (all of them together) do cover the whole spectrum of women body measurements and shape. The biggest concern is that they do not provide clear information to the end user. All of them identify their garment size with either small, medium or large or again with numerical codes such as 6, 8, 10, etc. leaving retailers and consumers to their own perception of what a size 8 should or could represent. We therefore assumed that a new customized size label that would detail measurements and shape would definitively help manufacturers, retailers and consumers at every level as well as being an initial step toward mass customization. This type of label (customized label) is what Pine called: mass customizing at the marketing level "market customized with standardized products" which is no need to recall also the lowest cost of customization. We do not think that manufacturers or retailers should change the way they cut and sew (measurements and shape) or that they should adhere to national standard sizes. We do believe that one way to help the whole apparel industry is only by changing the information provided on the size label. Conclusions We had, in the above text, shown (1) that women come in different sizes and shapes, (2) that the ready-to-wear industry does not adhere to national size standards yet (3) label their products with the agreed upon numerical code regardless of measurements, (4) more then 50% of the women bring at least two sizes of an identical garment to the fitting room and that there is still more than 40% of returns from catalogue order. On the other hand attempts to offer tailormade garments in the most price sensitive market segment (i.e. Levi’s) have not yet succeeded. We can therefore conclude, along with Kunick’s (1967) prediction, that the apparel industry sizing is encountering a high form of turbulence: chaos in the size label. We propose that the ideal approach to adequately serve the market is along the lines proposed by Pine (1993) to market customized services over a

806


standardized product e.g. to adopt a new size labelling system for the apparel industry reflecting variations in both body size and shape. To test the look and acceptance level of a possible "new" label (Figure 9), we conducted a mini-test with 36 women who participated in our pre-test and with 103 women who participated in the validation study: results showed that 87% of the pre-test and 78% of the validation participants respectively voted in favour of visual presentation with measurements mentioned information such as shown in Figure 9.

Figure 9: Potential new size and shape label.

The proposal of a new size and shape label system focuses on diversity and variety. With this new system the apparel industry can discard "one size fits all" and the small, medium and large or any numerical codes label and mass customize their collection simply by using a customized labelling system. A new size and shape label is all the apparel industry needs to jump into real mass customization and toward customer’s satisfaction. We believe that women deserve more information when it comes to size and that the label should provide relevant information. As Charles de Secondat de Montesquieu states in Esprit des lois in 1750: "There are certain ideas of uniformity which sometimes seize great minds, but which invariably strike the petty. They find in them a kind of perfection which they recognize because it is impossible not to discover it; the same weights and measures in commerce… But is uniformity always appropriate without exception?"


807

References Alder, K. (2003). The Measure of all Things, The Seven Year Odyssey That Transformed the World, London, Abacus edition. Anderson, L. J., Brannon, E. L., Ulrich, P. V., Presley, A. B., Worondka, D., Grasso, M. and Stevenson, D. (2001). Understanding Fitting Preferences of Female Consumers: Development an Expert System to Enhance Accurate Sizing Selection. National Textile Center Annual Report, I98-A08, Spring House, PA, 1–10. Anderson, L. J., Brannon, E. L., Ulrich, P. V., Presley, A. B., Worondka, D., Grasso, M. and Gray, S. (2000). Understanding Fitting Preferences of Female Consumers: Development an Expert System to Enhance Accurate Sizing Selection. National Textile Center Annual Report, I98-A08-1, National Textile Center, Spring House, PA, 1–10. Anderson-Connell, L. J., Ulrich, P.V. and Brannon, E. L. (2002). A consumer-driven model for mass customization in the apparel market. Journal of Fashion Marketing and Managament. 6(3): 240–258. Ashdown, S. P. (1998). An investigation of the structure of sizing systems: A comparison of three multidimensional optimized sizing systems generated from anthropometric data with the ASTM standard D5585-94. International Journal of Clothing Science and Technology. 10(5): 324–341. Ashdown, S. P. (2007). Sizing in clothing: Developing effective sizing systems for ready-to-wear clothing. The Textile Institute. Cambridge, England: Woodhead Publishing Limited. 384 p. American Society for Testing and Materials (ASTM D5585-94), webstore.ansi.org/ RecordDetail.aspx?sku=ASTM+D5585-95(2001), Visit 2008. Beazley, A. (1997). Size and fit: Procedures in undertaking a survey of body measurements- Part I. Journal of Fashion Marketing and Management. 2(1): 55–85. Beazley, A. (1998). Size and fit: Formulation of body measurement tables and sizing systems- Part II. Journal of Fashion Marketing and Management. 2(3): 260–284. Beazley, A. (1999). Size and fit: The development of size charts for clothing- Part 3. Journal of Fashion Marketing and Management. 3(1): 66–84. Blecker, T. and Adbelkafi, N. (2006). Complexity and variety in mass customization systems: analysis and recommendations. Management Decision. 44(7): 908–929. Bougourd, J. (2007). Sizing system, fit models and Target markets in Sizing in clothing Developing effective sizing system for ready-to-wear clothing, Edited by S. P. Ashdown. Woodhead Publishing in Textiles, 108–151. British Standards Institution (BSI), www.techstreet.com/info/bsi.tmpl, Visited January 2008. Burns, L. D. and Bryant, N. O. (2002). The Business of Fashion Designing, Manufacturing and Marketing. New York: Fairchild Publications. Caldwell, L. F. and Workman, J. E. (1993). New Product Development: Testing the Concept of Customizes Patterns. Clothing and Textiles Research Journal. 11(4): 1–6. Campbell, L. D. and Horne, L. (2001). Trousers Developed from the ASTM D5586 and the Canada Standard Sizing for Women’s Apparel. Clothing and Textiles Research Journal. 19(4): 185–193. Canadian General Standards Board (1984). Pants, Junior, Misses and Women’s Sizes - Dimensions, Office des normes générales du Canada, CAN/CGSB-49.211-M84. Ottawa, Canada. Canadian General Standards Board (1987). Canada Standard sizes for women’s apparel - trade sizes Office des normes générales du Canada, CAN/CGSB-49.203-M87. Ottawa, Canada.

808


Canadian General Standards Board (1987). Canada Standard sizes for women’s apparel - trade sizes. Office des normes générales du Canada. Ottawa, Canada. Canadian General Standards Board (1987). Supplement No. 1 to CAN/CGSB-49.203-M87 Canada Standard Sizes for Women’s Apparel - Trade Sizes. Office des normes générales du Canada, CAN/CGSB-49.203-M87. Ottawa, Canada. Canadian General Standards Board (1997). Canada Standard sizes for women’s apparel - trade sizes. Office des normes générales du Canada, CAN/CGSB-49.203-M87. Ottawa, Canada. Canadian General Standards Board (1997). Canada Standard sizes for women’s apparel - trade sizes. Office des normes générales du Canada, CAN/CGSB-49.203-M87. Ottawa, Canada. Chamberland, A., Carrier, R., Forest, F. and Hachez, G. (1997-98). Anthropometric survey of the Land Forces, Final Report, Defence and Civil Institute of Environment medicine (98-CR-15). Ontario: Department of National Defence. 275 p. Chun, J. (2007). Communication of sizing and fit in Sizing in clothing Developing effective sizing system for ready-to-wear clothing, Edited by S. P. Ashdown. Woodhead Publishing in Textiles, 220–245. Chun-Yoon, J. and Jasper, C. R. (1996). Key Dimensions of Women’s Ready-to-wear apparel: Developing a Consumer Size-Labelling System. Clothing and Textiles Research Journal. 14(1): 89–95. DesMarteau, K. (2000). Let the fit revolution begin. The Bobbin magazine. October, 42–56. DesMarteau, K. and Speer, J. K. (2004). Entering the Third. Apparel, The Bobbin magazine. January, 45(5): 28-33. Diffrient, N., Tilley, A. R. and Bardagjy, J. C. (1974). Human scale 1 / 2 / 3. Cambridge: Massachusetts Institute of Technology, The MIT Press. Eckman, M., Damhorst, M. L. and Kadolph, S. J. (1990). Toward a model of the in-store purchase decision process: Consumer use of criteria for evaluating women’s apparel. Clothing and Textiles Research Journal. 8(2): 13–22. Faust, M-E. (2003). L'utilisation des technologies de l'information et de la communication (TIC) lors de la fonction essayage vestimentaire. M.Sc.A. Ecole Polytechnique de Montreal, Quebec, Canada. Faust, M-E., Carrier, S. and Baptiste, P. (2006). Variations in Canadian women’s ready-to-wear standard sizes. Journal of Fashion Marketing and Management. 10(1): 71–83. Faust, M. E., Carrier, S. and Baptiste, P. (2006b). "Introducing a new labelling system for women’s readyto-wear", POMS Boston, April 30 – May 2. Faust, M. E., Carrier, S. and Baptiste P. (TBP). Women’s wear sizing: an analysis of the Canadian situation on pants' sizes (part II), being reviewed for publication by the Journal of Fashion Marketing and Management. Feather, B. L., Herr, D. G. and Ford, S. (1997). Black and White Female Atletes' Perceptions of their Bodies and Garments Fit. Clothing and Textiles Research Journal. 15(2): 125–128. Fiore, A. M. (2002). Effects of experiential pleasure from a catalogue environment on approach responses toward fashion apparel. Journal of Fashion Marketing and Management. 6(2): 122–133. Fiore, A-M., Lee, S-E., Kunz, G. and Campbell, J. R. (2001). Relationships between optimum stimulation level and willingness to use mass customization options. Journal of Fashion Marketing and Management. 5(2): 99–107. Fiore, A. M., Lee, S-E. and Kunk, G. (2004). Individual differences, motivations, and willingness to use a mass customization option for fashion products. European Journal of Marketing. 38(7): 835–849. Glock, R. E. and Kunz, G. I. (2000). Apparel manufacturing: sewn product analysis (3 Ed.), New Jersey, Upper Saddle River: Prentice Hall. 689 p.


809

Goldsberry, E., Shim, S. and Reich, N. (1996). Women 55 Years and Older: Part I. Current Body Measurements As Contrasted to the PS 42-70 Data. Clothing and Textiles Research Journal. 14(2): 108– 120. Goldsberry, E., Shim, S. and Reich, N. (1996). Women 55 Years and Older: Part II. Overall Satisfaction and Dissatisfaction with the Fit of Ready-to-Wear. Clothing and Textiles Research Journal. 14(2): 121– 132. Gould-Decauville, P., Bruere, C., Uhalde-Roux, C. and Khatar, L. (1998). Guide pratique des tailles dans 36 pays. Tome I, Etudes bimestrielles La Vigie internationale du vêtir- textile (2e édition). Clichy, France: Fédération de la maille. 98 p. Gould-Decauville, P., Bruere, C., Uhalde-Roux, C. and Khatar, L. (1999). Guide pratique des tailles dans 36 pays. Tome II, Etudes bimestrielles La Vigie internationale du vêtir-textile (2e édition). Clichy, France: Fédération de la maille. 98 p. Gray, S. (1998). In Virtual fashion, IEEE Spectrum. 35(2): 18–25. Gruber, R. (2005). Clothing stores shrink size labels to lure vain shoppers by Ben Whitford, jscms.jrn.columbia.edu/cns/2005-04-19/whitford-vanitysizing/ Hamel, C. and Salvas, G. (1992). C'est moi, ma personnalité, mon style, Québec: Éditions Communiplex. 310 p. Hart, C. and Dewsnap, B. (2001). An exploratory study of the consumer decision process for intimate apparel. Journal of Fashion Marketing and Management. 5(2): 108–119. Horne, L., Campbell, L. and Scholz, C. (1999). Older females as a market segment for well-fitting clothing. Journal of Fashion Marketing and Management. 3(3): 236–244. ISO (1976). www.standardsglossary.com, The ISO Standard Glossary Istook, C. L. and Hwang, S-J. (2001). 3D body scanning systems with application to the apparel industry. Journal of Fashion Marketing and Management. 5(2): 120–132. Istook, C. L. (2002). Enabling mass customization: computer-driven alteration methods. International Journal of Clothing Science and Technology. 14(1): 61–76. Istook, C. L. (2004). The virtual and Digital World of Textiles and Apparel. View 18 July 2004, www.tx.ncsu.edu/jtatm/ Kadolph, S. J. (1998). Understanding Standards and Specifications. In Quality assurance for textiles and apparel (Ch. 3 pp. 41-53). New York: Fairchild Publications. 608 p. Kerin, R. A., Hartley, S. W. and Rudelius, W. (2007). Marketing The Core. New York, McGrawHill/Irwin, International Edition, 2nd Edition. 127 p. Kinley, T. R. (2003). Size Variations in Women’s Pants. Clothing and Textiles Research Journal. 21(1): 19–31. Koontz, M. L. and Gibson, I. E. (2002). Mixed reality merchandising: bricks, clicks and mix. Journal of Fashion Marketing and Management. 6(4): 281–395. Kunick, P. (1967). Sizing Pattern Construction and Grading for Womens' and Childrens' Garments, 68 p. in Schofield & Labat LaBat, K. L. and Delong, M. R. (1990). Body Cathexis and Satisfaction with Fit Apparel. Clothing and Textiles Research Journal. 8(2): 43–48. LaBat, K. L. (2007). Sizing standardization, in Sizing in clothing Developing effective sizing system for ready-to-wear clothing, Edited by S. P. Ashdown. Woodhead Publishing in Textiles. 88–107.

810


Lee, S.-E., Kunz, G. I., Fiore, A. M. and Campbell, J. R. (2002). Acceptance of mass customization of apparel: Merchandising issues associated with preference for product, process, and place. Clothing and Textiles Research Journal. 20(3): 138–146. Lennon, S. J. (1992). Categorization as a Function of Body Type. Clothing and Textiles Research Journal. 10(2): 18–23. Loker, S., Cowie, L., Ashdown, S. and Lewis, V. D. (2004). Female Consumers' Reactions to Body Scanning. Clothing and Textiles Research Journal. 22(4): 151–160. Loker, S. (2007). Mass customization and sizing, in Sizing in clothing Developing effective sizing system for ready-to-wear clothing, Edited by S. P. Ashdown. Woodhead Publishing in Textiles. 246–263. McCulloch, C. E., Paal B. and Ashdown, S. P. (1998). An optimization approach to apparel sizing. Journal of the Operational Research Society. 19: 492–499. Mellian, S. A., Ervin, C. and Robinette, K. M. (1991). Sizing Evaluation of Navy Women’s Uniforms (U). AL-TR-1991-0116, ADA249782, Air Force System Command, Wright-Patterson Air Force Base, Ohio, 45433-6573. U.S. Department of Commerce National Technical Service. Melzer, J. E. and Moffitt, K. (1997). The head-mounted display designing for the User, (HMD). In Whitestone, J.J. & Robinette, K.M. (1997). Fitting to Maximize Performance of HMD Systems. 175–206. New York: McGraw Hill Publishers. or "Computerized Anthropometric Research & Design Laboratory (CARD)", Ch. 8, 23 p. O'Brien, R. and Shelton, W. C. (1941). Women’s measurements for Garment and Pattern Construction. Bureau of Home Economics, Textiles and Clothing Division, Miscellaneous Publication, No. 454, US Department of Agriculture and Work Projects Administration, Washington, DC. Ogawa, S. and Piller, F. T. (2006). Reducing the Risk of New Product Development. MITSloan Management Review. Winter 2006: 47(2): 65–71. Otieno, R. (2000). The role of garment sizing in creation of customer satisfaction: Indications from focus group responses. Journal of Fashion Marketing and Management. 4(4): 325–335. Pine, J. II (1993). Mass customization, The new frontier in Business competition. Boston: B. Harvard Business Press. 333 p. Rasband, J. (1994). Fabulous Fit, New York: Fairchild Publications, 176 p. Rasband, J. A. and Liechty, E. L. G. (2006). Fabulous fit Speed Fitting and Alteration, (2e edition). New York: Fairchild Publications. 432 p. Reda, S. (2000), "VeriFone and Russell Reynolds Associates top 100 Internet retailers", Stores, www.stores.org/archives/sept00cover.html, Rifkin, G. (1995). Levi Strauss Buys Custom-fit Software Concern, New York Time, query.nytimes.com/gst/fullpage.html?res=990CE2DF1E3DF935A35753C1A963958260 (visited 2008) Robinette, K. M., Daanen, H. and Paquet, E. (1999). The Caesar Project: A 3-D Surface Anthropometry Survey, Conference Paper 3D Digital Imaging and Modeling, Second International Conference, October 1999, Ottawa Canada. 380–386. Roebuck, J. A. L. (1995). Anthropometric methods: Designing to Fit the Human Body, Roebuck Research and Consulting. Monographs in Human Factors and Ergonomics. Santa Monica, California: Alphonse Chapanis, Series Editor. 194 p. Schofield, N. A. and LaBat, K. L. (2005). Defining and Testing the Assumptions used in current Apparel Grading Practice. Clothing and Textiles Research Journal. 23(3): 135–150. Schofield, N. A. and LaBat, K. L. (2005). Exploring the relationships of Grading, Sizing, and Anthropometric Data. Clothing and Textiles Research Journal. 23(1): 13–27.

811


[TC]2 (2004). Size USA Let’s size up America… The National Sizing Survey: Body Measurement and Data Analysis Reports on the U.S. Population Report. Cary, North Carolina: Prepared by [TC]2. 134 p. [TC]2 (2004). Size USA The US www.tc.2.com/what/sizeusa/index.html

National

[TC]2 (2004). 3D Body Scanner www.tc.2.com/products/body_scanner.html

Size

Survey.

Specifications.

Visit

Visit

March

March

18 18

2004, 2004,

[TC]2 (2004). Measurement Extraction Software and Size Prediction Development. Visit March 18 mars 2004, www.tc.2.com/products/extraction_software.html Tae, J. K. and Sung, M. K. (2000). Development of Three-dimensional apparel CAD system Part I: flat garment pattern drafting system. Int. Journal of Clothing Science and Technology. 12(1): 26–38. Tae, J. K. and Sung, M. K. (2000). Development of Three-dimensional apparel CAD system Part II: prediction of garment drape shape. Int. Journal of Clothing Science and Technology. 12(1): 39–49. Tae, J. K. and Sung, M. K. (2000). Optimized garment pattern generation based on three-dimensional anthropometric measurement. Int. Journal of Clothing Science and Technology. 12(4): 240–254. Ulrich, P. V., Anderson-Connell, L. J. and Wu, W. (2003). Consumer co-design of apparel for mass customization. Journal of Fashion Marketing and Management. 7(4): 398–412. United States Department of Commerce National Bureau of Standards (USDCOTS) (1958). Commercial Standard CS215-58 Body Measurements for the Sizing of Women’s Patterns and Apparel, A recorded voluntary standard of the trade, United States Government Printing Office, Washington, DC. United States Department of Commerce National Bureau of Standards (USDCNBS) (1971). Voluntary product standard PS42 70, Body measurements for the sizing of women’s patterns and apparel, United States Government Printing Office, Washington, DC. Whitestone, J. J. and Robinette, K. M. (1997). Fitting to Maximize Performance of HMD Systems or Chapter 7 of Head-Mounted Displays: Designing for the User, In Melzer, J. & Moffitt, K., (pp. 175-206). New York: McGraw Hill Publishers. or "Computerized Anthropometric Research & Design Laboratory (CARD)", Ch. 8, 23 p. Winks, J. M. (1997). Clothing Sizes International Workman, J. E. (1991). Body Measurement Specifications for Fit Models as a Factor in Clothing Size Variation. Clothing and Textiles Research Journal. 10(1): 31–36. Workman, J. E. (1991). Body Measurement Specifications for Fit Models as a Factor in Clothing Size Variation. Clothing and Textiles Research Journal. 10(1): 31–36. Workman, J. E. and Lentz, E. S. (2000). Measurement Specifications for Manufacturers' Prototype Bodies. Clothing and Textiles Research Journal. 18(4): 251–259. Yertutan, C. (2001). Problems of Young People Related to Anthropometric Measures of Ready-to-Wear Clothes. Journal of Qafqaz University, Ch. 8. 7p. Yoo, S., Klan, S. and Rutherford-Black, C. (1999). Petite and tall-sized consumer segmentation: Comparison of fashion involvement, pre-purhase clothing satisfaction and clothing needs. Journal of Fashion Marketing and Management. 3(3): 219–235. Yu, W., Fan, J., Ng, S-p. and Gu, H-B. (2006). Female Torso Mannequins with Skelton and Soft Tissue for Clothing Pressure Evaluation. In Thermal Manikins and Modeling, Sixth international thermal manikin and modeling meeting (613M) 194–201. Hong Kong: Edited by J. Fan The Hong Kong Polytechnic University.

812


Author Biographies Dr. Marie-Eve Faust research interest focuses on anthropometric data analysis to help manufacturers improve their size gradations and progress in mass customization applications; while working to overhaul the apparel size label system. She is Assistant Professor for Fashion Merchandising at the School of Business Administration at Philadelphia University. Besides she used to lead the ITC’s Global Fashion Management Master Degree seminar in Hong Kong, as with joint-partners schools FIT and IFM she partake in New York and Paris seminars. Dr. Faust is assistant professor at Philadelphia University in Fashion Merchandising. Before this Philadelphia position and her appointment in Hong Kong she taught at l’École supérieure de mode de Montréal and at l’Université du Québec à Montréal (UQAM) Management Faculty. She also held various positions in Canadian enterprises and business. Dr Faust holds a B.A.A. (Major in Accounting), a B.A. (Fashion Design & Management), from UQAM, a M.Sc.A. and a Ph.D. (Industrial Engineering) from l'Université de Montréal, École Polytechnique. Contact: [email protected] Serge Carrier has worked in the private sector both as a manager and as a consultant before joining academia. He is currently the Director of the École supérieure de mode de Montréal, a component of the Université du Québec à Montréal. His research interests centre on the worlds of fashion and the service industries. He has, in the course of his career, published or co-published more than thirty books, articles and monographs. Contact: [email protected]

2.4

Developing Considerate Design: Meeting Individual Fashion and Clothing Needs Within a Framework of Sustainability Sandy Black London College of Fashion, University of the Arts London, United Kingdom

Claudia M Eckert Department of Design, Development, Environment and Materials, Open University, United Kingdom

This paper addresses The Fashion Paradox – the economic importance of the fashion industry set against its inherent obsolescence and waste through constant change. A new methodology is being developed for designers to approach these complex problems, and to evaluate the impact of design decisions through the development of personalized fashion products. A new "Considerate Design" process model is being created through analysis of practical design processes by transferring tools and methods from engineering design within a framework of sustainability. This project responds to the rapidly changing context of fashion and positions the user at the centre of the design process. It will produce personalized fashion products using 3D body scanning and rapid prototyping techniques integrated with different production processes: the direct 3D manufacturing of seamfree knitwear; the making of bespoke hand-crafted bags, and the direct digital creation of body conformable seamless textile structures. The sub-projects each represent different levels of the industry, to test economic viability of products which are individually tailored to requirements, contributing to the development of mass customization. Considerate Design will reduce environmental impact of fashion products and consider both the end user and the entire product life cycle.

Considerate Design and "The Fashion Paradox" Before mass production, people had far fewer clothes which they valued and loved, often made themselves or made specifically for them, which were kept and maintained for many years by repairing and remodeling. With the rise of ready-towear fashions and the consequent demise of home dressmaking, the attention to personalized fit became the domain of a privileged few, as sizes were "standardized" by different manufacturers in different ways. In the UK, companies were still using size data gathered in the 1950s until the National Sizing Survey of 2001-2 established the new size and shape of the nation using 3D body scanning 813

814


technology (Bougourd et al. 2000; Bougourd et al. 2004; Treleavan 2004). Size information, is however interpreted by manufacturers according to their own production and economic values and constraints, with a bewildering array of brand-specific fit and sizing still leaving a great many individuals unable to find clothing which meets their needs. At the same time, globalization of production, increased competition and consumer demand have resulted in accelerated fashion cycles which in turn have led to a culture of "fast" and disposable fashions. The result is also a decrease in prices- clothing is relatively far cheaper than in previous decades (whereas housing, travel and transport costs have risen) – but at what ethical and environmental cost? Fashion’s inbuilt obsolescence is intrinsically unsustainable, but the desire for fashionable renewal is an inherent cultural construct; fashion is also a powerful economic driver, sustaining global industry and employment – a contradiction at the heart of contemporary fashion consumption which Black (2006, 2007, 2008) has termed "The Fashion Paradox" (A theme of the Designing for the 21st Century research cluster Interrogating Fashion: Practice, Process and Presentation. New Paradigms for Fashion Design in the 21st Century. www.interrogatingfashion.org). Considerate Design is an emerging concept which aims to reconcile consumer needs with the environmental impact of consumerism, and empower the designer to balance the often conflicting priorities and issues within the design process and the nature of fashion itself. Scope of Considerate Design Considerate Design for Personalized Fashion Products is a design-led project developed in response to the complex issues and critical debates surrounding fashion (See for example Interrogating Fashion above and the report Well Dressed? The present and future sustainability of clothing and textiles in the UK, Allwood et al., University of Cambridge Institute for Manufacturing UK 2006), due for completion summer 2009. The authors have, since developing their own concept, made contact with a similarly-named project in Sweden for the collaborative design of specific work-based and public interiors, such as nurseries, using a methodology which also addresses aesthetics, ecology, environment and economics (Refer to www.designmedomtanke.com). However, the Considerate Design process discussed in the present paper is differentiated by its focus on the potential for mass customization and personalization to offer solutions for more sustainable and local fashion; and by its function as a tool offering specific guidelines for the designer. A useful exchange between the two projects is however ongoing. In the context of this research, Considerate Design develops original methodologies for design and production of fashion products (exemplified in case studies of

815


fashion knitwear and fashion accessories) which are personalized to fit individual body shape, bringing the customer into the centre of the fashion design process, akin to user-centred design processes in product design disciplines. Importantly, this project addresses the major issue of sustainability in fashion through the creation of a new model for the Considerate Design process which a) considers the users needs, b) considers the environmental impact of materials and production methods, and c) considers the entire product lifecycle, including aftercare and disposal. Constructed as a proof-of-concept portfolio of three case studies, the project brings new approaches to different fashion practices, integrating both traditional and new technologies together with theoretical concepts from engineering design. The Considerate Design concept makes new links between sustainability, personalization and costs within the fashion design and production process (Figure 1). The concept is tested through three different case study scenarios: industrial production of personalized knitwear, bespoke handmade bags, and experimental pseudo-textiles using rapid manufacturing technology.

Sustainability

Considers design effort and economics of production

Costs

Considers environmental impact through the life cycle

Considers

Personalisation end user

Figure 1: Triangle of relationships in Considerate Design.

A two-fold approach is adopted to assist at different scales within the fashion industry: (1) for large scale manufacturing to compare costs and tasks, process modeling, using the P3 Signposting software tools developed by the Engineering Design Centre at Cambridge University (Refer to www-eng.cam.ac.uk/p3/) (Clarkson and Hamilton 2000), is adapted to the fashion industry (discussed in later section); and (2) environmental impact analysis using a simple accessible tool to identify and assist decision making, aimed at designers in small or larger companies. Practitioners therefore engage with theory through the analysis of their processes and guidelines for best practice, and theorists directly engage with the practical design process in the three case study contexts, which operate within very different market levels of the fashion industry, from mass production to bespoke.

816


The Need for Individualized Fashion The nature of fashion and its stylistic and cultural cycles results in fast changes driving ever faster industry response requirements. Underlying both the acceleration of fashion and mass customization is the desire to keep up with fashion, whilst simultaneously expressing individuality within the space of current fashionability. In order to satisfy an increasingly discerning and design-aware consumer looking for individuation in their choices of fashionable products, the concept of customization in clothing has developed strongly in the last decade, with demand for more personalized products forecast to continue rising. The market for fashion accessories has also seen continual growth over the same period as specific items offer differentiation or the new exclusivity of limited editions. With the emergence in the last decade of "fast fashion" retailers such as Zara and Top Shop, the strong dominance of seasonal collections is breaking up, and the power of buyers as mediators between customers and designers is decreasing (Eckert and Stacey 2001). There are multiple potential markets for individualized fashion as the mainstream fashion industry does not sufficiently address the special needs of niche groups – such as non-standard height, larger sizes, older children, the elderly or disabled, who do not want to be excluded from fashion. As purchasing behavior changes, design processes must evolve to respond: designers across different and multiple market sectors need to engage more directly with the tastes and requirements of their customers, necessitating significant changes. Ever accelerating fashion drives consumerism, but if people cannot find what they want, or if products are cheap, garments and textiles are kept for less time, increasing environmental impact from both production processes and waste. The growth of "value brands" and supermarket clothing in the UK has continued to drive prices down, with greater pressures passed further down the supply chain by turns to garment manufacturers, then fabric and fibre producers. This is often at the expense of ethical and environmental standards as shorter lead times are demanded and pressure is put on factories and producers (Refer to reports from campaign groups Labor Behind the Label, Fashioning an Ethical Industry, Clean Clothes Campaign, Pesticide Action Network, and the Environmental Justice Foundation. www.fashioninganethicalindustry.org). Traceability in the supply chain is now becoming increasingly important to the consumer of clothing, in the same manner as seen recently in the food industry with the growth of the organic market – some food companies have now begun to label the environmental impact of their products; American footwear brand Timberland trialled their Green Index scheme in a range of footwear in 2007, with the aim of labelling all footwear by


817

2015 (Timberland company presentation at the RITE conference London, 10th October 2007; www.timberland.com/corp/index). Technology and Fashion Although aspects of the fashion production chain, such as cutting and handling, are increasingly automated, most clothing and footwear is still manufactured manually, using craft skills and the sewing machine. However, research including European funded projects LEAFROG and SERVIVE, is beginning to address new processes for both automated and personalized garment manufacturing. The project LEAPFROG (European Technology Platform for the Future of Textiles and Clothing) examines production processes in the fashion industry and aims for the complete automation of the clothing production line integrating robotic systems and new technologies. www.leapfrog-eu.org . SERVIVE (Service Oriented Intelligent Value Adding network for Clothing-SMEs embarking in Mass customization) develops aspects of LEAPFROG for personalized service in clothing. A paradigm shift is taking place from 2D to 3D design and manufacturing processes as different technologies begin to impact, such as 3D visualization (eg Optitex system), direct 3D rapid manufacturing from 3D computer modeling software (eg Maya) and 3D body and foot scanning (eg Human Solutions, TC2 systems). Another example is advanced knitting technology that produces complete 3D knitted garments directly from the machine (eg Stoll and Shima Seiki). Seamless knitted products have already entered the lingerie and bodywear markets, but have been slow to enter mainstream knitwear production, due to cost and complexity factors. There is recognised potential for the use of this technology as a personalized service, with trial boutiques opened by leading Japanese knitting machine builder Shima Seiki, (Shima 2005). However, it has taken over 10 years for technical and design skills to begin to match machine capability in seamless knitwear (Black 2002, 2005; Sayer et al. 2006). The uptake of 3D body scanning technology into the clothing and fashion industries is starting to impact the area of made-to-measure and mass customization. Mass customization is gaining momentum as a viable possibility for the fashion industry – answering simultaneously the need for personalized product differentiation, previously only found in bespoke fashions, and cost efficiency of mass production (Heyd 2004). The intangible benefits of feeling and looking better in a better fitting garment can be made possible by linking body scanning to production systems and offering a quasi-bespoke customer service for more accurately targeted markets (Bougourd 2006). In addition, there are anticipated gains in minimising waste, reduced production and warehousing costs, and in

818


localised manufacturing (Walker 2006 for full discussion of contemporary thinking on sustainable design). The Spectrum of Customization The range of variables in the production of both basic clothing and seasonal fashion is still relatively high compared to mass production in other industries. The same product design, say a sweatshirt, or a fashion skirt, will be manufactured in a number of different colorways and across a range of sizes as a matter of course in each production run, the actual volume of which depends on the level of the market concerned. Supermarkets and high street brands may manufacture perhaps a thousand per style variation; a designer label manufactures in hundreds or even dozens; a start-up design company may operate in very small batch production, whereas a bespoke service works to a market of one. This fragmented production spectrum for ready-to-wear fashion, coupled with fast changing styles, has contributed to the difficulty in finding automated solutions and maintained the clothing industry as one of manual skilled production. Prices generally reflect volume – lower prices reflecting larger scale production – but as volume decreases the proportion of design and service cost as well as direct manufacturing cost rises dramatically. There are also price differential anomalies in the marketplace – a garment such as a tee shirt for example may be a very low priced bulk commodity item or sold at designer level with higher production values and different supply chain. Customization can be understood on several levels. A basic level of customization may start with a color preference in a sweater for example, or the addition of decorative but non-structural features (such as embroidery on jeans). Fashion and clothing ranges offer color and size selection in predetermined ratios, with information gathered at point of sale used to trigger "quick response" deliveries from the manufacturer, passing on the onus to hold stock from retailer to supplier. At the other end of the spectrum of customization is the hand-made bespoke suit or the couture evening gown, where tailored fit, luxurious fabric choice and dedicated service are at a premium resulting in exclusivity. These high quality luxury items are also likely to be kept and maintained longer than ready-to-wear items, and often handed down through generations. Between these two extremes, other working models may include for example, a wider range of standard offering in sizing, procedures for adapting standard designs, or the personalized garment.


819

Considerate Design The Considerate Design concept is a long term research agenda to change the mindsets of consumers and professionals in the fashion industry, which has started with a specific research project by the same name, to develop and test ideas. The aims of considerate design Considerate Design aims to exemplify a manageable approach to the paradoxical issues of transience and sustainability in fashion and clothing. It puts the emphasis on the ability of design and designers to influence the thinking and product development processes within fashion companies, creating a critical agenda at the core of fashion manufacturing and retail. The key research question to emerge is: Can the desire for new fashion be met more sustainably, through personalized fashion, harnessing emerging technologies and "Considerate Design"? Commercial success depends on high quality design and effective design processes which raises the question: how can an effective design process for Considerate Design be defined, and individual designers be empowered to improve processes through a Considerate Design culture? Questions addressing the role of the individual designer through Considerate Design are: what does this mean for me and how can I contribute? Approach of the project The meta project is structured as a portfolio of three case studies with the overall aim to develop the Considerate Design concept and test new design and production methods for creation of personalized fashion products – but using different technologies. Whilst the end products diverge, there is commonality in the use of individual bodyscan data and rapid manufacturing technology to derive a new methodology for the particular design process in collaboration with the customer. Selection of participants throughout each case study will cover a diverse range of body sizes, shapes, ages and abilities in order that considerate design is also inclusive. Through observation, interview and analysis, the process steps will be modeled to understand the observed design and production activity and to develop a generic process model, which could then be applied to other industry sectors. On a scale that assesses user needs against the degree of innovation, the three case studies within the present project each represent a distinctive position between the two parameters. These project scenarios have been selected to occupy different levels of the market: 1) the knitwear project will produce individual garments but using standard industrial technology for mass production; 2) in the second project

820


bespoke bags are individually handmade, produced in small batches, allowing for ease of customization; 3) the pseudo-textile project is experimental with no predefined market, but will create one-off pieces suitable for use as accessories at high designer level and couture in the first instance, with potential for more accessible markets. These sub-projects will be discussed in the following section The projects aims to develop tools that designers can use to achieve a "considerate design" in a two-pronged approach. An analysis of the multitude of factors which affect the considerate design process feeds into the development of a simple tool, which allows designers to compare the considerate design merits of different designs, as discussed below under the section Outlook. The project utilizes successful tools and techniques from the aerospace and automotive industry to assess the economical viability of offering customized products and aid designers in selecting the suitable customization strategy as discussed. Case Studies These three sub-projects have been deliberately selected as they represent different parts of the fashion market and different products: mass production in the knitwear; craft based design in the hand bags; and cutting edge innovation in the rapid prototyping. Knit to fit The project develops seamless garment knitting for comfort and personalized fit utilizing advanced knitting technology. The new paradigm here is the use of 3D body scan data for extraction of precise body measurements and translation into existing 2D CAD design systems integrated with the industrial knitting machines using Stoll 3D knitting technology. Utilizing measurement analysis for stretch garment fit (Watkins 2005), the design process and fit parameters are compared between on-body fit and a rapid manufactured actual size torso from the 3D body scan as a fit form, and between manual and scanned data. Comparative costings are made between customized 3D and fully fashioned knitting, and between the costs for different customized styles. Applications are in knitwear for fashion, sportswear, wellbeing and potentially in medical healthcare for monitoring of patients. Two recent European projects, WEALTHY and MyHeart have developed prototypes for medical monitoring, utilizing the flexible structures of knit construction. The commercially available Numetrex knitted sports top incorporates sensing of heart rate. This case study contributes to Considerate Design by personalized garment fit to user requirements, efficient use of materials by eliminating cutting waste, reduction in labor costs for assembly, reduced


821

overheads, localised "on demand" production feasible for bespoke service, and greater comfort contributing to longer garment use. Customized accessories The idea is to deliver bespoke functional bags ergonomically designed to fit the body. This project brings new materials and design process together with traditional craft making skills and materials for bags such as backpacks and computer cases. It tests a new paradigm of personalization of the bag form to individual body shapes, taken from a body scan, using rapid prototyping technology to translate forms into components. These are then encapsulated in materials and crafted into a range of bag shapes, with further options chosen by the user. Experimental materials used are metalized textiles which are initially conformable before an electro-deposition process is applied, combined with leather, industrial textiles, plastics and metals. The aspects of Considerate Design addressed include design personalized to end user requirements, efficient use of materials, longevity of products and localized production. Further applications would be suitable for ergonomic design for the elderly or in healthcare, with potential for the metalized textile products to have smart functionality. Evolving textiles Rapid Manufacturing for pseudo-textile structures which conform to the body. This proof-of-concept project tests out a new paradigm for a design process using generative software for 3D form finding, personalized to body scan data. Developments of genetic algorithms that optimize materials are already being used in the automotive, aerospace and construction industries. This project explores the use of such algorithms to create new textile structures, using a software package designed specifically for rapid prototyping output (www.complexmatters.com). This is investigated as a viable method for generating structures on a body surface, to give a flexible custom fitting seamless garment. It offers a new approach to design and a new discourse between the designer and the materials with which he works. Personalized design input includes unique shape bespoke fitted to an individual, coloration and surface patterning (Delamore 2005). Contribution to the Considerate Design process includes efficient use of materials, less waste, localised "on demand" production and fewer travel miles, a one stop process using less labor, thereby reducing overheads and carbon footprint.

822


Outlook: Supporting the Considerate Design Process The Considerate Design concept brings together notions from hitherto unconnected fields, such as sustainability, inclusive design and mass customization. Whilst the concept may be intuitive, designers need support in designing and assessing their products. Inclusive design (Keates and Clarkson 2003) has been concerned for many years with understanding the needs of elderly and disabled users and finding ways to design products that meet the needs of these groups, while providing good design for the wider society. While the emphasis of inclusive design has been on the design of better mass produced products for a wider audience, the issues in convincing companies to take up inclusive design indicate some of the challenges for considerate design. A survey of 101 UK companies, mostly in the design, manufacturing and retail sector, carried out by Goodman et al. (2006a) has shown the key barriers were the lack of knowledge, tools and a convincing business case, even though companies were motivated by social responsibility, demographic and consumer trends and saw inclusive design as a means to increase customer satisfaction and enhance their own brand. Considerable efforts are now being made to equip designers with tools to create inclusive designs (Goodman et al. 2006b). Inclusive design has mainly been operating in the product design area, which has a long tradition of user-centred design to generate products that meet user needs. Yet understanding those needs and making a business case is perceived as a major challenge. This problem is amplified in the fashion industry, which has traditionally been driven by fashion cycles, with little concern for the needs of particular user groups. Therefore designers will need support in understanding and assessing the needs of their users, while still providing them with the reassurance of offering fashionable products, but not immediately recognizable as being designed for the needs of a specific group. According to the particular context and constraints, considered "trade-offs" can be made which enable the best possible solution against the criteria thought to be the most important for each case, whilst alerting both the user and the designer to the alternative options. A visual representation of the considerate design footprint of the design will be developed as an additional communication tool (Figure 2). In the spider diagram the designers can access the relative strengths and weaknesses of different design options and think about trading them off against each other. Considerate Design has many dimensions, which define both the environmental impact, such as transport or disposal, and personal impact, such as comfort and cleanability of the product. These need to be traded off against conventional

823


measures such as design cost. Current research focuses on identifying those dimensions and developing a set of trade-off diagrams.

design 1 Transport

design 2

100 80 60

Maintenance/repair

40

Material

20 0

Design cost

Disposal

Laundry/cleaning

Figure 2: Considerate design profiles.

Economic viability of considerate design Perhaps the greatest challenge lies in finding an economically viable way to generate personalized designs in a considerate manner. Mass customization in the fashion industry is now being applied to the creation and fit of made to measure products, where basic design types are offered, but leaving the choice of style or materials to the user. This has been most successful for high end market products, such as specialized sports shoes (Delamore et al. 2005) or tailored suits, and products with a relatively limited range of styles and fabrics, such as jeans (Crawford 2005) or shirts (Byvoet 2005). One example, Brooks Brothers offer a "select shirt" service with a choice of 30 fabrics, 3 fit types, 6 collar and 4 cuff styles (www.brooksbrothers.com/selectshirts/) and a similar customized shirt service has been launched by Marks and Spencer (www.marks-and-spencer.com). In 2006 Nike introduced the Nike ID online system of customization for trainers, enabling colors and fabrics to be chosen and lettering to be added, and most significantly, for each foot to be specified differently. In response to customer needs, in autumn 2007 Nike instigated in their London flagship store a further personal designer service where face-to-face consultation and guidance was available (www.nikeid.co.uk), indicating adjustments to the service was required. The specific design of these products is created by the user in interaction with computer tools, choosing from variations on a basic design generated by a

824


designer. This requires a considerable degree of design automation. While this has clearly been successful in several cases the effort is considerable for each application, as to date the adaptation rules for particular products had to be mostly hand coded. Customization of garments is likely to require a considerable human design input for each design, which raises questions of the financial viability of mass customization as part of Considerate Design. It is therefore necessary to provide designers with support for understanding the processes they need to go through, and the costs and risks associated with each new design. The project draws on experiences from the engineering of complex products, where tools for planning design processes and assessing their costs and risks are far more advanced than in the fashion and textile industry, see Browning in Ramasesh (2007) for a review. The current project employs the Signposting approach (Clarkson and Hamilton 2000), to model processes through tasks and the input and output parameters. Unlike other modeling techniques, such as IDEF models (FIPS PUBS 1993) or Design structure Matrices (Steward 1981), the connecting parameters have qualifiers associated with them, which indicate the maturity of values, and the tasks have failure probabilities, so that the iteration in processes can be modelled probabilistically. The processes can be simulated over multiple runs so that a likely duration and cost can be established. This functionality is supported in the P3 environment (Wynn et al. 2006), which provides visualizations of design processes through flow charts and matrices and supports the analysis capability. By providing designers with a library of process building blocks, they can quickly describe the process for a particular design. After putting in cost figures for materials and means of production, designers can run simulations of their processes and estimate the time and cost of the product that they are developing. They can then critically evaluate both their design choice and design time in terms of cost and modify the design, if necessary, or increase their cost estimates. This has long been a significant problem for designers, who are notorious for underestimating the time and effort that is involved in design processes. Using simulation to assess approaches to customization Those in the best position to offer a quasi bespoke service are SMEs manufacturing in batches by semi-industrial methods, or individual craft-based businesses, whose production is flexible and can respond to individual orders. This model is represented by the bespoke bags case study in the present research. These businesses thrive on direct interaction and interpretation of customers' wishes and may operate on a cottage industry model, or by supervised small workshop


825

production. The ability to tailor orders to requirements distinguishes this batch production from volume manufacturing where uniformity of product over sufficient quantity is essential for economies of scale. However as mass fashion garments are in effect made individually, albeit in a production line, there is significant potential for customization at the right price and volume. One of the fundamental questions in customization is always, how much effort is put into customizing a design upfront, before any design is sold, and how much customization is carried out when an order is placed mirroring the spectrum of customization discussed earlier. This is of particular relevance to the businesses using batch production, where flexibility is possible for each customer, but built on a basic range which may be customized in size, style variations, colorways or a combination of these. Companies can pursue a number of strategies:

A bespoke design is generated at the point of purchase for each customer, here the options are to (a) completely redesign the product or (b) make cosmetic changes

An option package is developed upfront which is either (a) not changed further or (b) leads to the desired design with very minor modifications. Which strategy is most profitable obviously depends on the actual design effort involved in the number of the designs that the company expects to sell.

Figure 3: The main challenge of designing a patterned sweater is to ensure that motifs are not cut in an unsightly fashion (Eckert & Stacey 2003).

Eckert et al. (2008) reports on using process simulation to answer these questions using a simple knitwear design example. Figure 3 shows a garment with repeat

826


motifs. Although this particular design customization problem could be automated (for a discussion of the challenges see Eckert and Stacey 2003) it provides a useful example to illustrate the potential of process simulation to support the fashion designer. The designs look best when the motifs are not cut, however as a design is adapted to different measurements a lot of detailed tweaking of the design might be required to achieve this. Specifically we looked at the following options, as explained in more detail in Eckert et al. (2008).

A basic design is generated at setup cost C BS with an additional per-garment cost

o

A standard design knitted in a custom colour scheme. No design effort is required, but the machine must be configured and the garment knitted. The cost of this for one garment is C BC .

o

A standard design is customised for the customer’s individual body measurements and knitted in the desired colour. No additional design effort is required, but tasks associated with the detailed design must be revisited, along with the associated design iterations. The pergarment cost of adapting the measurements is C BM .

A standard design is generated up-front in 4 length measurements (short, medium, tall and extra tall) with four width settings each (slim, medium, large and extra large), such that each design is generated in 12 variants. The setup cost of generating these variants is CVS , where CVS > C BS . The customer has the following options: o

The design is not customised further. The best-fitting off-the-shelf variant in the desired colour is selected and sold to the customer. This incurs no per-garment customisation cost, i.e., CVC = 0.

o

The best-fitting variant is selected and customised to fit, then knitted in the chosen colour. This is similar to 1b above, but is likely to require fewer design iterations since it is possible to modify a variant which is closer to the final garment. It incurs a per-garment cost CVM where CVM < C BM .

The total cost for each of the two customisation strategies may thus be calculated, given the total number of garments N which will be produced and the fraction of


827

customers α who require customised measurements in addition to customised colour:

C B ( total ) = C BS +N [(αC BM + (1 − α )C BC ]

(1)

CV ( total ) = CVS +N [αCVM + (1 − α )CVC ]

(2)

Assuming that the customer pays a fixed price for the garment regardless of whether measurement adaptation is required, if the six values of C BS , C BC , C BM , CVS , CVC and CVM can be identified it is straightforward to identify the best customisation strategy for a given size of production run and fraction of customers requiring measurement customisation. In the following section, we illustrate how these cost components may be estimated and the best strategy identified through simulation of the applied signposting model (ASM) process detailed above. To analyze this a high level model of the knitwear design process has been generated as an ASM in the P3 environment, which describes the knitwear design process from the research in new design in a season to the production of the specific garments. The P3 has been based on a generic detailed model of the knitwear design (Eckert 2006; Eckert 1997). For the purposes of this paper, the top-most level of the original flowchart model was simplified to form the basis of an ASM simulation model. This required some modification to ensure the model was logically consistent and appropriately structured for simulation. The resulting model comprises 29 tasks and 9 decision points and is shown in Figure 4. Based upon the extensive case studies reported by Eckert (1997), in which over 25 knitwear designers and companies were interviewed or observed, cost values were estimated for the different customization options (see the table in Figure 5) and a monte-carlo simulation was run using the P3 functionality. Inserting the modal values from the table into Equations 1 and 2 above allows calculation of the modal customization cost for each of the two strategies for any values of N and α These results highlight that, while the bespoke strategy is always the best choice for low production volumes, a point is reached where it is more cost-effective to invest in up-front generation of variants; this is represented by the intersection of the two surfaces. The number of garments required to make the latter strategy viable depends upon the value of α , as well as the process-specific cost components. To illustrate, consider Figure 5, which shows cross-sections of the two surfaces for α = 0 and α = 1 . This plot indicates the minimum number of

828


garments required to justify the variant modification strategy (solid lines) assuming all garments are either measurement-adapted (light lines; α = 1) or colour-adapted only (heavy lines, α = 0). START

START

General Fashion Research in Companies

A

Fashion Research in Retail Chains

B

Briefing of Designers by Buyer Specific Design Research in Companies B

Research

Swatch Sampling D

no

Yarn SelectionC yes

Develop Design Framework no

Select Design Framework yes no Swatch Sampling D

Design

Detailed Design E no Like?

Swatch Sampling D

no

yes Discard yes

Meets Design Brief?

no

END

yes Technical Sketch

END

no no

Meets Design Brief? yes

Create Cutting Pattern F Consistent

yes

no no

Discard

yes

Create Fabric Sample G yes

Pattern PlacingH

Feasible Economical

no

no

Pattern Fits ? yes

Sampling

Internal Evaluation yes

no

Use Parts no

Buyer Presentation

Accept yes END

no

END

Alter no

END

Figure 4: High level flowchart of the knitwear design process (from Eckert, 1997) and the resulting ASM model (from Eckert, et al. 2008).

In the Knit to Fit Case study, this theory will be applied to the new paradigm of complete garment manufacturing which automatically produces knitwear garment by garment, irrespective of number required. In this scenario the input of labor is shifted strongly away from assembly to upfront design and pattern program specification, which may or may not be trivial. According to design requirements, a design may be modified from existing programs or designed individually at

829


much higher programming cost. Process modeling will enable these costs to be predicted more accurately. Anecdotal evidence from the Shima Seiki experience of customized service, obtained in November 2007 (Black S, personal communication, Shima Seiki store, Wakayama, Japan, Nov 2007), suggests that the variation model outlined above is most often applied to the standard knitwear offer, rather than complete garment knitwear.

Min Mode Max

C BS

C BC

12.8 28.8 76.8

2.5 2.5 2.5

Variant customisation

C BM CVS 3.7 7.0 45.6

69.6 103.2 156.0

400 350

CVC CVM 0.0 0.0 0.0

3.0 4.0 7.7

300 Total cost

Bespoke design for each customer

250 200 150 100 50 0 0

10

20

30

40

50

Figure 5: The cost values on the left hand side were run through a P3 simulation. Cross-sections through the resulting surfaces show the points when a different strategy becomes profitable.

Conclusions Design decision-making in the fashion clothing sector operates under a number of key constraints, notably high time pressures, remote manufacturing, saturated markets and increased competition. Dynamic supply chains create severe difficulties in achieving sustainable design, and responsibility is dissipated throughout the chain, with players at different points completely unconnected. Key decision makers are retail buyers, whose focus is on the right product at the right time and price, and designers whose focus is on the balance of style, aesthetics and cost. Communication between these interests determines economic success. The Considerate Design model will offer tools and guidelines for designers related to the entire lifecycle of a fashion product, with particular focus on the design stage, enabling informed decisions to be taken within the complex fashion product supply chain. It aims to empower the designer to balance the often conflicting priorities of user needs, design cost, environmental impact of materials and components, transportation, production methods, maintenance, laundry and aftercare, longevity and disposal, and to communicate these within the company to buying and sales teams. According to the particular context and constraints, considered "trade-offs" can be made which enable the best possible solution

830


against the criteria thought to be the most important for each case, whilst alerting both the user and the designer to the alternative options. The results will be two-fold: general guidelines for the best approach to carry out Considerate Design processes and the specific processes for individual designers modeled in the Signposting tool, which will enable them to reduce iteration, assess process risk and cost their design input before launching a project. They will also demonstrate the concept of Considerate Design as an underpinning to design work and product development, a form of environmental and ethical audit, contributing to the considerate "triple bottom line" of economic, social and environmental impact (Elkington 1997). One of the additional benefits of Considerate Design is the ability to be applied in mass production as well as for mass customization, highlighting the design process and associated costs for the benefit of more accurate information for the commercial sector. Due to serious overseas competition and decline in traditional markets, much of the UK textile industry has now moved into specialist niche markets such as technical textiles for sportswear or medical uses, products with identifiable added value. The effort has shifted from volume manufacturing to small editions or bespoke products with very high design input. This project supports this fledgling industry and addresses the need for similar value and differentiation in fashion products, through new craft and design methodologies. It could assist in retaining production and design capability in the UK and Europe by making Considerate Design economically viable. Through Considerate Design it is hoped to meet individual requirements guided by environmental and ethical principles which are applicable on both a local and global scale and at a personal and company level. The project also poses and attempts to answer the question – can personalized fashion and Considerate Design help to solve the fashion paradox?

References Black, S. (2002). Innovative Knitwear using seamless and unconventional construction, Proceedings of International Federation of Fashion Technology Institutes (IFFTI) 5th Annual Conference. Hong Kong, November 02. Black, S. (2005). Benchmarking of seamless knitting technologies. Proceedings of the 84th Textile Institute World Conference CDROM North Carolina USA, June. 2005. Black, S. (2006). Interrogating Fashion. In: Design Dialogues. Proceedings of Designing for the 21st Century. Tom Inns (ed). Dundee: University of Dundee, p. 4. Black, S. (2007). Interrogating Fashion: Practice, Process and Presentation. New Paradigms for Fashion Design in the 21st Century. In: Designing for the 21st Century: Interdisciplinary Questions and Insights. Inns, T (ed). London: Gower, 299–314.


831

Black, S. (2008). Eco Chic: the Fashion Paradox. London: Black Dog Publishing. Bougourd, J.P. (2006). Sizing systems, fit models and target markets. In: Sizing in Clothing. Susan Ashdown (ed). Cambridge:Woodhead Publishing, 108–146. Bougourd, J., Dekker, L., Grant-Ross, P. and Ward, J.P. (2000). A comparison of women’s sizing by 3D electronic scanning and traditional anthropometry. Journal of the Textile Institute. 91(2) part 2: 163–173. Bougourd, J., Treleaven, P. and Allen, R.M. (2004). The UK National Sizing Survey Using 3D Body Scanning. Eurasia-Tex Conference, Donghua University, Shanghai, China. Browning, T.R. and Ramasesh, R.V. (2007). A Survey of Activity Network-based Process Models for Managing Product Development Projects Production and Operations Management. Byvoet, M. (2005). Collaborative platform for e-custom fit. Case study shirtdotnet.com. Proceedings of 3rd MCP World Congress Hong Kong, Sept 05. Clarkson, P.J. and Hamilton, J.R. (2000). "Signposting": a parameter-driven task-based model of the design process. Research in Engineering Design. 12(1): 18–38. Crawford, A. (2005). Leveraging SizeUK 3D body data for mass customization and personalization. Bodymetrics case study presentation. Proceedings of 3rd MCP World Congress. Hong Kong, Sept 05. Delamore, P. (2004). 3D Printed textiles and clothing on demand. Intermesh RMIT Symposium presentation. Melbourne, Australia 2004. Delamore, P., Junior, V., and Lever, G. (2005). 3D Direct Manufacturing of made-to-measure performance footwear. Proceedings of Wearable Futures Conference CD ROM. University of Wales, Newport, Sept 05. Eckert, C.M. (1997). Intelligent Support for Knitwear Design. PhD thesis, The Open University, UK. Eckert, C.M. and Stacey, M.K. (2001). Designing in the Context of Fashion – Designing the Fashion Context. Designing in Context: Proceedings of the 5th Design Thinking Research Symposium. Delft University Press, Delft, Netherlands, 113–129. Eckert, C.M., Blackwell A., Bucciarelli L., Clarkson P.J., Earl C., Knight T., Macmillan S. Stacey M. and Whitney D. (2005). Comparative Study of Design – Application to Engineering Design Proceedings of the 15th International Conference on Engineering Design The Design Society, Melbourne, Australia, August 2005. Eckert, C.M. (2006). Generic and specific process models: Lessons from modeling the knitwear design process. Proceedings of TMCE 2006, April 18–22, 2006, Ljubljana, Slovenia. Eckert, C.M., Wynn, D., Clarkson, P.J. and Black, S. (2008). Process simulation to make personalization economically viable. Proceedings of TMCE 2008, Izmir, Turkey. Elkington, J. (1997). Cannibals with Forks: the Triple Bottom Line of 21st Century Busines. Oxford: Capstone. FIPS Publications 183. (1993). Federal Information Processing Standards National Institute of Standards and Technology USA, available at www.itl.nist.gov/fpspubs/idef02. Goodman, J., Dong, H., Langdon, P.M. and Clarkson, P.J. (2006a). Increasing the uptake of inclusive design in industry. Gerontechnology, Special Issue on Universal Design, 4 (4). Goodman, J., Langdon, P.M. and Clarkson, P.J. (2006b). Equipping designers for inclusive design. Gerontechnology, Special Issue on Universal Design. 4(4): 229–233. Heyd, J.L. (2004). Dream or reality? Potential savings in the clothing industry by means of the virtual model. IMB Forum Congress, Cologne, October 2004.

832


Keates, S. and Clarkson, P.J. (2003). Countering design exclusion: an introduction to inclusive design. London: Springer. Sayer, K.N., Challis, S. and Wilson, J.A. (2007). Seamless knitwear: The design skills gap, The Design Journal. 10(1). Shima, M. (2005). Developments of On-Demand Production Systems: Automated Production. International Federation of Fashion Technology Institutes 7th Annual Conference presentation, Tokyo, Nov 2005. Treleaven, P. (2004). UK National Sizing Survey using 3D body scanning. Proceeding of National Physical Laboratory DMAC annual conference, Birmingham UK, April 2004. Watkins, P. (2005). Custom fit pressure garment pattern profiling. Proceedings of Wearable Futures Conference CDROM.University of Wales, Newport, Sept 2005. Walker, S. (2004). Sustainable by Design, Earthscan, London Wynn, D., Eckert, C.M. and Clarkson, P.J. (2006). Applied signposting: a modeling framework to support design process improvement. ASME International Design Engineering Technical Conferences and Computers and Information in Engineering Conference, Philadelphia, USA, DETC2006-99402.

Author Biographies Sandy Black is a Professor in Fashion Design and Technology at the London College of Fashion, University of the Arts London, United Kingdom. She is Director of the LCF Centre for Fashion Science. Her research interests are knitwear, fashion and textiles design with particular emphasis on 3 dimensional aspects of design and realisation, incorporating mathematical principles; Inter-disciplinary design in social and cultural context; Innovation in knitwear particularly 3D and seamless construction; Intersection of arts and science with design; Sustainability issues in fashion and textiles. She previously designed and manufactured her own label, Sandy Black Knitwear, selling internationally Contact: www.fashion.arts.ac.uk/research | [email protected] Claudia M. Eckert has returned to the Open University as a senior lecturer in September 2008, after nearly 10 years as researcher in the Engineering Design Centre in Cambridge. She carried out her doctoral research at the Open University on intelligent support for knitwear designers to elevate problems in communicating and customising their designs. In Cambridge she researched engineering change and process planning and contributed to the development of tools for change prediction and process modelling, whose applicability to other application areas she is now investigating. One of her interests is comparison and transfer of best practice across different design domains. For example she is part of the considerate design for personalised fashion projects, where she applies process modelling and change prediction technique from engineering to assess the economic and environmental viability of textile design and customization processes. Contact: www-edc.eng.cam.ac.uk | [email protected]

2.5

Customized Garment Creation with ComputerAided Design Technology

Jing-Jing Fang Department of Mechanical Engineering, National Cheng Kung University, Tainan, Taiwan Chia-Hsin Tien Department of Mechanical Engineering, National Cheng Kung University, Tainan, Taiwan

In this paper, an application of computer-aided techniques to customize garment creation is proposed. Different from traditional CAD tools on planar pattern design, the authors, according to an individual body scan, present a parameterized procedure with minimum least-squared approximation to construct a basic bodice generated by trimmed NURBS surfaces. Then, some tools based on trimmed NURBS techniques are available to reshape and slice the bodices. Finally, the authors propose an innovative tailoring method to generate a flare sagging style.

Introduction In traditional garment manufacturing, designers' styling concepts are not directly translated to the finished garment. The success of the final garment depends on how well the pattern technologist transfers the designs to planar patterns; the process is empirical and instinctive. The garment made from these patterns may not entirely meet designers' requirements the first time; typically, either a revision on the pattern or a design change occurs. The whole process is often time consuming and limits the designer’s freedom of creative expression. Once the patterns are made, they still require grading and some dimensional alternations before being used to manufacture garments. For ready-made garments, a set of grades dimensioned for a population are chosen and then the patterns are scaled and modified using rules to create a set of patterns with these grades. However, a grade only represents a general dimension with a set range according to some fundamental measurements such as the bust, waist, hip, etc. In standard grading systems, the unique and complex dimension and surface configurations of individuals are not taken into account. Conversely, a customized garment has the advantage of the ability to provide closeness to each unique body, versus a ready833

834


made garment, but it also takes much more time and work due to necessary personal measurements and a final fitting assessment, generally requiring further manual adjustments. In order to alleviate the limitations mentioned above and improve efficiency, computer aided design (CAD) has been applied to the manufacturing processes of the garment industry. In early applications, CAD systems have automated the manual 2D pattern-making process. These systems have been widely used for a several decades and have continued to be developed in order to reduce the workload associated with conventional 2D pattern making. However, these systems merely free pattern technologists from conventional tools, such as pens and rules; the main issues of the entire manufacturing process still remain. Currently, as the technology of CAD and computer graphics advance, the new applications that will provide more functionality are being developed. These new systems simulate 2D meshed patterns draped on a 3D mannequin so that style designers and pattern technologists are able to directly modify each piece of the basic patterns in 3D to achieve the design specifications (Wang and Yuen 2005a; Cordier et al. 2005; Cugini et al. 2005; Luo and Yuen 2005). Moreover, in some applications, assessment of closeness to the body and virtual fit simulations (Choi and Ko 2005; Fozzard and Hardaker 1998; Wang et al. 2003; Wang et al. 2005b) are also integrated into these environments to reduce the complexity and workload of the entire process of developing patterns. Though the patterns are made in 3D, they still need to be mapped back into 2D for actual manufacturing. In general, this is accomplished using an approximating process consisting of flattening the meshes into a pattern with some reasonable constraints (Chong et al. 2005). Application Examples of the Trimmed NURBS Surface This section of the paper describes examples of applications of this study in CAD fashion design, including the use of the NURBS method, a description of the algorithm and calculation procedures, a demonstration of the interface design, etc. The following section introduces the foundation of CAD fashion design torso data, including the different types and important features of data points and explains the method of generating the basic bodice from the torso’s data points. This study uses two parameters to decide bodice styles and pre-sets several common styles for ease of operation. A method by which the user can specify girth features is also described. The next section discusses application examples of the NURBS method in shaping the bodice curves, including the shaping of basic single points along specific directions and the shaping of the boundary conditions


835

for the connection – the seams – between the front and back bodices. Next, we describe application examples of the trimmed NURBS surface method in bodice curve cutting, including the generation of the trimmed surface, (left/right) symmetric cutting, etc. The final section discusses the application of tiered draping and several demonstration results. Definition of torso structure Geometrically, clothes are surfaces placed onto bodies. The aim of customization is to design different clothing for different people. Hence, unique torso structures (Fang and Tsai 2003; Fang et al. 2007) can become the basis of unique clothing surfaces. A half-torso structure of 41 longitudinal lines and 61 latitudinal lines, as shown in Figure 1, can be generated from cloud data of real torso surfaces, captured by a torso scanner, by means of a search for geometrical features. This half-torso structure can then be used as the basis for upper torso garments and garments such as dresses that fall freely from the hip or crotch level. Longitudinal lines No. 0, 10, 20, and 40 are the center front line, left front princess line, left side seam line, and center back line, respectively; latitudinal lines No. 0, 10, 15, 20, 30, 38, and 48 are the girths of crotch level, hip, abdomen, waist, under-breast, bust, and under-armhole, respectively. Table 1 lists these important longitudinal and latitudinal feature lines that serve as the basis for pattern design for many styles of clothing. In actual application, this study divides the basic upper bodice into two pieces, front and back. To ensure a symmetrical model, it first compares the plane formed by structural points of the left-half bodice with the front and back central lines for a mirror reflection and the generation of structural points of the right-half piece. Structural points of the entire torso are then adopted for parameterization of the feature girths of the bodice and finally to combine the front and back bodice surfaces. Establishment of the basic bodice surface Before establishing the bodice surface, the missing data points in neck line and armhole girth are filled in for convenience when building NURBS surfaces. A generalized sloper is built based on parameter settings (a sloper is a basic pattern for a relatively fitted garment shape with no distinctive style features). The front and back bodice surfaces are built based on the torso structure data, and then the meshes enclosed by the collar girth and armhole girth are removed by trim curve.

836


Figure 1: Torso structure data. Table 1: Feature lines. Longitudinal feature No.

Feature

Latitudinal feature No.

Feature

0

Center front line

38

Bust girth

40

Center back line

30

Under-bust girth

10

Left front princess line

20

Waist girth

30

Right front princess line

15

Abdomen girth

20

Left side seam line

10

Hip girth

0

crotch level girth

837


The original torso structure data does not have data points in the neck line and armhole girth as shown in Figure 2. To make the surfaces smoother, we fill in data points within these areas. After filled, the torso data has a fixed number of points in the latitudinal and longitudinal directions; the result is shown in Figure 3. After the surfaces are generated in Figure 4(a), the curves of the collar girth and armhole girth are taken as trim curves in order to cut away surfaces within two enclosed areas, as shown in Figure 4(b).

(a) Original half torso

(b) Half neck line

(c) Armhole girth

Figure 2: Original torso.

This study utilizes two parameters to control the closeness to the torso of basic bodices: the parameter t r , which controls the closeness in the latitudinal direction, and the parameter tc , which controls closeness in the longitudinal direction. Let us first examine tr . Take the structural points of each girth, as shown in Figure 5 into consideration. The sloper that regards the original torso’s structural points as the closest fitting sloper (sinking into all of the hollows of the torso) and minimum envelope convex polygon as the loosest fitting sloper (spanning the torso hollows); the other slopers are all between them. This study makes use of a linear combination of the two aforementioned to produce various slopers, where the parameter t r is the parameter of the linear combination.

838


Figure 3: Torso filled with points. Collar Girth

Armhole Girth

Lap

Figure 4: (a) NURBS surface. (b) Cut away the areas enclosed by collar girth, armhole girth and lap.


839

Body points Envelope

Figure 5: Sketch of closeness to the torso in the latitudinal direction.

Figure 5 indicates that, among the structural points of any certain torso girth, the hollow points are structural points where the minimum envelope convex polygon will not sink in at the front and back of the envelope. We define the original group of torso structure points as B, the minimum envelope convex polygon as C, and the sloper generated as T; thus the sloper formula is:

Ti = (1 − t r ) Bi + t r Ci , 0 ≤ t r ≤ 1 , Ti ∈ T , Bi ∈ B, C i ∈ C Tk ,i = (1 − tr ) Bk ,i + tr Ck ,i , 0 ≤ tr ≤ 1

(1) (2)

In the direction of the cross section, a more closely fitted sloper will be generated the smaller tr is. Conversely, a looser sloper, will be generated the bigger tr is, while, Tk ,i , Bk ,i , Ck ,i represent the i-th points in girth k of sloper, torso and convex polygon respectively. When the minimum envelope convex polygon has been calculated, the sunken parts among the points in group B do not have corresponding points on the minimum envelope convex polygon. Thus, corresponding points need to be specially built. This is done by drawing a ray from the sunken point along the centroid and taking the point of intersection with the minimum envelope convex polygon as the corresponding point of the sunken point, assuring that all points in the B group of points have corresponding points on the minimum envelope convex polygon to form point group C. The sloper T can then be calculated correctly. Figure 6 demonstrates the influence of various tr on generated bodices and indicates that the curves of the torso’s cross section are less visible as tr grows larger.

840


tr = 0

t r = 0.8

t r = 0.4

tr = 1

Figure 6: Bodice with various parameters tr .

Figure 7: Description of degree of goodness of fit curves.

Another parameter t c is used to emulate the fit of statically-hung clothes, and can be interpreted as the extent of longitudinal hugging of torso curves. Take the princess line of the mannequin shown in Figure 7 as an example. Proceeding downward from the top of the mannequin to the bust point Pi , the distance between points and the torso’s central axis gradually increases, and the distance between points below the bust point and the centroid gradually decreases. A piece of fabric draped on the mannequin will hang naturally along the dotted line in the figure. t c is the very parameter used to control the linear combination of naturalhanging and form-fitting curves, which, unfortunately, cannot be expressed by a single formula. When examining Figure 7, we find that points above Pi cannot hang naturally along the dotted line like points below Pi due to gravity. Hence, the farthest distance away from the centroid must be recorded from a top-to-bottom process. Thus, each girth from neck line to bottom is projected onto the xy plane which is then enclosed within a minimal polygon M. The procedure to generate sloper Ti are:

841


Step 1. Project each girth onto xy-plane Step 2. Let polygon M be the topest girth Bn Step 3. Let k = n – 1 Step 4. If each point i of Bk inside the polygon M, then Tk,i = (1 – tc)Bk,i + tcMi Step 5. If each point i of Bk outside the polygon M, then Tk,i = Bk,i and let Mi = Bk,i Step 6. Let k = k – 1 and repeat steps 4, 5, 6 Base on the procedure, M will be enlarged gradually from top to bottom. The result of each calculation is the distance between each sloper point and the girth’s centroid; it is used to calculate the position of other sloper points. Figure 8 demonstrates the influence of different t c on the bodice generated. The smaller tc is, the more closely the clothes will hug the torso’s curves. Conversely, the bigger t c is, the looser the clothes will be.

tc = 0

tc = 0.8

tc = 0.4

tc = 1

Figure 8: Bodices with various parameters tc .

842


Figure 9 is the program interface for sloper settings. Figure 9(a) displays the sloper meshes with torso and current settings. Figure 9(b) provides settings of various parameters and the upper part of interface represents the three pre-set slopers provided, the blouse, the sheath dress, and the shift dress, respectively. These options will use pre-set t r , t c , and bodice length. t r and t c for the sheath dress are 0.89 and 0.3, respectively, and those for the shift dress are respectively 0.89 and 0.89. The two dresses' lengths extend to the crotch level girth, while the parameter settings for the blouse are the same as those of the sheath dress, but only to the waist girth. Two options at the bottom of Figure 9(b) can adjust the two parameters t r and t c and are termed as the latitudinal and longitudinal coefficients, respectively.

(a)

(b)

Figure 9: Interface for sloper settings.

843


The two parameters' meanings are evident from the terms themselves. For ease of operation, the parameters' value range is further divided into percentages varying from 0% to 100%. Figure 9(b) displays the choices for the torso girths, which are the collar girth, bust girth, waist girth, and hip girth, respectively. In the Blouse Parameter Setting window, one can change the length of the sloper girths and the bodice. The user can also refer to the torso information directly above the Blouse Parameter Setting window to adjust these values accordingly and directly specify the length of torso girths and the bodice. Figure 10 shows the result after pre-set slopers are fashioned into complete articles of clothing. As shown in the figure, the shoulders of the torso structure may penetrate the bodice slightly, which results from a smooth form-fitting bodice surface over the slightly uneven torso structure surface. After the sloper setting is completed, the sloper’s data points are taken as the basis for the combination of the NURBS surfaces. This study uses global least square approximation and refers to the distance between data points to specify parameter values (Chord Length Method) so that smooth surfaces can be generated.

Blouse

Shift Dress

Sheath Dress

Figure 10: Three pre-set dresses.

Clothes shaping This study provides two types of interfaces for clothes shaping. One utilizes single-point shaping directly on three-dimensional clothes and the other shapes two-dimensional curved lines on the specified horizontal cross section. For convenience of shaping, the shaping direction is confined to the x-y direction, or xy plane. As shown in Figure 11, the torso’s longitudinal central axis is the z axis,

844


the anterior and posterior is the x axis, and the lateral is the y axis. The user can choose the shaping direction. Moreover, the front and back pieces of the bodice are kept consistent during shaping, and the surface’s differential remains unchanged. Otherwise, two pieces of bodice may separate, overlap, or be rough at the joining seam. To achieve the two aforementioned targets, a constraint-based shaping for actual application of the NURBS method (Piegl and Tiller 1995) is adopted.

Figure 11: Definition of coordinates (xyz). n

A NURBS curve is defined as C (u ) = ∑i =0 Ri , p (u ) P i , where p is the degree, Pi is the control point, and the node vector is U = {u0 ,..., un + p +1 } , given the group of parameters {u s }, s = 1,..., S , and variation {∆Ds(k ) } of the curve at each order of these parameter positions. k theoretically represents a k-order differential, up to p + 1 order differential. However, it is meaningful only when k =0, 1, 2, which represent the variations of position, inclination, and curvature, respectively, when u = u s . As mentioned in the NURBS book (Piegl and Tiller 1995), if compared with the use of Ds(k ) to represent K-order differential when u = u s , s = 1,..., S , then

845

VOLUME 2: APPLICATIONS AND CASES n

Ds( k ) = C ( k ) (u s ) = ∑i = 0 Ri(,kp) (u s ) P i

(3)

Given that only control points could be changed, the above formula plus the variation {∆Ds(k ) } will be

∑

n i=0

R i(,kp) ( u s )( P i + ∆ P i ) = D s( k ) + ∆ D s( k )

(4)

It is concluded from the above two formulas that:

∑

n i =0

Ri(,kp) (u s )∆P i = ∆D s( k )

(5)

Hence, when the variation of differentials of different orders is limited, the variation of corresponding control points can be obtained based on Eq. (5). Eq. (5) is formed as a N+1 simultaneous equations which can be written into a matrix form and solved by optimization method such as the least square method. Surfaces can also be adopted with a similar method. In the single-point shaping used in this study, the point chosen for shaping is S (ui , vi ) and shaping is confined along the x direction. ∆Di( 0,0 ) = ( ∆x,0,0) thus needs to be specified. Similarly, ∆Di( 0,0 ) = (0, ∆y ,0) and ∆Di( 0,0 ) = ( ∆x, ∆y ,0) confine shaping along the y direction or xy plane, respectively. Proper and smooth seams of the front and back pieces thus cannot be perfect because the area of joining between them can be regarded as a NURBS curve. This method is only suitable for boundary conditions at discrete positions; it cannot be ensured that all positions of the curve can meet this requirement. This study takes enough points at the same intervals at the seam locations. Since the longitudinal direction is taken as the y direction of the surface’s parameters and since when surfaces are combined, one differential variation along the longitudinal line is 0, i.e. ∆Di( 0,1) = (0,0,0) , smoothness at these positions can be ensured. If enough points are taken, a good effect can be realized but this requires more calculation. Another clothes-shaping method used in this study is the adoption of twodimensional cross sectional girth. The operating interface is shown in Figure 12; the horizontal cross section determines the girth for current shaping. Arrows ↑ and ↓ on the interface control the height of the cross section. The pre-set girth can move the cross section directly to the height of the specified girths, including bust, under-breast, waist, middle waist, and hip, which makes it more convenient for users to switch between these girths that need shaping. The girth at current cross section is shown in the box at the top right of the interface shown in Figure 12. Here, the user can directly conduct bodice shaping, where the styling function is realized.

846


Figure 12: Girth shaping.

Figure 13 and 14 respectively demonstrate two shaping methods: single-point and entire enclosure scaling. While the user shapes the girth, the three-dimensional bodice’s surface changes accordingly because the two-dimensional girth shaping also uses an identical method for three-dimensional shaping of bodice’s surface, equivalent to three-dimensional shaping confined to the xy plane, but with different operating interfaces.

Styling

Figure 13: Single-point shaping.


847

Figure 14: Entire enclosure scaling shaping.

Application of trimmed NURBS surface in style scissoring This study uses a trimmed NURBS surface function to duplicate the function of scissoring or cutting the edges of the clothing. The studies related to trimmed NURBS surface refer to actual applications. The original surface does not change because of the trimming of NURBS surfaces; only parts cut away do not appear. Also because of this, the user can still change the shape the cutting line after scissoring. However, scissoring lines are NURB curves that exist on a parameter plane and are difficult to shape directly in pattern space. This study adopts the point inversion method (Hewitt et al. 2003) to solve this problem. As shown in Figure 15, the point inversion method can transform points that the user chooses during bodice scissoring into points on the parameter plane. Assume the bodice’s curve is S (u, v ) = ( x, y , z ) . Then the scissoring line on the parameter plane is C (t ) = (u, v ) , and the composite function S (C (t )) is the cutting line in model space, hence, S −1 ( x, y , z ) = (u, v ) can represent point inversion. The scissoring curve in the model space shapes one point P0 = ( x0 , y0 , z0 ), P0 ∈ S (C (t )) to the other point P1 = ( x1 , y1 , z1 ) . Point inversion Q0 = S −1 ( P0 ) and Q1 = S −1 ( P1 ) is used to obtain corresponding points on the parameter plane. When Q0 ∈ C (t ) , the shaping position of the scissoring line C (t ) is obtained from t 0 = C −1 (Q0 ) . Hence, the final movement

848


shaping is C ' (t 0 ) − C (t 0 ) = Q1 − Q0 . Therefore, the single-point shaping on a NURBS curve can also be used to shape the scissoring line on parameter plane.

Figure 15: Trim curve shaping.

Flare draping of tiered skirt This process utilizes the concept of displacement of surfaces. Adding displacement to the bodice surface creates tiered shapes. Assume the original surface is S (u, v ) ; and the displacement function is W (u, v ) . Make N (u, v ) the normal vector of S (u, v ) at (u, v ) , then the surface produced is (6) S ′(u , v) = S (u , v ) + W (u , v ) ⋅ N (u , v ) Special attention should be given to the fact that W (u, v ) is a scalar function, which ensures the shape’s variation along the bodice’s normal vector. Thus a designer only needs to choose a desired function of displacement instead of directly shaping a flare draping surface which is much more difficulty. This study uses a simple sinusoidal function to ensure periodical displacement variation of W (u, v ) and to develop wave shapes. The formula is

W (u , v) = (1 −

v ) ⋅ d ⋅ sin( f π u ) R

(7)

849


Parameter R represents the scope of wave, d represents the maximum displacement of waves away from the original surface, and f represents the wave frequency. Simple wave shapes are made only at the lower draping area. This mathematical statistics method can be used to develop a variety of more sophisticated shapes. Figure 16 demonstrates the result of the paired use of scissoring and tiered draping, while Figure 17 shows several styles designed by application of three-dimensional scissoring.

(a)

(b)

(c)

(d)

Figure 16: Draping and three-dimensional scissoring: (a) basic bodice; (b) scissoring; (c) wave-like draping; (d) wave-like draping with scissoring.

(a)

(b)

(c)

(d)

(e)

Figure 17: Possible design styles (a) bateau collar; (b) square collar; (c) vect collar; (d) deep V-neck vest; (e) thin shoulder ribbon.

850


Discussion and Conclusions This study applies trimmed NURBS surface to three-dimensional CAD fashion design and develops NURBS surface’s shaping and scissoring teaching methods. It contributes to the process by developing its numerical methods and practical applications:

The study establishes two parameters for slopers and develops a method of sloper parameterization based on a torso structure model and specified feature girths, to generate basic slopers with different styles.

It accomplishes another shaping method for single-point NURBS surface shaping to get a smoother shaping effect. It also takes into consideration boundary conditions to be the shaping direction and smooth connection of front and back pieces, and applies this method into the shaping of the bodice’s surface.

This study adopts trimmed a NURBS surface to demonstrate the scissoring effect of the bodice surface in fashion design and uses the point inversion and projection method to shape the parameter plane. The shaping of the scissoring line after initial scissoring is possible.

This study makes use of the periodical function for displacement of the bodice surface and creates simple wave-like draping. This method also provides a basis for various styles in the future. The incomplete solution for many problems related to the geometric aspect of NURBS’s applications is a continuing area of investigation for academic research all over the world. Moreover, automatic workflow of fashion design and manufacturing is still far from full maturity. Transformation of the 3D bodices into 2D patterns that accurately reproduce the shapes illustrated here, identifying the appropriate locations for additional seams and other shaping devices, and incorporating the physical action of fabric grain interacting with the torso surface and with gravity is an area that will require more study. However, this preliminary work is an important step in the process of providing a useful, innovative, and intuitive tool for fashion designers.

Acknowledgments The study is funded by Ministry of Economy Affairs, ROC, under the project number 95-EC-17-A-19-S1-053.


851

References Wang, C.C.L. and Yuen, M.M.F. (2005a). CAD methods in garment design. Computer-Aided Design. 37(6): 583–584. Cordier, F., Magnenat-Thalmann, N. and Volino, P. (2005). From early virtual garment simulation to interactive fashion design. Computer-Aided Design. 37(6): 593–608. Cugini, U., Fontana, M. and Rizzi, C. (2005). 3D virtual apparel design for industrial applications. Computer-Aided Design. 37(6): 609–622. Luo, Z.G. and Yuen, M.M.F. (2005). Reactive 2D/3D garment pattern design modification. ComputerAided Design. 37: 623–630. Choi, K.J. and Ko, H.S. (2005). Research problems in clothing simulation. Computer-Aided Design. 37(6): 585–592. Fozzard, G.J.W. and Hardaker, C.H.M. (1998). Towards the Virtual Garment: Three-Dimension Computer Environments for Garment Design. Int.Journal of Clothing Science and Technology. 10(2): 114–127. Wang, C.C.L., Wang, Y. and Yuen, M.M.F. (2003). Feature based 3D garment design through 2D sketches. Computer-Aided Design. 35(7): 659–672. Wang, C.C.L., Wang, Y. and Yuen, M.M.F. (2005b). Design automation for customized apparel products. Computer Aided Design. 37(7): 675–691. Chong, K.W., Hinds, B.K. and McCartney, J. (2005). Pattern flattening for orthotropic materials. Computer-Aided Design. 37(6): 609–622. Fang, J.J. and Tsai, M.J. (2003). Feature based data structure for computer manikin. USA patent 7,218,752, 2003/11/4~2023/11/3. Fang, J.J., Leong, I.F. and Tsai, M.J. (2007). Automatic torso features extraction from marker-less scanned human torso. Computer-Aided Design. 39(7): 568–582. Piegl, L. and Tiller, W. (1995), The NURBS Book, Springer, Berlin. Hewitt, W.T. and Ma, Y.L. (2003). Point inversion and projection for NURBS curve and surface: control polygon approach. Computer Aided Geometric Design. 20(2): 79–99.

Author Biographies Jing-Jing Fang is an Associate Professor in the Department of Mechanical Engineering in National Cheng Kung University, Taiwan. She leads her research team working on the area of digital anthropometry, 3D garment and pattern design, image-guided surgical planning and navigation. Her research interests are geometric modeling, object-oriented design, and virtual reality applications. She received her BS and M.Sc. in Applied Mathematics from Taiwan, in 1984, and Ph.D. in Mechanical and Chemical Engineering in Heriot-Watt University, Britain, 1996. Contact: vr.me.ncku.edu.tw | [email protected] Chia-Hsin Tien is currently a PhD candidate in the Department of Mechanical Engineering in National Cheng Kung University, Taiwan. He obtained his BS and M.Sc. degrees in 2003 and 2005, respectively, in the same department. His research interests are in the areas of computer graphics and computer aided geometric design. Contact: [email protected]

2.6

A Case Study in Personalized Digitally Printed Clothing Philip Delamore London College of Fashion, University of the Arts London, United Kingdom Jennifer Bougourd London College of Fashion, University of the Arts London, United Kingdom

The project presented in this paper focuses on the digitisation of the clothing product development process incorporating 3D body scanning, automatic pattern generation, visualization, digital printing and embroidery. The aim was to introduce custom print and embroidery to the existing Bodymetrics Digital Couture offer in two London retail stores. The first objective of this project was to extend the couture jeans offer in Selfridge’s, and the second objective to introduce customizable suit linings for Nutters in Harrods. The current virtual data flow with the inherent processes is described and a detailed account of the research is given in two stages and illustrated with examples of the use of digital print and embroidery in each case. A description is given of the steps constituting the first stage, which involved the consumer in print and embroidery selection and the automatic extraction of his or her measurements from which a prepared pattern was produced. An engineered print was visualized on a 3D virtual image of the customer, before being cut and manufactured. For stage two of the work an account is given of the methods used for re-proportioning imagery to automatically fit different pattern sizes. The problems of CAD workflow, fabric stability during the printing process, and print registration during single ply cutting for each case are outlined, together with a summary of progress and outstanding issues.

Introduction During the past 100 years the apparel and footwear industries have almost traveled full circle from traditional bespoke clothing, through mass production to mass customization. Oliver et al. (1993) described traditional bespoke clothing as being shaped by three factors – convenience, service and selection – which "allow a person to create a unique style without having to go shopping". This clothing was made to meet the needs of individuals but it is a labor-intensive, craft-based process, available only to an élite group. Mass production emerged from the industrial revolution, opening access to ready-to-wear clothing to the wider population and, although the traditional process of mass production is continually 852


853

being enhanced through the use of new technologies, it is nevertheless being challenged for a variety of reasons.

Increased competition in the market place

Changing consumer behavior.

Designs being replicated by competitors.

A high percentage of returns.

Over supply of very similar goods Consumers are now better educated, more demanding and discerning. Fralix (2001) suggests that they seek greater variety and immediate, personalized service. In the same way that mass production replaced bespoke clothing at the turn of the twentieth century, mass customization is expected to replace mass production in the twenty-first century. It is regarded as a new paradigm, based on creating variety and customization through flexibility and responsiveness and is perhaps, as Fralix (2001) proposes, a practice that combines the best of the craft era with that of the mass production era. Tseng and Piller (2003) see the apparel and footwear industries as being forerunners of the application of mass customization. They have the potential to address all three possible dimensions of customization; fit, functionality and aesthetic design and, with the advent of new and enabling technologies, the opportunity to bring mass customization to fruition. One of the first retail companies to offer custom clothing using these new technologies was Brooks Bros. of New York. Many other retailers in the US, mainland Europe and, to a lesser extent the UK, now offer these services, although product types are confined to classic styles such as jeans, shirts and tailored suits. A UK company called Bodymetrics has two custom clothing units in London; one set up in Selfridges’s department store, offering digital personalized jeans, the other in Harrods department store, specializing in digital bespoke suits for women. The authors worked with this company to introduce digital print and embroidery to both their jean range and their jacket linings. Aim and Context of this Research The overarching aim of this continuing research is to create a totally digital system for the product development and manufacture of made-to-measure and personalized clothing. In stage one digital print and embroidery was introduced to an existing Bodymetrics jeans range and animated design approval. In stage two (reported in this paper) we have extended that offer to include digitally printed linings for women’s suits with a method to re-proportion images for differently-

854


sized garments and sought to resolve some of the problems reported in stage one: the use of pigment dyes to obviate fabric instability (inherent in the digital printing process) using reactive dyes, and further investigate vision recognition systems enabling accurate single-ply cutting of digitally-printed pattern sections. Our starting point for this research was the existing Bodymetrics digital personalization of jeans. This retail custom apparel process typically uses 3D whole body scanners and visualization systems. It has the following stages:

Figure 1: Custom clothing current digital data flow.

Current data flow (Figure 1, developed from the flow process as described by David Bruner, TC2) is supported by a proprietary data tracking system, where the outcomes of each stage are uploaded and progress monitored from the entry of the customer’s details at onset to jeans being delivered to the customer at the end. Retailer or manufacturer can access the progress of the garment at any stage. The current digital data flow has six stages: design selection, size and shape capture, pattern generation, virtual try-on and cutting and manufacture. Design Selection. Anderson et al. (1997) identified four options through which a consumer may collaborate in the design process. Design options with standard sizes, co-design, totally custom and clothes clones. Of those four, co-design most closely resembles the activity undertaken by Bodymetrics, where a professional sales person guides the customer through a choice of bespoke and couture jeans


855

with either a skinny fit with a shallow yoke, or a super skinny fit, with a deep yoke and a curved fit. Figure 2 shows body rise depths and leg shapes – straight, drain pipe and boot cut with leg lengths. The materials are available in several colors, in two weights (12 and 14 oz) with two percent elastane. In addition, there is a non-stretch fabric range available for men. The trimmings include contrast or self-stitching, with either traditional or crystal studs.

Figure 2. Available styles and materials.

Size and shape capture. The 3D scanning unit is used to automatically capture 3D shape, in the form of a point cloud, and up to 140 measurements. Both are verified visually prior to being uploaded for pattern generation. Pattern Generation. A proprietary system, which uses both shape and measurement data is combined with a commercial pattern alteration system, using standard CAD pattern processes to generate patterns. Static Virtual try-on. The seamless 2D pattern is then "wrapped" in real-time around a static 3D virtual image of the customer to verify fit, shape and style of the garment envisaged. Interim adjustments may be made to style and fit prior to generation of the final pattern, lay plan and manufacturing specification. Cutting, manufacture and delivery. Lay plans are uploaded for processing by an automated, single ply cutter and a flexible, quick-response assembly system used to manufacture the jeans. All jeans are packed in a proprietary box for dispatch: either to the retailer for collection, or directly to the customer’s home. The tracking system is used, not only to store customer data but also to monitor progress through the production process.

856


Stage One Introduction of digital print and embroidery to existing personalized jeans offer We now turn to the ways in which the process described above was extended for stage one of our research. We selected two fashion models, one male and one female, to deputize as customers for the project. Each model was scanned in the Bodymetrics TC2 3D body scanner. A slim fit jean was selected for each, so that a suitable non-stretch pattern could be generated for a cotton drill fabric that was of a similar weight and construction to denim. Pattern preparation. The patterns received from Bodymetrics as .dxf files were imported into Adobe Illustrator for viewing. A preliminary pattern was exported to a plotter so a toile could be cut to check the fit – not normally part of the manufacturing process, but identified as desirable to check garment fit prior to developing the print. The Illustrator files were adjusted to remove seam allowances, though it was recognised that future iterations of patterns could be generated without seam allowances. Shrink testing. Textile ink-jet printing systems require pre-treated fabric substrates in order to permanently fix colorants and give good fastness. However, as indicated by May-Plumlee (2005), due to pre and post processing, these preparedfor-print fabrics (PFP) are inherently unstable. We therefore conducted a series of shrink tests to establish the dimensional stability. An 8 cm square grid was printed across the width of the fabric, which was steamed and washed; this was repeated three times so that an average shrinkage formula could be calculated and applied to the final artwork before printing. Print design selection. Two approaches were employed for the print and embroidery design for the prototypes (Figure 3). For the men’s jeans, artwork was selected from a student design project; and, for the women’s jeans, a Research Fellow developed the artwork. Following design selection the artwork was imported into a CAD package to be engineered to fit the pattern pieces. Engineered Print. An engineered print is a design tailored to fit the pattern pieces of a garment in such a way that, when assembled, there is a degree of continuity so that the image flows unbroken around the body. Many historical craft designs for print and embroidery are tailored to fit the garment. Notable designers of the 20th Century such as Sonia Delaunay, Emilio Pucci, and Gianni Versace favoured engineered screen-printing and, more recently, Tristan Webber, Jonathan Saunders and Basso & Brooke have employed the technique using digital textile printing (Figure 4). None of the existing proprietary fashion/textile software packages allow for the automated matching of a single image through the pattern


857

pieces of a garment. CAD programs were used to generate suitable artwork for the print. Digitized outlines were imported from pattern pieces without seam allowances and a shrinkage formula was applied for use as a template where the artwork could be rescaled and manipulated to engineer the pattern pieces.

Figure 3: Selection of print design.

Figure 4: Engineered print designs by Philip Delamore for Tristan Webber. Reproduced by kind permission of Christopher Moore Ltd.

858


Embroidery was discussed as an alternative approach to print, but it was decided to use it only as an embellishment to the print for the women’s jeans. Print design approval. 3D computer animation allows visualization of the prototype for design assessment and approval prior to printing the fabric created from either measurements or a body scan. For this study we introduced a dynamic visualization (Figure 5). Optitex supplied a browser-based 3D viewer and a 3D animated catwalk visualization for the print approval. A final lay plan was created for the most economical use of fabric and exported for color management. Color management. Using a color management system and a suitable textile inkjet printer, a number of swatches were first printed and post-processed for color approval. Some color management was necessary for the men’s print as the red/green balance was incrementally adjusted and reprinted. The lay plans for both jeans were then printed.

Figure 5: Optitex 3D viewer showing jeans on female avatar. A UK manufacturer who is supplying a major high street retailer is using this virtual approach to evaluate design and fit.

Post processing. Following steaming and washing, the fabric was embroidered for the women’s jeans. Samples of CAD CAM embroidery, demonstrating a range of techniques, were presented and discussed by the research team. It was identified that simple small blocks of color could be effective with the digital skyline print and reference the concept of architecture, whilst maintaining the abstract quality


859

of the design. The digital image was imported into the embroidery CAD CAM software program to form a background with small blocks of satin stitch digitized in relation to the print image. A paper printout of the combined design was used to register the embroidery onto the digitally printed textile. The design was embroidered using the CAD CAM computerized embroidery machine. Cut and assemble. The prototype jeans were manually cut and assembled using traditional methods, fitted on models and photographed. (Figure 6). Stage two – Digital-print linings and workflow development While the digital product development process was demonstrated in stage one, there were several issues identified within the digital workflow that require manual intervention at this stage – CAD workflow, fabric stability, print registration for single-ply cutting – and as such, make the process too labor, and therefore cost, intensive to be commercially viable for mass customization at this time. Digital print lining. Nutters, of Saville Row, founded by Tommy Nutter in 1968, re-launched in Harrods department store as part of the Bodymetrics bespoke couture offer for women’s tailoring, including Vivienne Westwood, with the suit worn by Bianca Jagger for her wedding. It was proposed that a new offer could include a digitally-printed lining which would give the customer added value through an enhanced retail experience and a unique personalized product.

Figure 6: Men’s digitally printed jeans, with lay plan.

"Co-design" suggests that, by involving the consumer in the design process, there is increased customer satisfaction and, consequently, increased brand loyalty. It has been noticeable that many of the online design interaction interfaces such as

860


Nike-iD and Mi-Adidas have now developed into store concepts, as the value of having agents to guide the customer through the design process has been recognized (Figure 7).

Figure 7: Nike iD store, New York.

The traditional process of having a bespoke suit made could be seen as the model for these new concepts, i.e. the choice of fabric, color, lapel, cuff, button etc and the process of fitting and alteration are in effect forming a bond between the customer and the product. In by-passing some of these processes using technology, such as 3D body scanning and automated pattern extraction, it could be argued that the links between the customer and the product are being weakened and that by reintroducing additional co-design processes, these bonds may be reinforced. In addition, for those customers that have not experienced having made-to-measure clothing before, the co-creational aspect of the retail experience may prove worth the premium that has to be paid for these products at present. The potential to then close the gap between such personalized experiences and move to a Mass customization model may be achieved. Following a consultation process it was decided that a prototype system would guide the customer through a simple process, where a personalized tattoo artwork


861

could be created for the lining of a suit jacket, and a number of customizable items chosen by the customer and facilitated by the agent would give each customer a unique design which could be visualized in real time to facilitate text and images to be flowed into a pre-determined template, and then uploaded and digitally printed (Figure 8).

Figure 8: Concept development.

For stage one the jeans were developed with a single prototype design for each gender. But in a live retail situation it was envisaged that a range of designs and colorways would be offered to personalize each pair to the customer’s requirements. In the case of the suit linings, it was proposed that a single print design would be offered with a modular approach to personalization. Within the design layout a number of customizable items are offered in the way a tattoo would be personalized for the individual from "flash art". E.g. the name of a loved one, a star sign etc. (Figure 9). The lining would then be cut and sewn using traditional methods. Digital Workflow. There were three stages in the digital workflow development for each case: (i) CAD workflow; (ii) fabric stability; and (iii) print registration and single-ply cutting.In order to customize an engineered print to fit the individual, we have developed a methodology to allow an image to be reproportioned to fit to differently sized patterns, as there is currently no proprietary software to manage this process (Figure 10).

862


Figure 9: Artwork for printed suit lining.

Figure 10: Re-proportioned right back lining pattern pieces with print.

Current systems allow a repeat to be flowed into a pattern that may be adjusted in real-time. It is noted that existing 3D CAD systems allow for images to be wrapped or projected onto a surface, and UV mapping techniques currently employed to make a 2D map which represents a 3D model. This map can be associated with a texture, and subsequently used to wrap the image onto the 3D surface. This has been identified as an approach to be further examined with dynamic movement of the customers scan for design visualization.


863

The most significant limitation of the inkjet textile printing process used for jeans and suit linings is fabric stability associated with the number of processes in pretreatment, printing and post processing. Using reactive dyes requires pre- and post-treatment, which has significant effects on color and shrinkage, and require significant time to be taken on testing and processing the fabric. Two approaches to address this limitation have been identified. One is to use Pre-shrunk fabric treated to receive existing reactive printing system. The other is to use a pigment printing system, which does not require pre-treatment and has a less rigorous postprocessing requirement (Figure 11).

Figure 11: Pigment inkjet system for placement printing. These pigment systems, including those with nano-scale pigments, offer a reduction in processing time and increased fabric stability.

In order to accomplish the identified digital workflow it was recognised that the manual cutting process needs to be replaced with an automated system, and it is proposed that single ply cutting be incorporated into the fully digitized process. Existing systems allow for camera recognition, using edge detection for leather cutting, and stripe and check matching for automated cutting of nested patterns. A suitable optical recognition system has been identified and tested to enable the engineered patterns to be cut from a lay plan, using visual recognition of the printed pattern pieces, as opposed to importing a cutting template that is applied to

864


pre-printed fabric (Figure 12). This method ensures a more cost and time-effective solution to printing.

Figure 12. Computer-controlled, single-ply cutting system. Visualization of lay plan for digitallyprinted jeans.

Recommendations and Conclusions Having trialed two case studies for personalization of clothing, we can recommend that there is a need for significantly more time to be invested in testing these processes in order to refine and improve them. The digital stages of product development need improved Graphical User Interfaces to engage the consumer and enable them to interact with archives or upload designs. While this is relatively simple for a T-Shirt (where the product is pre-assembled), printing engineered patterns for a wide variety of garment types should be possible and desirable, if it can achieve the aims of producing better fitting clothes, reducing waste, and delivering greater customer satisfaction. Dedicated CAD modules are needed to merge 3D body information with 2D design information to enable simple design communication and adaptation to suit the needs of the customer in collaboration with the designer. This blueprint then requires embedding within a robust system to translate these engineered digital patterns into physical ones. Recommendations for further work to continue the development of the proposed digital workflow include testing new and emerging pigment printing systems. These may allow fewer processing steps and greater fabric choice; and further investigation of automated cutting and manufacturing processes for different


865

product types. Significant progress has been made in the development of a digital workflow for personalized products. There are several processes still to be addressed in order to move towards localised production, with an aim of providing a one-stop solution for a retailer/manufacturing unit. This would offer a significant reduction in the time to deliver a personalized product to the consumer who seeks a sustainable and ethical solution to the purchase of customized clothing. Acknowledgments The authors would like to thank the following: Ceri Isaac, Polly Kenny, Suran Goonatilake, Jon Walters, Tania Fauvel, Rachel Du Preez, Anna Sjoberg, Mons Lindstedt, Patricia Matzdorf, Sharon Blackford, Megan McGuire, Rachel Bradburn, Annie Robinson, Sonja Boussou, Oliver Furlong, Abibatu Fofanah, Jason Quaglia, David Mason, Ricardo Matos, Julie Webb. The authors would also like to acknowledge the co-operation of the following organizations: Bodymetrics, Union Models, Gerber Technology, AVA, Optitex, Fashion Business Resource Studio, Nutters, e-Mode, Gerber Technology.

References Anderson, L., Brannon, E., Ulrich, P., Marshall, T. and Staples, N. (1997). Discovering the process of mass customization: a paradigm shift for competitive manufacturing. Annual Reports and Research (National Textile Center, USA). Fralix, M. (2001). From Mass Production to Mass Customization, Journal of Textile Apparel, Technology and Management. 1(2): 1–7. May-Plumlee, T. (2005). Behavior of Prepared-For-Print Fabrics in digital Printing. Journal of Textile Apparel, Technology and Management. 4(3). Oliver, B., Mahoney, M. and Shim, S. (1993, Winter) Profile of male made-to- measure customers: Body characteristics and purchase selection. Clothing and Textiles Research Journal. 11(2): 59–62. Tseng, M. and Piller, F. (Editors, 2003). The Customer Centric Enterprise: Advances in Mass Customization and Personalization. New York: Springer. Introduction to Part VI.

Author Biographies Philip Delamore is a Senior Research Fellow at the London College of Fashion, University of the Arts London, and director of the Digital Fashion Studio which encompasses 3D scanning, CAD, Digital Printing and Rapid Prototyping. He is an associate of the Centre for Fashion Science and the Textile Futures Research Group, with interests in the creative exploration of new technologies across disciplines. He is currently

866


involved in a range of research projects which bring together experts from the human, material and computer sciences to develop new approaches to design practice. Jennifer Bougourd, Senior Research Fellow at the London College of Fashion, University of The Arts, London. Chief Designer for major UK manufacturer before entering higher education. Engaged for over 10 years in research involving partnerships with education, Government and industry. Became manager of the Centre for 3D Electronic Commerce – a project that pioneered use of 3D scanning technology in the UK and organised SizeUK, the first size survey for fifty years, which provided a blueprint found useful by other countries. Following membership of eTCluster, European Standards and Eurasia-Tex projects She helped establish "3D Direct" at LCF, a unit supporting advanced 3D technologies. An active researcher in digital fashion, mass customization, anthropometrics, health and ageing studies; a member of the Centre for Fashion Science and Chair of The Textile Institute Special Interest Group on Fashion Technology. Continues to publish and broadcast on developments in the field, at home and overseas.

3

Mass Customization in Architecture and Construction

867


3.1


Amir E. Piroozfar (Poorang) School of Environment and Technology (SET), University of Brighton, United Kingdom Olga Popovic Larsen School of Architecture, Royal Danish Academy of Fine Arts, Denmark

Despite the relatively long history of inadvertent using of mass customization in the building industry, very little and scattered systematic attempts have been made to apply it within the field knowingly. Almost none of them have successfully avoided the predicament imposed by the dominance of its pure manufacture interpretation or have been able to obliterate the failure of its predecessor, mass-production. This chapter attempts to set the scene for customization in the building industry with special reference to building envelopes. It investigates a series of projects in which the approach can serve the purpose of a customization approach to the design, fabrication and implementation (DFI) processes of building envelopes. Some existing examples have been chosen and investigated to show how the notion should and could get adopted by the industry. Whenever applicable, comparisons to the strategies in the manufacture industry have been given to help keep the track and support the main idea of knowledge transfer. In the end the chapter sums up the findings and comes up with some general suggestions that would improve the notion within the building industry.

Introduction Customization has been used for some time now in the car and computer industries. As such, this approach has offered better quality end products, greater customer satisfaction as well as greater profits. Yet, customization is not a wellknown paradigm in the building industry. It is even less known when it comes to the specific areas in the construction industry such as its application onto building envelopes. However, in recent years some architects have been using approaches that are very close to customization to fulfil specific tasks. This evolutionary process of introducing customization into the building industry has not happened overnight but is a result of a gradual symbiosis between the involving bodies in this industry and as a result of an ever-changing atmosphere which rules the whole business of the built environment. The buildings, which are addressed here, are 869

870


not particularly chosen because of any of their architectural aspects, positive or negative. They are merely important to this research to show how the idea of customization can be comprehended, formed, and developed within an architectural project with special focus on building envelopes. In addition, this chapter will provide a further set of possible benefits which customization can potentially offer, if employed efficiently, to other sectors of the construction industry. In other words, the chosen buildings are good examples that show how a customizable envelope can encourage higher standards in design, more efficient components production processes as well as better controlled assembly and construction process. These as a result create buildings that perform better with regard to longevity, maintenance and operability. Aims and Objectives The current chapter is an implication of a research project on application of mass customization on the design, fabrication and construction processes of building envelopes. It aims to draw a clear outline of the existing solutions which may be used as platforms to develop a customizable envelope system (or set). Although what customizability means in its source fields, how it should get interpreted and comprehended in the destination domain, what are the real obstacles on the way of this knowledge and technology transfer or what additional actions need to be taken to make this approach work and why, are all very important questions, they will not be dealt with here. To emphasise the fact that mass customization strategies should be different in the two domains, however, some differences will be highlighted. What is targeted here is to ponder on what is going on in the building industry with special reference to the Design, Fabrication and Implementation (DFI) processes of building envelopes, to have the experience of mass customization in the manufacture industry in mind, and to investigate what can be learnt from the building industry to improve the notion of mass customization which has been adopted from an absolute outer discipline, on the DFI processes of building envelopes. Of the different strategies presented in this chapter some have tight links to those of the manufacture industries whereas the rest have very limited connections to the classical approaches of customization, if any. The choice of a strategy and how to develop it very much depends on each individual case. Whether it is for an existing envelop system provider to improve an existing system, for an existing architectural scheme to dismantle the façade and adopt a new customizable one, or for a producer/client who wants to develop a new product/scheme from scratch benefiting most from what mass customization can offer are the main factors


871

which determine the most suitable mass customization strategy to be taken. This point is very important that choosing the strategy is the first step to further a customizable approach. The technical, operational and executive issues should be the next steps to be dealt with and resolved. Why Is Mass customization Different in the Building and the Manufacture Industries? Traditionally there are some striking differences (and even contrasts) between building and other industries. These differences extend back to the early history of architecture, handicrafts and artefacts. They are rooted in the very nature of the building area and those of the others rather than just the approaches to craftsmanship, the evolution of them, the ways of making or fabricating and the delivery phase of the final product. The two industries are fundamentally different both in terms of the final product and the production process. Understanding differences is believed to provide a promising route to the successful application of the notion in the building industry with special reference to the building envelopes. The flow of work can be seen in Figure 1.

Figure 1: Comparative flow of work in building and other industries (Piroozfar 2008).

872


The size of final product The size of the final product in the building industry is by far larger than that of the other industries. The issue of size subsequently will take some other dependent issues into account each of which is more or less affected or determined by the size of product. When talking about the size of a product, one may mean the absolute dimensions of it or by contrast the relative proportions of its component or even more radically its size and proportions comparing the other products in the same range or in other areas. Having the latter point of view building industry has been producing a range of the largest human artefacts ever (except for the particular products in some other industries such as aviation or ship industries which are extremely limited in variation and number in comparison to building industry). Proportional/dimensional relation with the customer One of the first areas which are definitely determined by size is the dimensional dialogue between the user and the product. Dimensional dialogue means the ways in which the customer makes their utilitarian interaction with the product. In a rather mathematical sense this interrelationship can be perceived in two different ways. The user/customer is either inscribed in the product or circumscribed about it. The terms "inscribed" and "circumscribed" are here intentionally employed in their very technical (geometrical) meanings to magnify the difference. By "being inscribed in" it is meant that the customer can easily be placed or located within the product (like a building, a ship or an airplane) whilst "being circumscribed about" means that he or she can enforce their direct physical (and mental) dominance or control over the product (like a computer or a mobile phone set). In other words in the former case the perception process of the product can hardly be completed from outside. The complete acuity needs usually an inner spatial perception as well. In this case having an overall view/perception about an object is hardly probable. By contrast, in the latter case not only is the user wholly in touch with the product but he/she is in control of it from outside. This widely affects the production process in terms of customizability as well as its design/ production process by the people in the respected industries. Mobility The other aspect fairly inspired by size, is mobility of the product. The mobility itself is quite a size-wise issue. Building (except for specific buildings for particular mobile uses) by very nature is an immobile artefact while almost all end products of other industries are moveable. Immobility inherent in a building gives


873

it an inflexibility which puts serious restraints on its alteration based on the new emergent market demands. On the other hand this characteristic will make the building dependency on its context inevitable. Other products can be easily produced globally and subsequently used globally. But buildings although can hardly be produced globally they can never be used globally. They are to be used locally or in a very optimiztic view regionally (if the cultural and ethnical backgrounds are not taken into account). As a result of the product-context relationship, the equation of functionality-aesthetics will be affected in two distinct domains. In better words the building industry’s by-products are rather gravitational-based therefore, their production process is consequential. The notion of variation Building industry has dramatically changed during its recorded history. Alteration is the best survival strategy for each and every industry. In addition to a diachronic alteration, which is crucial for long term existence, a synchronic change is an unavoidable factor for being able to survive within the market against the rivals. This synchronic alteration is referred to as "variation in theme" of a product. The point, however, is that the concept of variation is totally different in building industry comparing to that of the others. This is first of all because the building industry is more complicated than other industries not in terms of technology, production process or the product itself but because of the consequential nature of its production process. The process in building industry is a rather linear process that can be hardly broken into non-linear sub-processes. The second reason is that the number and nature of the factors involved in variation of a product are themselves by far more colorful in building area. In other words the number of the components to be configured by the customer to achieve a customized product in an ordinary product such as computer are limited and usually do not deal with its constructional specification or infrastructural requirements. But in a building almost every aspect which can receive a customization, from a customer’s point of view, is related to and will affect its construction deeply both in degree and form. Thus, the notion of variation, providing an appropriate underlying infrastructure to effectively offer it to customers and at the end of the process achieving it is too different from other industries and in some senses too difficult to achieve. Product life cycle and product life Although a great effort has been put during the last two or three decades to upgrade the meaning of "product life" to the notion of "product life cycle", there are some slight differences which may potentially affect the idea of mass

874


customization in building industry. Having this difference in mind the product life (and its life cycle) will give the issue of customization distinct levels of importance. If a product is supposed to last and serve its purposes for a longer period of time the need of being repairable, reconstructable, rehabilitatable and even somehow reconfigurable based on the changing nature of its users' needs and preferences is extremely crucial for it. A car comparing to a home is a good example in which the fairly shorter life period makes the nature of customizability of difference from that of a home. On the other hand the product life cycle (the process of raw material acquisition to recycling the used materials) necessitates a considerate policy and decision making process about customizability. In other words the longer the product life (cycle) is, the more complicated yet crucial the process of the customization would be. The flexibility of customization itself and the extent to which the product can get customized not only at the very stages of design and production (up to delivery stage) but during its serving period (performance) are both a matter of its life cycle (servicing period). Costs The investment rate, the dispersal of capital over a project, its profitability and profit margins in building and other industries are not the same. Economical input and output of a project and the turnover terms cause a fundamental differentiation in the two areas. From this point of view building industry is much safer yet reluctant to change whilst other industries such as car or computer industries are much more risky yet flexible to changes of the market demand, inflation rate and stock exchange fluctuations. Relatively long term turnover periods in building industry make the transition happen in a more judicious and intangible way. This intrinsic difference hence, will affect the areas in which customization may be more effectively applied in building industry. What is meant by this is that the area of authenticity for the customization of a product in building industry is not necessarily the same as those of other industries. Economies of scale The other issue that makes the approaches to customization in building industry essentially differ from those of other industries is the economies of scale. While Ford model T as the first precedent of the mass-production paradigm reached the record of over 1.5 million cars in less than ten years (1908-1916) the building industry could hardly ever pass several thousands of the identical housing units in its zenith of application of mass-production in modern era 30 years later. In other words it can be claimed that the building industry (in terms of final products, not the production of building materials and components) has never effectively


875

experienced the notion of mass-production or passed it over successfully. One of the other determinants in this regard is what Pine (1991) cunningly refers to in his MSc dissertation: the size of the firms which hold the market share in different industries. Quoting from "innovation and small firms" he brought up four examples of building industry (in 1990) in which 67.65% to 84.09% of market share is under control by small firms (firms with less than 100 employees) – over 96% of firms in those industries. This will face the issue of customization with even more complicated problems and not viable to be achieved in the similar ways of the other industries. The customer needs and expectations The expectations and needs of the customers from a widely perceived form of a building – a house – as a human shelter are by far broader than the other industrial products. The range of these needs and expectations are much more colorful and heterogeneous than those of the other industries. So in this sense building deals with a range of objective to subjective goals to provide for its users while the subjective achievement of the final products of other industries is chiefly limited to the aesthetics. Functionality is supposed to be achieved in its highest degree in both areas. Well-being is highly aimed to be provided by and within the buildings. Well-being means all the aspects which are related to the quality of life, work or entertainment within indoor spaces. This will render the notion of customization as a highly critical issue to be achieved in the building industry. First customers vs. further owners The building because of its very nature can not be removed, disposed (or nowadays recycled) and then replaced with brand new products in as fast a cycle as the by-products of other industries can be. The financial restrictions are the main players in this scene. As a result, a quite intrinsic yet still challenging issue emerges. A building has not been and is not supposed to be designed and made as many times as a new customer rises. So the issue of further owners will emerge particularly in the field of building industry. By contrast "further owner(s)" is not such a dominating factor in other areas mainly because each and every customer does prefer to buy a new product and in some limited cases (such as buying a used car) the number of the choices in the supply chain is so vast that can potentially serve any taste and budget in the demand side almost as widely as the new products' market. This is a challenging aspect that makes customizability in the entire product life period as important an issue as in their pre-making period. Although this feature (being customizable even after being produced) is not quite unfamiliar to the other industries, but regarding the nature of the buildings it is

876


completely different in the building industry. This is simply because when the owner of a building is going to be changed, almost all of the pre-defined or pre-set factors of value are deemed to be fundamentally changed. No preferences still remain in value or even more radically exist anymore. The order-make sequence Industrial products have usually benefited from a pre-designed basis (or chassis/platform in a general and a subjective sense) in which the customers' points of view is not directly determining. This basis might not even be sensible or tangible for the customers, users or even the dealers. Limited range of customization (in different faces which are referred to or labelled through different approaches) is applicable to the common basis which is shared between the products of the same group (Adidas sport footwear) or even between the members of the different groups (Black & Decker tools). Some other products are wandering in between (Swatch watches) in which the alteration range makes the issue of identical products with different components or different products with similar components somehow vague. By contrast, first-customer-ordered projects in building industries are mainly made-to-order or at least tailored-to-order products in a wider sense. In other words the sequence of order, design, make, marketing and sale is different in building and other industries. In other industries a rather nonclassical approach to this sequence (design/ marketing/ order/ make/ sale) is targeted and fairly achieved yet in building industry still the classical sequence (marketing/order/design/make/sale) is dominated, in which sale process could move before the make or even the design process based on the type of contract/ agreement. However, needless to say that even the very classical approaches in other industries (marketing/ order/ design/ make/ sale for products of niche market and design/ make/ marketing/ order/ sale for mass-produced products) were still by far different from those of building industries. The position of order process is a determinant in the feasibility of customizing a product. The nearer the process is to the beginning of the chain, the more feasible and sensible the applicability of the mass customization would be to the production process (strictly speaking about the building industry). This confirms again that building industry hasn't passed the era of mass-production successfully while other industries passing that era, experiencing its inefficiencies and having it as an invaluable feedback in their disposal, are trying to make a new equilibrium to respond to the ever-changing nature of market which is very prone to the colorful range of products offered by niche market but still deeply desperate and athirst for the affordable prices promised and provided by the market of mass-production. In this regard building industry should rely on the feedbacks from other industries.


877

The fallacy of modularisation in the building industry The other problem in employing the notion in the building industry which puts on some serious restrictions is misunderstanding of the notion of modularity. In architecture as one of the pioneers in devising and employing the notion of modularity, the exact word of "MODULOR" was first coined by Le Corbusier, the Swiss architect in the late 1950s. What was really meant by Modulor was fundamentally different from the notion of modularity itself at the very same time in other industries or from its current modern interpretations. Not only did the complexity and mathematical nature of the original notion (Modulor in architecture) cause an exceptionally complicated confusion very soon but it consequently led to an entire distortion in the notion between Corbusier’s colleagues, the precursors of modern movement (in architecture) and the next generation. The blight legacy of this misunderstanding was not limited to this but soon afterwards showed the way towards a blind resistance against the acceptance of modern interpretation of and approaches to the notion of modularity in the field. It was just during the past decade that the precise understanding of the meaning of modularity started being both speculatively and practically perceived in building industry. Having the basic differences between the two domains – the source and the destination for the application of mass customization – different buildings have been chosen from all around the world to explain how different strategies of customization can or may be applicable to the building envelopes. The projects are not necessarily chosen based on their architectural significance. They are chosen to best serve the purpose of this chapter. Therefore, having them discussed and criticised, will be from the mere customization point of view and will not reflect their pure architectural values or even their performance. In addition, they are not one of a kind or the only examples. They are believed to be one of the best exemplars to meet the features of the strategies explained. Here are most likely customization strategies to be taken in the field of building envelopes, investigated through some built projects: Customizable Façades Sometimes a course of envelope is designed with some changeable specifications such as dimension, direction, openness, angle, transparency or luminance percentage of the façade which can give different performance capacities to the envelope as well as a distinct look to the building from outside. Most of these customizable façades have been developed and designed to fulfil a physical or technical task rather than just serving the purpose of mere architectural aesthetics

878


for the building. In this sense, they may be regarded as the hard-customizable façades. However, the meaning of a customizable façade can also be stretched over a soft-customization. For instance a building façade can be customized using different cladding to allow for more controlled day-lighting during the day (hardcustomization) while it can be customized using different flood-lighting patterns over its nightscape (soft customization). These characteristics are usually some configurable high tech cladding components which in most cases are controlled automatically, but can also be components that can be controlled and operated manually by the users of the building. L'Institut du Monde Arabe (See Figures 2, 3 and 4) and Bibliothèque nationale de France (See Figures 5, 6 and 7) both in Paris are examples of this type of customizable façades. The first building (Figure 2) utilizes a prototypical façade component with which the translucency of the surface can be controlled automatically (Figure 3). The main functionality of the building is cultural. It includes different types of exhibition spaces and galleries, meeting rooms, lecture halls and study spaces.

Figure 2: L'Institut du Monde Arabe, general view, Photo: A. E. Piroozfar.

These spaces, because of their particular uses, need a range of different lighting specifications which the architects have aimed to provide by different combina-


879

tions of natural and artificial lighting. Based on different factors, such as climatic conditions, level of available day-lighting, desirable lighting patterns for the inside spaces regarding the function of the space in general or with specific needs when a short time application or exhibition is being carried out, metal-bladed apertures on the external façade control the level of incoming day light (Figure 4).

Figure 3: A prototypical façade element, Photo: A. E. Piroozfar.

Figure 4: pneumatic control of steel apertures, Photo: A. E. Piroozfar.

These are centrally controlled by the building operation and maintenance team. However, it very much affects the outer appearance of the building looking like an impenetrable large-scaled safe which contains a very precious treasure of art (perhaps more traditional or classical) or a friendly cosy and welcoming atmosphere for just sitting and unofficially discussing the contemporary art exhibition which is held inside. Both aesthetically and functionally the customizable façade plays a big role in representing the characteristics of the building. The second example (Figure 5) utilizes a slightly different approach. It provides a virtually infinite number of combinations of unbounded positions of the configurable vertical timber blades which serve as shading devices (Figure 6). Despite the normal location for these devices on the outer surface of a façade, in this building they are housed inside the building. Having them accommodated within the course of building envelope, changing them will have the least formal affect on the building visual appearance. This will provide the users – whether they are permanent, long-term occupants such as the library staff or temporary, short-term users like the book readers – with the ability to exactly and flawlessly adjust the amount of light based on the natural daylight, the sky condition, the season, the time and the function of the spaces inside, the lighting standards for each space and finally the personal desirable light and acceptable glare for the specific task

880


they are carrying out. It will also inspire the building’s outlook, ranging from absolute dark shining translucent blocks to four L-shaped opaque cream units embracing the central inner green courtyard (Figure 7).

Figure 5: Four L-shaped blocks forming the main idea of the Bibliothèque, Photo: A. E. Piroozfar.

Figure 6: Timber blades forming shading devices as a customization tool, Photo: A. E. Piroozfar.

Row 1

Row 2

Row 3

Row 4

Row 5 Figure 7: Controlling the incoming day-light and heat by occupants will result in increasing dark transparency on the exterior skin and will render the building with an "ever-changing façade" characteristic, Photo: A. E. Piroozfar.


881

Component-Sharing Customization This approach is supported by the fact that there are wide ranges of pre-made factory-built components which can be ordered or simply found off the shelf or in the producers' catalogues. These components can be shared between infinite numbers of different projects. Figure 8 shows how sash windows are shared between a range of relatively different arrangements of a prototypical façade. Taking this fact and applying it back-to-front i.e. if the components are available off the shelf, they can be used as insets in the different arrangements of surrounding walls; different end products which will be using similar components. On one hand this is not a new idea within the building industry because a palette of different pre-made elements has been at hand for many years. On the other hand the number of pre-made component providers is so vast that the notion of customization might not seem very much applicable in the building industry, comparable to how it is in the manufacture industries. Despite these two facts, though, as the time passes and labor costs continue to soar, the notion of customization is not only infrastructure-wise but cost-wise, productivity-wise, and efficiency-wise more justifiable and even inevitable.

Figure 8: Component-sharing customization; the notion of a component in the building industry is not as industrialised as it is in the manufacture industry. This means that the shared component may be produced in a number which is not comparable to what it means regarding the economies of scale in other industries, Drawing: A. E. Piroozfar.

Contrary to what is widely comprehended as component-sharing customization in the manufacture industries, which can be evolved easily and smoothly to component-swapping customization (sometimes with no clear boundary in between), in the construction industry it can hardly evolve to anything comparable to component-swapping customization. This is because the construction industry has never embraced the idea of platform design; namely developing a chassis with interchangeable parts – the principal approach which has been in use from the early days of industrialisation in car industries. In fact there is no sensible platform in the building industry on which a selection of different compartments can be mounted to give the final product different touches or distinct looks. This is partially due to the dimension variation which is inherently imbedded in the construction industry and the tradition of one-off production. Current research

882


suggests that this can and should be cured if a sensible progress and an in-time evolution in the mindset within the industry occurs appropriately and widely. Open Customization This idea comes from a preliminary interview with architect John Atkinson (HCD Architects, Sheffield, UK). The interview was carried out for a research project that its implications have been partially used to provide this book chapter. Although at that time there was no clear idea of customization, personalization and customer-based approaches in design and construction, the theme, which was discussed there, if well-understood and properly developed will provide great potentials for customization in the building industry. The term "opencustomization" is selected for this approach for several reasons. First of all, one of the main obstacles on the way of customization in the building industry is that the building and construction industry is gravitational-based. This means that construction (of a building), despite the production (of a manufactured product), is hardly likely to be broken into sub-processes and carried out with no consequence. Major progress in breaking the overall process of construction into smaller subprocesses and tasks has been already made. However, the nature of the building as an immobilised product that very much depends on its physical site, on which it should get erected, is still dictating an established trend in the building industry. This nature compels a consequential approach which is hardly avoidable if the whole tradition of building is not about to undergo a significant change. However, taken the existing approach for granted, there are still some short-term innovative remedies or applications which can break this gravitation dominance. One of these solutions which can be employed with regard to the building type, design specifications and the available technology is to both functionally and structurally free the envelope from any task but wrapping the building. One of the outcomes of this been the invention of curtain walls. Freeing the course of envelope from structural task, however, is not a novel idea within the industry, yet it is not enough to provide all requirements for devising a customizable course of envelope easily and successfully. The other reason for choosing the term "open-customization" is that this approach, if well-developed, can promise an absolute freedom in changing or altering the envelope course later in the service life of the building almost as independently and openly as possible with almost no intrusion into the inner life of the building. Thirdly, although roof and façade are both parts of the envelope in a building, by nature they behave differently. Traditionally the roof is deemed to be independent on the walls which may form the façade. This tradition is less dominant nowadays


883

but the physical (not functional) dependence of roof on walls is still overshadowing the outlook in the building and construction industry. In this sense breaking the tradition of this physical dependence of roof on the walls can open up an opportunity for customizing the façades without necessity of intervening in the roof. Selected projects to investigate open customization are the Sage, Gateshead, UK (See Figures 9 and 10) and a bus station in Hamburg, Germany (Figure 11). The first building typifies an approach in which the envelope is liberated from any other task but providing an outer shelter (Figure 9). Limiting the physical contact and the structural dependence of the outer skin on the inner structure has been successfully achieved due to the functionality of the building, the form of the external shell and the architectural design specifications (Figure 10). In addition, the site and the context in which the building is located have had a crucial effect on how the scheme has been formed. The approach can not be directly generalised to any other building application but has great lessons to learn about how the open customization can be perceived and employed.

Figure 9: Sage Gateshead; forming the shelter around the functional structure inside. Open customization through divorce between the intervening tasks, Photo: A. E. Piroozfar.

Figure 10: Sage Gateshead; the finished shell provides a house literally to any independent activity inside. Reasonable customisation is possible both inside and out without interfering with the other side, Photo: A. E. Piroozfar.

The second example is a bus station in Hamburg. In this building, doubling the roof, the architect has separated the weatherproofing task of the envelope from the visual implication of the roof (Figure 11). In other words, the functional and visual tasks of the outer skin have been utterly split. In this sense the functional compartment can be repaired, improved or replaced merely in favour of performance or maintenance of the building whereas the outer slot can be dismantled,

884


moved, or replaced with any other similar or different course for the sake of the aesthetics and appearance of the building.

Figure 11: Bus station, Hamburg. Second reading of open customization: separation through doubling the structure and elimination of dependence. The outer skin splits both over its technical tasks and in its formal and visual performance, Photo: R. Piroozfar.

In addition, the specific approach to customization in this building can potentially serve the purposes of the adaptability, the objective which has been on the agenda for a long time. Curtain Customization As discussed in open customization, curtain wall was one of the ways to free the façade from the load-bearing task in a building. It was mainly invented in the Chicago School by laying very heavy masonry wall wherever they were needed. The only but main difference between these walls with their predecessor, loadbearing masonry walls, was that they could be easily laid everywhere on a floor because no structural load was expected to be carried by them. As the time passed and the technology progressed, the massiveness of the covering layer became an absolute redundancy and started fading out. So the new generations of the curtain walls emerged to fulfil a notably broad range of tasks. The very fact that whether these curtain walls are the only covering layer of a building or they form a secondary cosmetic layer to the main envelope will put the challenge of customization somehow on stake. In other words, if more than one major task is expected to be carried out by the external curtain wall, this may well affect the degree of customizability that the building envelope may or can accommodate.


885

The perfect condition for achieving a full range on customization is to have the curtain wall separated from the main wall course wherever possible. Having done that, the inner main wall will have a big influence on the required or possible technical aspects of the external covering system while the outer curtain can mainly become a palette to meet the specifications entailed to a customized and/or customizable product. Nevertheless, changing the specification of this independent curtain wall might not still be as easy as changing the color of a car based on the customer order. Jakob-Kaiser-Haus, a new parliamentary building in Berlin, illustrates perfectly how curtain customization may form and be employed on a course of an envelope (See Figures 12, 13 and 14). The building façade is formed of two different selfregulating envelope courses (Figure 13). The inner course fulfils perfectly the tasks of a building envelope both technically and visually. Being visible through the second additional course on the outside, has required it to look like as perfect a building face as possible (Figure 14). Yet the secondary envelope is added to give the building a more modern and coordinated look to its neighbouring buildings and to its context (Figure 12). A very thin glass skin shapes the final outline of the building using the steel cantilevers projecting out on the main façade. The number and the frequency of the cantilevers are arranged in a way that the system can remarkably offer high versatility in form. Using no framing as the glazing, the unique employed approach meets the daunting task of wrapping the building very smoothly in one of the most susceptible materials by nature.

Figure 12: Jakob-Kaiser-Haus: Exterior view, Photo: A. E. Piroozfar.

Figure 13: Curtain wall and the main façade, Detail drawing: A. E. Piroozfar.

886


Figure 14: Jakob-Kaiser-Haus, view from outside, Photo: A. E. Piroozfar.

Compartmentalised Façade If a façade is broken to some small compartments, different combinations of those compartments may well provide different variations of the façade. This will be a modularised customization strategy in a narrow sense. It may vary from one project to another, from one producer to another, and from one façade system to another. Therefore, modularity is very much bound to its context. However, the method of compartmentalisation is very important in how flexible and adaptable the façade system will be. What is really crucial is to make a sensible balance between the size of compartments and the number of them. Equally importantly, attention should be given to the joints between compartments, the inter-links between these elements, how they meet the building main façade course (if there is an independent one), the building structure, the roof, and the ground, when the system is to be devised. If they are small enough and the system consists of necessary and enough basic elements, it can guarantee a sensible variation range while keeping the price within a reasonable range. If the number of the compartments grows out of control, although the final façade can more or less look like a niche building façade, costs will rise dramatically in design, production and assembly stages and the construction time will increase. If the size of the compartments grows, the number will decrease and the façade can look like a componentised façade or at its worst towards a prefabricated panelised façade as it was implied originally by the modern movement. In this case if the concerns that will be explained in componentised façades are not considered carefully, the variation will decrease and a repeated pattern will re-emerge; similar to dull and boring façades which once dominated the prototypical architecture of modern era.


887

On the other hand and at its best, this approach, resembling Lego blocks in which a bus modularity exists, can provide in-budget sensibly near-to-niche massproduced façade system; an efficiently developed mass-customizable façade system. However, the common connection pattern which exists in Lego system is not fully achievable. In other words, the schemes are not always easy enough to be simplified as they are in a mini (building) model which is made using Lego blocks. However having an early settlement between the design, production, assembly and construction parties to reconcile the complexity of architectural design, the innovation in component design both individually and in relation to the other parts, and the assembly/construction concerns will dramatically improve the system. Having a systemic mindset and/or approach for architect from the early stages of design will potentially help make huge improvements. Some very good examples in this category are available worldwide such as the new office developments in Potsdamer Platz, Berlin (Figures 15 and 16). Nonetheless, because there is still a long way to go to be idealistically comparable to other industries or what is really required by today’s building standards in compartments design (specially when it comes to joints), in almost all of them an attempt has been made to keep the joint design failures apart from intervening in the expected quality and standards of the final product i.e. the building envelope set. This simply has resulted in decreasing a pure compartmentalised system to a secondary course in which the systemic approach to the idea of customization can be slightly differently comprehended. Selected building is an office building in Potsdamer Platz in central Berlin. The compartmentalised façade is formed of terracotta tiles in combination with ceramic linear elements (Figure 15). This Lego-typed façade, as discussed before, is not appropriately capable of fulfilling the technical tasks of a façade on its own. Therefore, it has been added as a secondary course and its systemic role is very similar to what was addressed in curtain customization. An inner simple and most likely dull glass façade is covered with a colorful yet consistent touch of a ceramic layer which offers different degrees of openness based on the requirement of the spaces which it is covering. Although the inner layer will safely take care of the waterproofing and weatherproofing tasks as well as providing a comfort zone inside, there are still some understandable concerns about the joints including the joints between the compartments, the joints between compartments and the support structure or links between the envelope system and the main body of the building wherever applicable (Figure 16).

888


Figure 15: Office Building, Potsdamer Platz, Berlin, the general view, façade is made of compartments which can be effectively used in offering customization, Photo: A. E. Piroozfar.

Figure 16: Office Building, Potsdamer Platz, Berlin, the compartments joint details, Photo: A. E. Piroozfar.

Componentised Façade The idea of big panels to cover a notable area was put to practice when the high formability of precast concrete and an unprecedented call for housing came together during the 1950s and 1960s. Componentised façade can be assumed as one of the successors to the precast concrete façade panelling systems of that era. Initiated by the concrete mouldability, the idea got furthered using different major techniques. Panelised façades and componentised façades can use different techniques of wet or dry backing for a put-together set of different materials and components to form mid-range or large-sized panels. While the term panelised emphasises more on panel as a whole and denotes preliminary concrete panels, the term componentised highlights the idea of interchangeability which can be employed and promoted by this approach benefiting from the idea of customization. A componentised façade normally is a panelised system with different supporting systems in which literally there should be no limitation on the type of materials or combination of components in one panel. The most crucial challenge faces when it comes to variation on form. Usually different systems are a bit slow and/or incapable of adopting curves. However, the existing systems based on the level of demand can employ some new techniques or modify some existing ones


889

to accommodate non-planar surfaces. These componentised façades can represent one of the clearest types of customization of an envelope course in which different types of pre-made small and mid-sized components can be chosen and put together. Adding a supporting system in form of dry framing or wet mouldable supporting backings will then form a large panel which can be made in a fullycontrolled condition of a factory and then to be transported to site and to be assembled. Although the fact that these panels are factory made is a huge step forward, very little effort, if any, has been made to improve the on-site interrelation between the components i.e. joints, dimension coordination, tolerance design and movement pattern of adjacent panels. A new housing project in London Docklands, on the River Thames' bank and near the famous Calatrava’s pedestrian bridge, very well represents a componentised façade (See Figures 17 and 18). Fairly restricted on form, this approach has provided a virtual platform for different materials and components to be put together and jigged or moulded using steel or concrete backing.

Figure 17: New development, Docklands, Thames' bank, general view, Photo: A. E. Piroozfar.

Figure 18: New development, Docklands, Thames' bank, Detail, Photo: A. E. Piroozfar.

Despite the limitation on form, some of the producers have managed to use the accumulative resultant of the technique and knowledge they have gathered over time in personalizing the infrastructures of industrialised envelope system and have developed on-off solutions based on this system. These solutions, which have a nature of niche products in the manufacture industry, use the very basics of mass-produced façade products yet combine them with more inspiration and creativity for every individual project to offer unique products which might be the first and the last one of their kind.

890


Discussion and Conclusion Although very little has been done on customization in the building industry, the application of such industry-initiated paradigms seems to be promising a big selfawareness and a huge leap forward. There are no applications of customization in the construction industry in general and in building envelopes in particular yet. However, there is a lot to learn from scattered experiences around the world regarding customization and its application on building envelopes. There are some initiatives that may help facilitate the way of customization in the building industry. However, to make this happen the mindset in the profession is the one that most needs a revolutionary reform first. A couple of projects were described as part of this chapter which can be used as archetypes to develop an organizational approach towards customization in the field. It is clear that the notion of customization within the building industry, its limitations, its target areas, opportunities, strategies, are very different from those in the manufacture industries. This is not surprising because the two disciplines themselves are utterly different. The success of implementing a broader concept of customization in the building industry will not be achieved unless both the industry and the mindset of designers start changing. To achieve mass customization, the building industry will need to make changes and a leap forward, towards a more agile, versatile, fast-responding, customer-centric, and just-in-time response attitude and take the respective action to fulfil this need of clients.

References Piller, F.T., R. Reichwald and M. Tseng (2003). Proceedings of the 2003 World Congress on Mass Customization and Personalization (MCPC 2003): Competitive advantage through customer interaction. Technische Universität München, Munich. Proceedings of the 2007 World congress on Mass Customization and Personalization (MCPC 2007): Extreme Customization, Massachusetts Institute of Technology. Pine, B. J. (1991). Paradigm shift: from mass production to mass customization: MSc Dissertation, Cambridge, MA., MIT Sloan Sschool of Management. Pine, B. J. (1993). Mass customization: the new frontier in business competition. Boston, Mass., Harvard Business School. Piroozfar, A. E. (2005). Mass-Customization: A new customer-based approach to design and delivery system in the building industry, Part I (The Approach). PhD Students Research Seminars. The University of Sheffield. Piroozfar, A. E. (2005). Mass-Customization: A new customer-based approach to design and delivery system in the building industry, Part II (Achievements). PhD Students Research Seminars. The University of Sheffield.


891

Piroozfar, A. E. (2005). Mass-Customization: applicability and complications in the building industry. 3rd Interdisciplinary World Congress on Mass Customization and Personalization MCPC2005, Hong Kong, China. Piroozfar, A. E. (2008). Mass-Customization: the application on design, fabrication and implementation (DFI) processes of building envelopes. PhD Thesis, The University of Sheffield, School of Architecture, Sheffield, United Kingdom.

Author Biographies Dr. Amir E. Piroozfar (Poorang) has recently joined the School of Environment and Technology (SET), University of Brighton as a senior lecturer and researcher. Prior to taking up the post in 2009, he has been researching, teaching and practicing both as an architect and an urban designer in different countries for nearly 15 years. He finished his PhD on "Application of Mass Customization on Design, Fabrication and Implementation of Building Envelopes" in the School of Architecture, University of Sheffield in 2008. His research interests include Mass customization in Building Industry, Digital Fabrication and Digital Craftsmanship, Building Envelopes (Theory and Practice), BIM applications, VR (Virtual Reality) and AR (Augmented Reality), Urban Façades (Theory and Practice), Urban Morphology and Urban Evolution, Sustainability with special reference to the Third World countries (Sustainable Environment, Sustainable Development, Sustainable Design), and Fuzzy Logic (the application on qualitative data assessment in built environment). Contact: [email protected] Prof. Dr. Olga Popovic Larsen leads the Structures in Architecture research and education group at the Royal Danish Academy of Fine Arts School of Architecture in Copenhagen. Before coming to Denmark for 13 years she was at The University of Sheffield School of Architecture in UK where as an Associate Professor she was researching and lecturing. Her research is into fields that bridge Architecture and Structural Engineering, such as Structural Morphology and Advanced Structural Systems including Reciprocal Frames and Tensegrities. Special interests include Sustainable aspects of Mass customization and Design for Disassembly and Adaptability. Contact: [email protected]

3.2

Mass Custom Design for Sustainable Housing Development Masa Noguchi Mackintosh School of Architecture, The Glasgow School of Art, United Kingdom Karim Hadjri School of Planning, Architecture and Civil Engineering, Queen’s University Belfast, United Kingdom

The societal pressure on sustainable housing development is on the rise. Homes need to be socially, economically and environmentally sustainable in response to the wants and needs of individual homebuyers/users as well as society. However, existing housing design approaches being applied by today’s homebuilders barely lead to the accomplishment of the sustainability agenda. Mass customization was seen as one of the potential means to tackle issues arising in achieving the housing sustainability. Based on the notion, a systems approach to sustainable homes and the interactive design communication tool were introduced. This study led to a suggestion that along the line with the research on mass customization, the way to mass-personalize a house after occupancy, which may need to involve inclusive design approaches, should be examined for the delivery of truly sustainable homes that satisfy the dynamic and diverse market demands and requirements for housing over the lifetime.

Introduction Since "Sustainable Development" was initially advocated by the World Commission on Environment and Development (WECD) in 1987, our society has urged the building industry to produce sustainable homes which contribute to reducing energy and material consumption. D'Amour (1991) claims that "housing is an environmental industry." Housing is constructed with, and operates on products from surrounding environment. In response to changing social values, demands are being made of the housing industry to fewer resources, both to build and to operate structures. Indeed, builders are beginning to employ resource saving strategies as marketing tools (Friedman et al. 1993). Housing manufacturers in Japan have been gaining a worldwide reputation for their unique design, production and marketing approaches to the delivery of sustainable mass-customized housing. The prefabricated housing sales in Japan dominate 12.4% of the market share today. Total 1,290,391 houses were newly built in 2007 and among them, 892


893

160,347 houses were prefabricated (JPA 2008). Moreover, in response to the societal pressure on sustainable development, the housing manufacturers started producing net zero-energy-cost homes through a net metering arrangement that enables the occupants to use their own power generation to offset their electricity consumption over a billing period (Richard and Noguchi 2006). Their industrialized housing is often equipped with a number of renewable energy technologies, such as a solar photovoltaic (PV) power generating system, an air-source heat pump and a combined heat and power system, and they tend to be installed as standard equipment rather than options today. Contemporary consumers may no longer be satisfied with minimum quality housing which merely corresponds with the adequacy, suitability and affordability problems. Housing choices can be seen as major lifestyle and investment decisions. In other words, today’s homebuyers are looking to purchase a customizable house at an affordable price that adapts to the societal demographic changes with regard to socio-economic profiles. In order to satisfy consumers' individual needs and demands for contemporary homes, builders are urged to consider the application of an innovative design approach to housing. Traditionally, homebuilders are practicing three design approaches: production, semi-custom and custom design (Smith 1998; Noguchi and Hernàndez 2005). Production builders are organized for high volume construction and they usually produce ready-built model homes that are designed on a speculative basis. Builders, who apply the semi-custom design approach for their housing development, are often called semi-custom builders since they combine characteristics of ready-built and custom-built homes. In order for the semi-custom builder to consider clients' desires, requirements and expectations for housing, the design modification of a model house selected by the users is carried out based mainly on dialogue between the sales person/designer and the homebuyer. However, the dialogue that necessitates a number of meetings may still take a long time to reach the final design of a house in question. Custom builders start from a blank sheet of paper or computer screen to create a one-of-a-kind home. The user participation approach can be considered as the optimum way to customize a new home in response to the wishes and needs of individual homebuyers. However, the longer time required to design and build combined with lost economies of large-volume work leads to the higher prices typical for custom homes (Sadeghpour et al. 2006). In short, homes need to be socially, economically and environmentally sustainable in response to the wants and needs of individuals as well as society. However, existing housing design approaches being applied by today’s homebuilders (and

894


architects alike) barely lead to the accomplishment of the housing sustainability agenda. In order to tackle issues arising in achieving the hosing sustainability, the emerging notion of mass customization was introduced to homebuilding operations. The following sections brief the notion and explore the potential means for the delivery of sustainable homes tomorrow. Mass Customization Concept Mass Customization is an oxymoron. The term is composed of two opposite notions: mass production and customization. The notion was anticipated in 1970 by Alvin Toffler in his book entitled Future Shock. Toffler (1970) asserted that maximum "individual choice is regarded as the democratic ideal" and expressed anxiety at the emergence of more standardized mass culture and lifestyles in the future. In 1987, the term was eventually coined by Stanley M. Davis in his book entitled Future Perfect. Davis (1987) delineated the concept as follows: "The world of mass customizing is a world of paradox with very practical implications. Whether we are dealing with a product, a service, a market, or an organization, each is understood to be both part (customized) and whole (mass) simultaneously… For mass customizing of products, markets, and organizations to be possible, the technology must make it economically feasible in every case." Furthermore, in 1993, Joseph B. Pine II profoundly systematized the general methods of mass-customizing products and services. Pine II (1993) regards mass customization as "a synthesis of the two long-competing systems of managements: the mass production of individually customized goods and services." In many industries, this innovative concept has already been introduced to product design in order to accommodate the unique demands of each consumer. Mass customization is based on user participation in the product design decision-making process. Thus, before discussing how the concept can be applied to the housing delivery process, the meaning of user participation, which is still vaguely understood today, should be reviewed. Meaning and Criticism of User Participation Generally, users, community or citizens participation is understood as a means to meet inhabitants' requirements in terms of housing design and planning, as well as improvement of landscape and public amenities. Its success depends on the extent to which full collaboration of professionals, local authorities and users in the participatory process is effectively carried out. Although the concept of participation is a few decades old, there is still great concern about its interpretation and


895

use. Habraken (1985) indicates that participation has two meanings in opposite directions. Firstly, it is seen as a means to enable users to make decisions; thus, this transfers the decision-making power to the inhabitants. However, the process requires essential changes within the administrative structure. Secondly, it is perceived as a change of approach, however, within the same structures. The users' comments are taken into account and are guaranteed to be implemented (Habraken 1985). Participation is also identified as a "face-to-face aggregation of individuals who share a number of values important to all, that is to say a purpose for them for being together" (Beheshti et al. 1985). However, participation has more chance for success in societies where citizens are free in deciding the future of their environment and their concerns are taken into consideration by local governments. It is also related to user control over decision-making which here again requires fundamental changes within the established political and administrative structures (Beheshti et al. 1985). Malpass (1979) claims that public participation has lost its role of a way to communicate users' interests and concerns to decisions-makers, arguing that participation should be "a way of discovering differences of opinion and conflicts of interest." Participation is also identified as "the means by which the victims of design decisions could influence the decision-making process" (Hellman 1973). Some professionals believe that participation in design is simpler than in planning; however, it is crucial that architects should learn how to deal and efficiently communicate with participants (Hellman 1973). Although many professionals have attempted to implement user participation in order to produce, for instance, responsive environments or redevelop unpopular estates, views concerning the validity of participatory processes are still divided. Criticisms are related to the fact that they are time-consuming, complicated, costly and affecting progress of administrative work. Hamdi (1991) argues that despite some successes, participatory processes do not always lead to user satisfaction and more efficient maintenance of buildings and open spaces. That is to say that the validity of user participation is still creating cause for concern in many countries. In this sense, participation in design can be criticized for being unable to fulfill users' housing requirements. The goals of participation need to be clarified and the value should be explicit. The following section identifies a mass custom design system model for the delivery of quality affordable homes with due consideration of the meaning and criticism of user participation.

896


Mass Custom Design System Model One of the successful mass customization approaches that can be applied to the homebuilding process is the modularization of housing components and the variations accomplished through the mass customization process can be quantified in the light of Set Theory, a branch of mathematics developed by Georg Cantor. The total number of possible ordered pairs (or combination) of given standard housing components can be simply calculated. To bring the concept of mass customization into effect, a total coordination (or systems) approach to considering products and services within the housing delivery process needs to be taken into account since to design, build and market a home requires consideration of these two aspects. In short, mass customization, when it is considered as a set of systems for designing, producing and marketing a product, is impossible if either customizable products or communication services are absent. To formulate the means to mass-customize homes, the system model was developed (Noguchi 2001; Noguchi and Friedman 2002a; Noguchi and Hernàndez 2005). To discuss the potential applicability to the delivery of quality affordable homes, this generic mass custom design system model was first introduced at an internal symposium on urbanism, which was entitled Prospectiva Urbana en Latinoamerica held in Aguascalientes, Mexico, on 31st August 2001. This televised symposium was organised by the local government and over 500 audience members from the academia, construction industry and government joined the discussion. The mass customization (MC) was visualized simply by making use of a conceptual analogue model as follows: MC = f (PS)

(1)

In this model, the service sub-system (S) concerns communication techniques that lead users to participate in customizing their new home while the product subsystem (P) covers production techniques that aim to encourage housing suppliers to standardize housing components for mass production. In mass-customizing homes, user participation, as described above, is vital; therefore, housing suppliers need to offer design support communication services to their clients. Designconsulting staff and appropriate communication tools are required to facilitate user choice of standard components (Noguchi and Friedman 2002b). These fundamental design service factors can also be integrated into a comprehensive model: S = f (l, p, t)

(2)


897

In this model, the service sub-system (S) is supported by the existence of the location (l), personnel (p) and tool (t) factors. Even though these elements are necessarily interrelated, most homebuilders and housing manufacturers have already been applying these during the design stage. An important part of mass customization is that the user directly determines the configuration of their home from choices given as client input during the design stage. However, this cannot be achieved without the standardisation of housing components for the volumetric, exterior and interior design arrangements. These components should be organized in a visually attractive way in a component selection catalogue that enables clients to easily choose from many options given—as well, the value of each component choice may need to be explicit. Basically, housing components can be divided into three categories: volume, exterior and interior. These can be considered the main elements of the product sub-system (P) which can be explained by the following conceptual model: P = f (v, e, i, o)

(3)

The volume (v) components are used to configure the space of housing that determines the number and size of each room while the interior (i) and exterior (e) components serve to co-ordinate decorative and functional elements that customize a home. In addition, "o" denotes other optional features such as air conditioning, home security system, emergency call buttons, handrails, dishwashers and other electrical appliances. Some optional features can be offered before or after occupancy with due consideration of inclusive design approaches to housing the elderly and disabled users. Both sub-systems can be considered the indispensable functions of mass-customizing homes. Generally, the unit production cost can be reduced through standardization and/or mass production of housing components. As well, the design cost associated with a number of meetings between the designer-builder and the buyer to reach the customization of a house can be reduced when the process accompanied with user participation in choosing housing component options is well standardized. It is also known that the higher the rate of in-factory completion of housing components, the more the product quality can be maintained under optimum conditions inside the factory where materials are not exposed to adverse outside climate (Hullibarger 2001). Moreover, the higher rate of in-factory completion helps control the elapsed time for the unit production and this in turn influences the labor cost. The design process, where the aforementioned mass customization system is brought into full play, can be considered mass custom design. The existing elements (i.e. parts of a whole) can be standardized while the myriad combina-

898


tions of these parts still provide great scope for creativity. Homebuyers can directly choose from the given options. The combinations of user choices from mass-producible standard housing components result in customizing houses—viz. these homes have been termed "mass custom homes" (Noguchi 2001). The mass custom design approach achieves the high level of standardization in housing components from which homebuyers participating in the design decision-making process can directly choose. The user choices of standard housing components paradoxically increase the level of design customizability (Table 1). Table 1: The levels of standardization and customization by housing design approach (Source: Noguchi and Hernández, 2005). Design Approach

Standardization Level

Customization Level

Production

High

Low

Semi-custom

Middle

Middle

Custom

Low

High

Mass Custom

High

High

The application of the mass custom design approach to reaching the customization of sequential housing units has the potential to reduce the design cost by 30-60% due to the economies of scope which can be achieved by the combinability of standard housing components (Noguchi 2004). Mass custom homes in Japan Japanese housing manufacturers have been successful in mass-customizing their industrialized homes. SANYO Homes Co. Ltd. produces their mass custom homes, winning the Good Design Award 2004 (Figure 1). Sekisui Heim (or Sekisui Chemical Co., Ltd.) brings their wood- and steel-frame modular housing system into full play, marketing over 55,000 net zero-utility-cost mass custom homes to date where a PV system is often installed as a standard feature rather than an option (Figure 2). PanaHome Corp. is successful in the solar communities consisting of mass custom homes equipped with renewable energy technologies, such as a PV system, an air-source heat pump and a combined heat and power system. In their mass housing development, every house is designed and built after the sales so that the functions of each dwelling unit built on the lot of the client’s choice are tailored to the wants and needs of the individual buyers (Figure 3). Misawa Homes Co., Ltd. is known as the first zero-energy mass custom home


899

manufacturer in Japan and their initial research on an eco-energy housing system using solar energy is dated back to 1974. In 1998, the company succeeded in commercializing the world’s first 100% self-sufficient zero-energy home. Sekisui House, Ltd. maintains the top sales of steel-frame mass custom homes in Japan, gaining the net sales of $13,112,487,000 (US $) and the net income $514,770,000 in 2007. The company is recognized as the first company in the Japanese housing industry as to the creation of a recycling system in 2005 that achieves zero emissions of waste material at the construction sites of their newly built homes.

Figure 1: Logia-Type E. (Source: SANYO Homes Co. Ltd., 2008a).

Figure 2: Sekisui Heim unit assembly line.

900


Before actually making a contract with the client, Japanese housing manufacturers are willing to give a variety of catalogues to the client, visualize the client’s image of the house in question, and estimate the total price (Noguchi and Friedman 2002b). At the marketing and design stages, the manufacturers usually provide their clients with four types of catalogues: general catalogue, commodity style catalogue, technology catalogue and design component selection catalogue.

Figure 3: PanaHome-City Seishinminami solar community. (Source: PanaHome Corp., (c2007)).

The general catalogue contains information on the company’s profiles and commodities. The commodity style catalogue provides detail information on specific housing types. The technology catalogue visualizes the value of innovations. The design component selection catalogue encourages the user choices of standard housing elements, which are inevitable in mass-customizing homes. The first three catalogues are usually provided during the marketing stage while the design component selection catalogue is used during the design consulting stage. The component selection catalogue corresponds with the housing styles selected and helps clients choose standard components for the exterior and interior arrangements of the home. The catalogue describes the material, size, color, texture and functions of each component; however, it tends not to include any prices. Additionally, Japanese manufacturers use a computer-aided design system for the creation, modification, analysis and optimization of a design. The virtual image of the house is erected based upon the housing components selected by the client. Then, the manufacturer also provides a cost estimate. Once the client is satisfied with the plans, the manufacturer will finalize the design and, at last, enter into a contract with the client.


901

An important part of the mass custom design approach is that the user directly determines the configurations from choices given as client input during the design stage. This could hardly be achieved without the standardization of housing components for the space, exterior and interior design arrangements. The concept of component standardization can be illustrated with LEGO® building blocks. A number of simple modularized blocks can be connected in a variety of ways because of their interlocking tabs and holes. Likewise, Japanese housing manufacturers offer a variety of standard housing components, which can be mass produced, to their clients and then encourage them to participate in combining the components to customize their new home. The design feedback suggested by the manufacturer extends to the considerations of future changes of family structures as well as potential renovation patterns of the house over the lifetime. The services may aim to support social and economic sustainability in housing. These design options are visually presented in the component selection catalogue so that clients can easily make their buying decisions. Mass customization is an effective means to provide optional features that help users to customize their end product to be purchased; however, there are some hidden standard features that aim to maintain or elevate the minimum level of product quality that the company expects to achieve. Environmental design features that buyers tend to opt out due to the high prices are the typical examples in today’s homes since the value of the user choices is still unclear. However, in response to societal needs for sustainable homes, green aspects should also be incorporated into the housing delivery process. The following sections introduce design and marketing strategies that help to initiate and maintain the sales of environmentally-friendly sustainable mass custom homes being built in Japan and Canada. Greening mass custom homes Japanese housing manufacturers has been implementing a marketing strategy that is aimed at supporting the commercialization of their green mass custom homes. In fact, the housing manufacturers started installing expensive renewable energy technologies and attempting to materialize the environmental sustainability by their new tectonic integration (Figure 4). The manufacturers tend to invest heavily in advertising and educate their clients to appreciate the distinguishing features of their high-cost and high-performance (i.e. high cost-performance) housing where a variety of amenities (including renewable technologies today) that drastically improve housing quality is installed as standard features rather than options. Their

902


quality-oriented production approach is rooted in the high cost-performance marketing strategy (Noguchi 2003).

Figure 4: Wa no Nagomi type-R (Source: SANYO Homes Co. Ltd., 2008b).

Japanese housing manufacturers emphasize that they have been producing betterquality homes for about the same price as conventional ones. In fact, the selling price of their high-quality industrialized houses is about 8% more expensive than that of conventionally-built moderate-quality houses (Noguchi 2003). The costperformance marketing strategy can also be seen in other industries—e.g. the automobile industry. Although today’s automobiles can be produced with lower production costs than those in the past, their selling price does not seem to be affected dramatically by higher productivity and new cars are still generally regarded as expensive. However, the list of items now offered as standard features in new cars, such as air conditioning, a stereo set, airbags, remote-control keys, power steering, power windows, and adjustable mirrors, were offered only as expensive options in older models. Clearly, the quality of newer models is much higher than that of older models. The same is true for the housing industry in Japan. Quality-oriented production contributes towards the delivery of high costperformance housing in which high-tech modern conveniences that are installed as options in conventional homes are now available as standard equipment. There is a debate about whether or not a PV system, for instance, should be installed in housing as a standard feature rather than an option. Barbose et al. (2006) indicate that the optional approach to PV sales in new homes has several distinct disadvantages from a PV deployment perspective. The most fundamental problem is a complex buying decision that homebuyers need to make for adoption of a PV system. This becomes contingent on each individual homebuyer who makes separate decision about PV amidst all of the other decisions involved in


903

buying a new home, most of which are much better understood. Additionally, sales staff must receive a much higher level of training so that they can explain the PV system and its benefits to potential homebuyers. The installation scheduling can also become more complex and prone to delays. Offering PV as an option ultimately may not be a profitable business model for large production homebuilders since the lower number of PV systems likely to be installed may increase the transaction costs (Father et al. 2004). PV systems equipped with housing reduce carbon dioxide emissions when the house comes into operation. Renewable energy technologies that alleviate negative impacts on environment can be installed as standard features while other housing design features (e.g. volume, interior and exterior components) that affect the functionality or usage of a house should remain options. Low-cost passive energy techniques that optimize the use of clean free sunlight for heating and lighting space, for instance, can be incorporated into volumetric sunspace and exterior window options. The quality-oriented design and production approaches reflect Japanese housing manufacturers' success in initiating and maintaining the mass sales of net zero-energy-cost mass custom homes today. Interactive Mass Custom Design Communication Tool In 2005, the Canadian team composed mainly of Concordia University’s engineering students participated in the Solar Decathlon housing competition, aiming to showcase their low-energy solar-powered house, called Northern Light. This PV solar house was the only Canadian entry to have competed in the Solar Decathlon, a week-long event that took place in Washington, D.C. in October, 2005. The house was featured by a 7kW PV system that covered the rooftop area of approximately 74 m². The solar modules were rated at the conversion efficiency of 13.9%. In addition to the generation of electricity, the PV system was also designed to capture heat in order to supplement the indoor space heating (Pasini 2005). The house consisted of a small-sized bed room, a living room and a kitchen/dining space and was initially designed on a speculative basis. Thus, the design customization was totally neglected before the concept of mass customization was introduced. To reduce the on-site construction time, the house was built by making use of a prefabricated modular housing system that helped lessen exposure to site nuisances such as bad weather, theft and vandalism. In order to enhance the innovativeness of architectural designs, the first author was assigned to mass-customize the house and he developed an interactive mass custom design (MCD) communication tool in collaboration with the local illustration firm. The design tool was proposed to support the initial design

904


decision-making and the future modification. It is a digitalized interactive design component selection catalogue developed with the aim of value visualization of each standard housing component option given to the design decision makers. On a computer screen, the design image can be created instantly in response to the users' direct choices of housing components. Indeed, the interactive design communication facilitates design decisions. Due to the project’s time and budgetary limitations, the standard design options were confined to the volumetric and exterior components. With consideration for the disabled wheelchair access, the volumetric design components encompassed two basic options for the entrance design: a ramp access and a porch entrance (Figure 5). Additionally, a horizontally extended skylight that affects both the interior volume and the level of natural day-lighting was introduced featured as a solar option. As for the exterior components, standard design options were given to the walls, fascias and window and door frames. Each component extends the visual variations achieved by the combinations of different materials, colors and textures (Figure 6). In this project, PV modules were installed as standard features; however, color options were given to the panels that could be integrated seamlessly into the rooftop. After the competition, the PV solar housing prototype was re-built on the university’s Loyola campus in Montreal where the engineering students can continuously examine the energy performance in the local contexts. Consumers are aware that advertisements are far from trustworthy and they continually check what they see and hear in advertising against their own experience and the experiences of others (Schiffman 1999). Experience is more reliable than other sources of information so that consumers may wish to confirm the advertisement’s claims by examining the product themselves before they buy it. The Northern Light house is now open to the general public and exhibited permanently so as to sharpen the consumers' awareness of low-energy mass custom homes (Figure 7). Building a house consumes a large amount of energy during construction and after occupancy. A considerable amount of waste materials is generated in the process of housing development. Hullibarger (2001) suggests that more waste materials are generated by on-site construction than by in-factory production which reduces or eliminates site interruption such as bad weather, site disturbance, theft and vandalism. The proposed interactive mass custom design communication tool encourages housing producers to further standardize their home-building components. The product and process standardization may create opportunities for increasing the level of in-factory production as well as reducing, reusing and recycling resources towards sustainable housing development.


905

Figure 5: Interactive mass custom design communication tool. Volumetric options: porch entrance (top) and ramp and solar access (bottom).

906


Figure 6: Interactive mass custom design communication tool. Exterior component options: stone veneer (top) and wood siding (bottom).


907

Figure 7: Canadian PV mass custom home, North Light, built by Concordia University.

Conclusions The notion of sustainable development tends to link the collective aspirations of the world’s people for improved living conditions and a healthy environment with the need to reconcile conflicting perspectives on the economy at present and in the future. The homebuilding industry is no exception and builders today are requested to deliver homes that correspond with the social, economic and environmental sustainability targets. Mass customization was considered the effective means to create options from which the users can choose. User choices are critical in how a house is designed, constructed and operated. Without the user participation, homes may never achieve the sustainability agenda. On the other hand, too many options given to users may paralyze their choices and the vagueness of the value concerning component selections may make them troubled over the buying decisions. The interactive mass custom design communication tool introduced in this study facilitates user choices and helps visualize the value. The given options can also encompass the design principles rooted in inclusive design approaches, affordable housing strategies, passive solar techniques and active renewable energy technologies—those that help achieve sustainability goals specified. Mass customization aims to customize a house at the time of sale so that it may barely affect the way to personalize space after occupancy. Accordingly, the effect of mass personalization on housing evolution after occupancy may need to be explored further. The growing patterns of space in response to the usage should be incorporated into the development of housing design options given to the users

908


before and after occupancy. Homes need to be customized and personalized so as to meet the timely needs and dynamic demands of buyers/users over the lifetime of housing. Acknowledgments The authors would like to express their sincere gratitude to Sekisui Chemical Co., Ltd., Misawa Homes Co., Ltd., PanaHome Corp., SANYO Homes Co., Ltd., and Sekisui House, Ltd. for their kind cooperation in providing precious information on their products and services in the course of this study, as well as their generous support for the Zero-energy Mass Custom Home Mission to Japan that the first author had organized over the last three years.

References Barbose, G., Wiser, R. and Bolinger, M. (2006). Supporting Photovoltaics in Market-Rate residential New Construction: A Summary of Programmatic Experience to Date and Lessons Learned. Berkeley: Lawrence Berkeley National Laboratory and Clean Energy States Alliance. Beheshti, M.R. and Dinjens, P. (1985). DPC, 85 International Design Participation Conference. Open House International. 10(1): 3–4. D'Amour, D. (1991). Sustainable Development and Housing Research Paper.1: The Origins of Sustainable Development and its Relationship to Housing and Community Planning. Ottawa: CMHC. Davis, S.M. (1987) Future Perfect, New York: Addison-Wesley. Father, B., Coburn, T. and Murphy, M. (2004). Large-Production Home Builder Experience with Zero Energy Homes. Proceedings of the 2004 ACEEE Summer Study on Energy Efficiency in Buildings. Pacific Grove, California. American Council for an Energy Efficient Economy. Friedman, A., Cammalleri, V., Nicel, J., Dufuax, F. and Green, J. (1993). Sustainable Residential Development: Planning, Designing and Construction Principles ("Greening the Grow Home). Montreal: McGill University, School of Architecture, Affordable Homes Program. Habraken, N.J. (1985). Who is Participating? Towards a new professional role. In Beheshti, M.R. ed. (1985) Design Coalition Team. Proceedings of the International Design Participation Conference. 22–24 April, 1985. Eindhoven, Netherlands. 1: 1–10. Hamdi, N. (1991). Housing without Houses: Participation, Flexibility, Enablement, New York: Van Nostrand Reinhold. Hellman, L. (1973). Housing and Participation. Built Environment. 2(6): 328–332. Hullibarger, S. (2001). Developing with Manufactured Homes. Arlington: Manufactured Housing Institute. Japanese Prefabricated Construction and Suppliers and Manufacturers Association (JPA) (2008). Prefab Club www. purekyo.or.jp/3-1.html retrieved on 13 March, 2008. Malpass, P. (1979). A Reappraisal of Byker-Part 2: Magic, myth and the architect. The Architects' Journal. 169(20): 1011–1021.


909

Noguchi, M. (2001). The Mass Custom Home: Vivienda Personalizada Masiva: Optimización de la Tecnología de la Vivienda Canadiense y Japanesa en Latinoamérica. Simposium Internacional: Prospectiva Urbana en Latinoamerica. Aguascalientes, Mexico. 31 August 2001. Noguchi, M. and Friedman, A. (2002a). Mass Custom Design System Model for the Delivery of Quality Homes-Learning from Japan’s Prefabricated Housing Industry. Proceedings of International Council for Research and Innovation in Building and Construction. CIB W060-096 Syllabus Joint Conference: Measurement and Management of Architectural Value in Performance-Based Building. Hong Kong. 6–10 May 2002, 229–243. Noguchi M. and Friedman, A. (2002b). Manufacturers-User Communication in Industrialized Housing in Japan. Open House International. 27(2): 21–29. Noguchi, M. (2003). The Effect of the Quality-oriented Production Approach on the Delivery of Prefabricated Homes in Japan. Journal of housing and the Built Environment. 18(4): 353–364 . Noguchi M. (2004). A Choice Model for Mass Customization of Lower-Cost and Higher-Performance Housing in Sustainable Development. Unpublished Ph.D. Dissertation. Montreal: McGill University, School of Architecture. Noguchi, M. and Hernández C. (2005). A „Mass Custom Design“ Approach to Upgrading Traditional Housing Development in Mexico. Journal of Habitat International. 29(2): 325–336. PanaHome Corp. (c2007). PanaHome-City Seishinminami II. Kobe: PanaHome Corp. Pasini, M. (2005). PV/T Simulation of the Canadian Solar Decathlon House. Varennes: CETC-Varennes, Natural Resources Canada. Pine II, B.J. (1993). Mass Customization: The New Frontier in Business Competition. Boston: Harvard Business School Press. Richard R.B. and Noguchi, M. (2006). Japan Solar Photovoltaic Manufactured Housing Technical Mission 2006. Varrennes: CANMET Energy technology centre, Natural Resources Canada. Sadeghpour, F., Moselhi, O. and Alkas S.T. (2006). Computer-Aided Site Layout Planning. Journal of Construction Engineering and Management. 132(2): 143–151. SANYO Homes Co., Ltd. (2008a). SANYO Products Lineup: Logia-Type <www.sanyohomes.co.jp/products/logia_e/index.html> Retrieved on 19 August, 2008.

E

SANYO Homes Co., Ltd. (2008b). SANYO Products Lineup: Wa no Wagomi Type-R. www.sanyohomes.co.jp/products/wanonagomi/index.html. Retrieved on 19 August, 2008. Schiffman, L.G. (1999). Consumer Behavior. Upper Saddler River: Prentice Hall. Smith, C. (1998). Building your Home: An Insider’s Guide. Washington, D.C.: National Association of Home Builders. Toffler, A. (1970). Future Shock. New York: Random House. World Commission on Environment and Development (WCED) (1987). Our Common Future. Oxford: Oxford University Press.

Author Biographies Dr. Masa Noguchi is Lecturer in Architectural Technology and developed the Mass Custom Design system model for the delivery of sustainable homes. His design contribution led a housing manufacturer in Canada to the commercialisation of the nation’s first

910


net zero-energy healthy housing that won the federal government’s sustainable housing completion. His Mass Custom Home research extends to the integration of inclusive affordable housing design techniques, passive energy and environmental systems and renewable energy technologies. He is the initiator and co-ordinator of the Zero-energy Mass Custom Home Mission to Japan where in total 17 industry executives, 6 government officers and 14 academics from the UK, France, Canada, China and Japan joined to date and visited the state-of-the-art production facilities of leading housing manufacturers in Japan. Moreover, he was the co-host for 39 series of Japanese community TV program entitled ARIGATO on CH Montreal, assigned to introduce the essence of Japanese business culture. Contact: www.masscustomhome.com | [email protected]. Dr. Karim Hadjri is Senior Lecturer in Architecture and the coordinator of the Design Research Unit at the School of Planning, Architecture and Civil Engineering, at the Queen’s University of Belfast, UK. He graduated as an architect in 1985 in Algeria. He was awarded a Master of Philosophy in 1989 and a Doctor of Philosophy in 1992 from the Joint Centre for Urban Design at Oxford Brookes University, England. He has worked as a scholar in the United Kingdom, the UAE and Saudi Arabia, and managed academic units and research centres in both Cyprus and Colombia. His teaching and research interests are concerned with housing studies, CAD and urban design. He has produced numerous publications and designed a number of buildings in Cyprus and UAE.

3.3


Amir E. Piroozfar (Poorang) School of Environment and Technology (SET), University of Brighton, United Kingdom Olga Popovic Larsen School of Architecture, Royal Danish Academy of Fine Arts, Denmark Hasim Altan School of Architecture, University of Sheffield, United Kingdom

Customization is well established in manufacture and service industries. By contrast, it is still relatively new in the building industry. The concept is even less known when it is to be investigated with regard to some controversial issues such as sustainability. With a special emphasis on sustainability within the context of the built environment, this chapter focuses on the areas which have been less studied or researched so far i.e. the noneconomic impact of customization on its context. Giving brief accounts of sustainability and customization, the chapter addresses a modern method of construction (MMC), in which the notion of customization can potentially be embedded, and how it can improve the sustainability agenda. Findings of the comparative study of two built projects, which have used different methods of construction, have been used to show how the benefits of a customizable MMC could contribute to sustainability in the built environment. The cases have been analyzed using Ecotect and Envest to establish their environmental performance and the results have been discussed to indicate the potential contribution of customized approaches to a more sustainable built environment. Finally, some routes are suggested to conduct further research in this field.

Introduction Mass customization is a relatively new concept in the building sector. Customization as a production method and in itself dates back to the 1960s. It superseded mass-production, is now very well established, and has been applied successfully to the computer, automotive, textile and other industries. The concepts (of customization and particularly mass customization in their specific and expert account) show a significantly shorter background in the building and construction industry. Despite some recent attempts to attribute other concepts such as flexibility or adaptability to customization in the building industry, it should not 911

912


be mistaken for any other such notions. This is because, first of all, customization (in the building industry) requires some certain levels of technological infrastructures which had not been at hand at the times it is attempted to have dated back. Secondly and equally importantly, the philosophy of customization had not yet been conceived in those days, due to the capacity of market economy, the nature of supply-demand chain, and the customers' expectations at the time. Mass customization is often understood as mass production of individually customized products and services. In its more sophisticated form this means meeting each customer’s individual wants and needs exactly, but at a price comparable to those of standard mass produced goods. The notion comes from the car and computer industries but has been accepted very swiftly across a wide range of the service and the manufacture industries from education to eyewear industry. It is still by far in need of thorough research within the building industry. Sustainability as a concept, on the other hand, has been known for a very long time. On a very broad and simplistic level sustainability is about preserving the world’s resources and minimising the negative environmental impacts. As such sustainability, it generally implies maintaining life into the long term future. This is neither a comprehensive definition, nor is it exclusive. This description, very much like the notion of sustainability, is controversial, open to interpretation and broad enough to include everything yet to imply very little. In this chapter some more exclusive and detailed definitions will be addressed which will help clarify our specific points of view more precisely. The given accounts of sustainability and mass customization have particularly been chosen in order to well serve the specific purpose of this research. They can by no means be considered as all-inclusive an account as they can be in a pure exploratory study. Aims and Objectives The main aim of this paper is to investigate the relations between customization and sustainability in the building industry (design and construction). There is a general preconception about mass customization when applied to other industries that it uses extra resources and/or puts them on a "standby" demand line to be recalled or ordered by potential customers at some stage, if and when needed. That would imply that more resources are used and in this sense it is against the strategies of a sustainable production. Although it may be true that in some cases materials are in "standby" position, one can argue that mass-customized production processes save resources because they comply with customers needs and as a result may extend the customized products' life span, usefulness, replaceability,

913


etc. The above argument is applicable to the building industry. Therefore the objective of this paper is to show that, if understood and employed well, not only mass customization is not against the goals of sustainable development, but it also can help the construction industry take a faster pace towards it. Sustainability and the Construction Industry The Brundtland Commission Report on Sustainable Development (1987) gives a very broad, disputable yet all-inclusive definition of sustainable development. It suggests sustainable development as "meeting the needs of the present without compromising the ability of future generations to meet their own needs". It was the first formal definition which implies the understanding that social, economic, and environmental issues are all inter-related. In that sense sustainability involves the above three different aspects: Environment, Society and Economy (Figure 1). A sustainable society can only thrive on a sustainable environment as its broader context and a sustainable economy as its driving power. Environment cannot reach a balanced sustainable situation where there is no sustainable economy or a sustainable society to endorse it. In a long run sustainability in economy can only be achieved where a sustainable society subsists in a sustainable environment. Environment

Sustainability Triangle

Society

Economy

Figure 1: A popular way of understanding sustainability is the triple-bottom line of environment, economy and society.

Any violation to one or more of the three will have swift damaging effects or long-term de-balancing impact not only on the other two but on the whole system. In a more practical way, a country (or society) with a strong economy should remain so without damaging the environment or dehabilitating another society or their local environment. In this sense, for instance, having planted special crops for extracting bio-diesel fuel somewhere in the South America although might

914


seem a very green or sustainable action, in fact is one of the most unsustainable acts which might have been committed because it cuts the food supply which may well affect poor African countries. In addition it also devours millions of hectares of jungles over a period of months; a green reservoir which has been created over millions of years and acts as a breathing lung for the whole Planet Earth. The UK Government has defined sustainable development as a progress which recognises (Sustainable Construction Task Group 2003) the needs of everyone; effective protection of the environment; prudent use of natural resources; and maintenance of high and stable levels of economic growth and employment. "What is a sustainable environment?" is a question to be carefully considered and contemplated on. Although the other aspects of sustainability should not be overlooked as they are major players in making balance in the sustainability triangle, the most important impacts on the environment directly are initiated by the sustainable environmental acts, policies and strategies. Coming back to the focus of this research, the built environment, it ought to be stated that sustainable environment includes sustainability in both the built and natural environment. From our specific point of view, sustainable built environment has some basic requirements which Smith and Williams define as: Carrying capacity, Thresholds, Biodiversity, Health, User-friendly, Equity, Governance (Smith et al. 1998; Williams et al. 2000). Sustainable built environment consists of sustainability in design and construction. Another important setback, which yet remains to be further investigated, is the question of what sustainable design is and what design for sustainability is? Are they different? If yes, how do they differ? Although this is not directly related to current discussion, appreciating the importance of being more precise about the two, we make some suggestions to build our approach upon. Sustainable design in general addresses the long-term social and environmental impacts of a product throughout its complete lifecycle. Foster and Partners as an architectural firm (1999) suggested that sustainable design is the "creation of buildings which are energy-efficient, healthy, comfortable, flexible in use, and designed for long life". Few years before in 1996, Building Services Research and Information Association, (BSRIA), described sustainable construction as the "creation and management of healthy buildings based upon resource efficient and ecological principles" (Edwards 2005). Janis Birkeland in "Design for Sustainability" (2005) gives an account of the impacts of construction. He indicated this impact to be on forests, water, carbon dioxide, greenhouse gases, energy, resources, and landfill:


915

Forests: The Earth’s surface has lost 50% of its forest cover (Brown et al. 1996). Buildings alone account for one quarter of the world’s wood harvest and (in the US at least) over 50% of wood is used in the building industry (Roodman and Lenssen 1995).

Water: Fresh water may be the next scarce resource in the next century. Buildings consume one sixth of fresh water supplies (Brown et al. 1996). The US Environmental Protection Authority (EPA) has identified over 700 regular pollutants in drinking water, 20 of which are known carcinogens (Zeiher 1996).

Carbon Dioxide: In the past 100 years, the level of carbon dioxide in the atmosphere has risen 27%. One quarter of this is attributable to burning fossil fuels to provide energy for existing buildings. Energy use in buildings in the UK, for example, is 48% of total CO2 emissions (Pout 1994).

Greenhouse: Building account for one third of one half of total greenhouse gases emitted by industrialised countries each year (Roodman and Lenssen 1995).

Energy: The energy used in construction alone is approximately 20% of annual energy consumption (Tucker 1994). The total energy consumed in building operation, construction and services in the UK is 66% of annual energy consumption (Vale and Vale 1994).

Resources: Buildings account for over 40% of the world annual energy and raw material consumption (Roodman and Lenssen 1995).

Landfill: Building wastes account for 44% of landfill and 50% of packaging waste in industrial nations. Brian Edwards (2004) explains sustainable materials as "materials and construction products which are healthy, durable, resource efficient and manufactured with regard to minimising environmental impact and maximizing recycling."

The Manufacture and Building Industry: Differences and Similarities There are differences and similarities between the building and the manufacture industries. They are important in terms of application of customization and to provide a more sustainable insight into the industry. The building industry is different from other industries mainly because its response to change is slow (Figure 2). It is reluctant in deployment of new ideas and approaches even if they are already established in other disciplines.

916


Figure 2: Pioneers in customization; building sector lags far behind (adopted from Kieran 2004).

The main issues in which differences and similarities can be traced between the two major disciplines – building and manufacture industries – are (Piroozfar 2005):

Philosophy, history, and background

Size of the final product

Dimensional relationship with the user

Mobility

Variation

Product life cycle

Costs and investment flow

Economies of scale

Needs and expectations

First customer vs. future owners

Order-make consequence

The fallacy of modularisation

The safety margins

Decommissioning process: dismantling vs. demolition

Unused materials (the compartments order process: top-down vs. bottom-up)

Sustainability as a New Challenge Although sustainability has shown to be a necessity for a longer occupancy of the human-beings in the planet and being acknowledged as a serious topic on the


917

growth and development agendas, it seems not to be as welcomed or easily accepted as a guideline by everyone as it should be. In brief, the extra cost, it imposes in the short term on the different business sectors is the biggest obstacle on the way of applying more sustainable building designs and construction methods. A measure project was commissioned by the European Commission called "The Future of Manufacturing in Europe 2015-2020: The Challenge of Sustainability (FutMan)" to investigate the future of manufacturing in regard with the issue of sustainability in the future (Miles et al. 2003). FutMan Scenarios were developed between June and October 2002 by 50 experts from academia, industry and policy makers. The objective of the project was to draw an imaginative picture of potential socioeconomic developments and future technologies based on the subjective views and judgements of the participant experts. The scenarios highlight important trends, possible trend-breaks, critical challenges and opportunities and present four possible visions of manufacturing in Europe in 2015-2020 (Figure 3).

Figure 3: The scenarios on the future of manufacturing in Europe 2015-2020 (adopted from Geyer et al. 2003).

The FutMan scenario exercise focused on four manufacturing sectors – electronic components; measuring, precision and control instruments; basic industrial chemicals; and motor vehicles. As a result most of the information provided in the scenarios refers to these sectors. The scenarios are structured along two qualitative dimensions of change. The first dimension relates to the modality of policy making. It includes issues such as geo-political developments, the balance between central decision-making and subsidiarity in Europe, and the rate of co-

918


ordination between different policy areas. The second dimension refers to prevailing public values, consumer behavior and demand patterns. The dimension also includes issues of public acceptance of new technology and backing of policies in support of sustainable development (Geyer et al. 2003). It is important to point out that in just one scenario out of the four (Global Economy), there is a direct account of customization: "Manufacturers focus on customization and individualization". Perhaps this scenario is the least concurrent with the idea of sustainability. This is because both dimensions of change are at their least agreement with what has been defined as a sustainable pattern for the future of manufacturing. In the Global Economy scenario, the policies are loose and the demand pattern has individual characteristics. However and despite this negative implication of customization on sustainability which has been pointed out by FutMan as a challenge in manufacture industries, one can argue that there are promising attributes of customization that if comprehended appropriately and accordingly, could stimulate a more sustainable development. Our paper investigates the implications on the built environment. Can Customization Influence Greater Level of Sustainability in Building Design and Construction? As a result of moving towards a more manufacture industry type of approach to production process, the potential fields in the building and construction industry which can be improved in terms of sustainability, if mass customization is going to be employed, can be envisaged as:

Improvement of the production process in the built environment

Increasing building life span

Work flow control

Quality control

Progressing adaptability by encouraging interchangeability

Ease of dismantling/repair/replacement of the compartments with shorter life spans or the entire building

Lowering the embodied energy in the product life cycle [including acquiring, processes, product operational period (service life), and decommissioning procedure In 2003, The Sustainable Construction Task Group (SCTG), Chaired by Sir Martin Laing provided a report, using a range of sources, in which critical points about the British construction industries have been addressed (Figure 4 shows the


919

share of the building sector in the UK carbon emissions in 2000). It asserts: "Some (experts) estimate that in the UK we are using 3 times our global share of resources…servicing of buildings produces about half of UK carbon emissions.

Figure 4: 2000 UK carbon emissions: 152.5 million tones using IPCC estimate (Sustainable Construction Task Group 2003).

The report indicates 1.5 million people, and 8% of GDP in the UK are involved in the construction industry. 6 tonnes of materials per person are used in the industry each year. 70 million tonnes of waste (of which 13 million are unused materials) are used per year and over 44,000 mega-litres of water are used by the industry every day (Sustainable Construction Task Group 2003). All these figures show that a big gap still exists within the industry to be filled with regard to sustainability. However, a need for change has been already felt and any action in this regard seems to be highly appreciated and supported. It then addresses four main themes to be explored: Investment, Design, Building Services, and Building Fabric. The major questions about the three of above themes can help current research find out what aspects of the construction industry are more likely to be improved by customization: Design, Building Services and Building Fabric. The questions about design are: Are most designers ready to consider the community as the client? What is needed to encourage the management community to become more involved in design? The two above questions seek to establish a more customer-based approach to design; the motivation underlying some very specific strategies in customization.

920


And the concerns about Building Services and Building Fabric are around:

Well insulated, air tight structure

Efficient, responsive heating and controls

Appropriate glazing and shading

Controlled ventilation

Energy efficient lighting and appliances

Passive or energy efficient cooling

Efficient water heating

User involvement during the design, remaining committed and knowledgeable during use

Passive solar design, with renewable energy incorporated as all the above are optimized. The above concerns call for a higher quality fabric (materials) and services in the buildings. To be more precise they implicitly address the need for higher quality building envelopes. In fact many of the above features can be provided by a high quality exterior cladding system. Provided the fact that moving towards customization necessitates a more modern, flexible, agile yet customer-based approach to the process of production rather than the product itself, it can provide a higher quality, longer life span and more predictability for and control over the final product. These objectives are in line with what is targeted by sustainable construction.

Edwards (2005) suggests different life spans for building components, buildings and infrastructures:

Building finishes: 10 years

Building services: 20 years

Buildings: 50+ years

Infrastructures (roads, railways): 100+ years

Cities: 500+ years The difference between life spans requires a different pace of replacement for the less steady/durable compartments of a building during its service life. Having said this, building resembles a car which needs periodical parts replacement. Customizability can potentially call for some principles which are not yet known or properly welcomed in the construction industry. These principles will automatically improve the interoperability between the compartments at same level or the


921

elements at different levels. This will both improve the real-time performance of the building and its repair, improvement or enhancement processes. The Potential Facilitators and Obstacles to Customization The established methods of construction are extremely wasteful yet very desirable, safe, stable, and unwilling to change. This basically roots in several reasons and differentiates the construction industry from any of its counterparts in the manufacture industries. First of all, it is encouraged by the traditional time lag in construction industry compared to other manufacture industries. Secondly, the nature of the business, which is dispersed virtually over millions of providers, eliminates the main driver of sensible competition from the list of the market initiatives. This problem with today’s level of both computer and internet technology is no more a serious and threatening problem. The consequential procedure of construction is the next obstacle. This dooms the building to be built on a step-by-step basis in which no real compartmentalisation is actually possible. On the other hand, lagging behind other pioneering industries in both customization and sustainability is not a minus point for the construction industry. It can help it observe the drawback in other manufacture industries and prevent same traps. In addition it will give the building industry a good opportunity to learn from its sister industries. Finally this time lag will potentially provide a thinking gap for the building sector to be able to avoid purely employing of its sister industries' experiences but to transfer and adopt them in line with its very own characteristics, nature and trends. Customization and the Modern Methods of Construction (MMC) Call for a more responsive action in the construction industry was initiated in 1994 by the Latham Report (1994) followed in 1998 by the more executive Report of Eagan’s (1998). As a result, a vast number of public and private bodies (ranging from professional to executives and from research to policy making entities) started thinking and acting more responsively. Since then so-called MMC (Modern Methods of Construction) has been initiated in different sectors of the industry. Some of these methods are nothing but a new look of the established or classical industrialised methods while the others are utterly new approaches. Despite the very good record in new and innovative MMC’s which are available or at their different stages from conception to completion, literally there is no independent approach in which mass customization has been counted for as the main incentive. Many may argue that customization cannot be taken as a sole

922


initiative in building industry but their argument very much resembles similar arguments whenever a revolutionary idea comes around and threats the safety margins of the industry. Generally speaking, a customizable approach to construction industry is similar to its ancestors i.e. systemised approaches (Including: pre-fabrication, industrialised building solutions and any approach which can be tagged as an off-site or semioff-site solution). In this sense it potentially can and should be able to benefit from all the implications of those systems hard-ware wise. Yet, it very much differs from them in terms of soft-ware and firm-ware which are available at hand nowadays comparing to 40 years ago. Having said that, lots of new possibilities have started emerging that make the customization-driven approaches too competitive to be easily neglected. Customization as a driving force will not have the very short term and visible effect on the appearance but it will start a new era of thinking and a new paradigm in the traditions of building industry. This very much needs an evolution (or more likely revolution) in the mindset and ethos within the industry. Using the level of currently available technology in combination with a strategy of customization – to tailor each individual product based on a specifically personalized order – has offered a lot in the building sector. However, it might not yet be fully possible to take each and every customer’s order and treat them as they are treated in other sectors. But having had their specific needs and preferences heard and taking them into account in DFI (Design, Fabrication, and Implementation) processes in the building and construction industry is still a huge achievement even if it is done indirectly through an independent designer’s architectural flair. The chosen exemplar as a representative of a potentially customizable approach to DFI is believed that on one hand has well benefitted from the advantages of an industrialised, mass-produced, off-site method. On the other hand, a very welldeveloped communication procedure has been established to realistically stick to the generic scheme of the architectural design. All of these aspects along with the potential flexibility of the developed system to fast respond to the call for change make this project an exemplar for further development on the theme. Analysis of Case Studies Using two different case studies, the investigation has been carried out through the performance of two buildings: one where customization strategies can be traced in the design and construction methods and the other one where traditional construction methods have been applied. The University of Sheffield’s new Student Health Centre project in Sheffield has been used as a representative of the established or


923

traditional methods of construction. While there is no direct example which can be referred to as an example of application of the customization in construction industry knowingly and purposefully as it is meant in other industries, we have chosen a project whose approaches are very close to customization and potentially can provide an archetype of this approach. This building is Mundy School in Derbyshire. The most important reason for the choice of the buildings for this study is that both of the buildings are almost of the same age. This provides more than one benefit for the study. First of all they are not significantly different from each other in terms of performance because of their age. Secondly, in the design process and construction for both of them same sets of codes and building regulations were used. Also they are located relatively close to each other. This makes some comparisons more meaningful. The other collective advantage of this choice is that the two buildings belong to the same height category, one with one story and the other in two stories. Apart from the role of the local authorities which has been eliminated from the comparison process, the two buildings both have used fairly similar materials. And the final reason for this choice is the fact that they are somewhat comparable in terms of their net area. The only disadvantage of this combination is the function of the two buildings which is different. In the study we have taken into account this difference in our analysis and have eliminated its potential downsides by comparing each of them against their own categorical benchmarks when we are inspecting the measures of sustainability. Brief introduction to the projects The first project which is used in our comparative study is a replacement school building for the existing Heanor Mundy CE (Church of England) Junior School, which was built in 2005 with a new modular-based building system called SCAPE developed by CLASP (Consortium of Local Authorities Special Program)in strategic partnership with Skanska UK. The new Mundy School was built over a period of 38 weeks in a new site near the old school. SCAPE is modular-based system with an emphasis on flexibility, sustainability, off-site fabrication, natural ventilation and light, and low energy costs (Figure 5). In terms of construction the design was considered as a pilot scheme for the consortium’s updated system, SCAPE. The external wall comprises of panels fabricated off site. The Kawneer windows utilized a new profile developed specifically to meet the needs of education. The configuration of the opening lights was carefully engineered to ensure the maximum natural ventilation could be achieved. The tartan grid of the external wall is achieved by a jointing section. Higher levels are formed with a Trespa rain screen. The roof is covered by aluminium standing seam sheet. The

924


roof is insulated and based on a roof deck spanning between the curved beams. The roof eaves are extended to provide some sun shading by means of a bracket. Great care has been taken in the design details to ensure that the insulation is continuous with cold bridges eliminated and to ensure that air leakage is below that required by the Building Regulations. Internally the walls are constructed of two layers of plasterboard to both sides of the partition. An analysis of the buildings carbon emissions calculated using a method devised by the DfES (Department for Education and Skills) showed them to be 3.55kg C/m2/year where the measure for excellent was 5kg C/m2/year. The second project is the new Student Health Centre for the University of Sheffield, which was designed to provide healthcare services for the university students. The building is selected as an example of the traditional/established methods of construction with an emphasis on green design features, such as natural light and energy efficient natural ventilation system as well as the use of environmentally friendly components and finishes (Figure 5).

Figure 5: Mundy Junior School and University of Sheffield Health Centre. Photo of Mundy School courtesy of CLASP (Photo of Health Centre by A.E. Piroozfar).

The project won two prizes in 2005 including the top prize in the education/medical category of the RIBA Yorkshire Awards and a special prize from Ibstock, for the building’s brickwork. The RIBA Yorkshire Award was given in recognition of the building’s excellent design which incorporates glass blocks and floor-to-ceiling windows for natural lighting, a reception area with oak detailing, and an energy efficient natural ventilation system with solar shading. The building is built traditionally with external walls made out of a non-British sourced brick in combination with timber. The projected flat roof in combination with horizontal bris-soleil provides high levels of shading. Broad windows provide the street looking façades of the building with natural light while a double height floor to


925

ceiling window opens the main waiting area to a small garden outside. Emphasis has been put on elimination of artificial lighting and ventilation. The construction process As mentioned earlier the University of Sheffield’s new Student Health Centre uses traditional/established methods of construction. There is no need, therefore, to describe in any further detail the construction process of this building. Mundy School, by contrast, uses an innovative method of construction. The project was aimed to meet the standards and requirements set for a benchmark school project for the 21st Century by DfES (Department for Education and Skills): "The Mundy Junior School is a pilot project between the Consortium and Derbyshire County Council which brings together new technology to create a high quality teaching environment. Derbyshire County Council is the lead designer supported by the Consortium’s Concept Architect. Skanska are the contractor… To embrace sustainability, the design includes features for recovery of grey water and off site construction techniques" (CLASP 2006). The project has a straight-forward modular plan which facilitates its off-site production strategies and future extension and/or alteration in the project if required. The factory-made system, SCAPE, is developed to fulfil the specific needs and requirements of school environments. Relatively shallow in-situ concrete foundations were used due to need to support the single-story building above (Figure 6).

Figure 6: In-situ foundation (left) and the guide concrete elements used as pedestals for the exterior wall panels (right) (Photos courtesy of CLASP).

In the further construction stages of the building more off-site techniques were being employed. A pre-made support structure (Figure 7) has been used to provide a framework to support the wall panels and has very little role in load transfer. They mainly contribute in transfer of the lateral forces in coordination with a

926


bracing system. It forms the steel skeleton made of cold-formed steel sections with bolted joints (Figure 7). A curved Warren Terrace forms the main structure of the roof. These are made off-site, brought to site and assembled.

Figure 7: Cold-formed skeletal structure to support the external walls (left) connection details of a bedding for roof terrace at a vertical support (right) (Photos courtesy of CLASP).

Figure 8: Fabricating the wall panels on jigs (left) a finished panel waiting to be transported to the site (middle) and layer detailing of an external wall panel (right) (Photos courtesy of CLASP).

Figure 9: Wall panels assembly on site (left) finished roof of aluminium standing seam sheet (right) (Photos courtesy of CLASP).


927

The main exterior walls, which also form a major construction stage, are made of off-site pre-made panels with the initial size of 3.6 m x 3.6 m (Figure 8). They are made in controlled conditions of a factory on jigs with a range of variation on theme based on the requirements of architectural scheme. The panels use a coldframed gridal backing. This support structure provides a relatively open platform to adopt different variations of the outer façade. It then accommodates a layer of insulation, a concrete thin backing and the outer layer of terracotta. Lifting the panels however has necessitated some extra support on the full-sized panels. "Technology has been developed to increase the prefabricated elements in the external wall. A clay block faced panel developed by FIS Systems Limited has been used together with a unitised window from Kawneer. Special attention has been given to the routing of services and the layout contains movable walls all to make the spaces adaptable" (CLASP 2006). The panels are then transported to the site, lifted by a crawler crane and assembled in place (Figure 9). In this sense the construction process is very much like manufacturing a car in a higher scale production line, in the open space and for just one product. Further alterations of one or more than one of the envelope assemblies are easily achieved at any time during the construction or even after the completion of the building. The roof is made on site and it consists of corrugated aluminium sheets insulated from inside with special attention given to the elimination of air-leak and cold bridges (Figure 9). Our initial assumption is that the weakness of the building in terms of high Operational Environmental Impact, which we will explain later, is due to the clash between the prefabricated walls and established methods of constructing the roof specifically at joints. More detailed study, however, is needed to confirm or dismiss this hypothesis. The approach In order to establish a measurement of sustainability we have decided to evaluate the two projects using building energy and environmental evaluation tools. These tools allow designers to assess both energy and environmental impacts of a building. In this study Ecotect and Envest have been used to analyse the buildings, because they are relatively simple and are widely used in the UK to carry out both energy performance and environmental impact analyses. Also they use an open platform structure which allows the user to evaluate different buildings in regard to the available data. The aim of this analysis is to predict the likely energy performance and environmental impact of the buildings with respect to the employed construction methods. Comparisons of the outputs against each other as well as to the energy benchmark

928


of typical and same purpose built buildings have been carried out afterwards. In both case studies, the building energy use was based on the heating and cooling loads in the building with assumptions of operational schedules and the basic environmental and cost impact of various strategies for heating, cooling and operating buildings were predicted. The advantages of using such tools have allowed us to easily compare the two projects selected as the case studies with different construction methods (CIBSE 1998). Energy performance analysis One of the studied aspects about the two buildings is how they perform in terms of energy use. This forms a major contribution in how environmental friendly a building behaves during its service life. It is more than evident that the less the energy a building uses, the more sustainable it can be assumed to be. The tool: Ecotect – opportunities and limitations Ecotect is a complete energy and environmental analysis tool. It is only used for energy performance analysis in the select case studies (Ecotect). Energy performance within Ecotect uses the Chartered Institution of Building Services Engineers (CIBSE) Admittance Method to calculate heating and cooling loads predicting the likely energy loads for the buildings. In both case studies, 3D modeling has been provided and detailed material properties were assigned to the constructions and operational schedules of occupancy, internal gains, and filtration have been assumed based on commonsense. In Ecotect, the thermal simulation engine provides a range of thermal performance analysis options and the CIBSE Admittance Method is used to determine internal temperatures and heat loads (Ecotect 2008). This thermal algorithm is very flexible and has no restrictions on building geometry or the number of thermal zones that can be simultaneously analyzed. As with any calculation method, it is necessary to strike a balance between accuracy and simplicity. The Admittance Method is widely used around the world and has been shown to be an extremely useful design tool. It is not as physically accurate as some of the more computationally intensive techniques such as the response factor or finite difference methods; however for the purposes of design decision-making, the Admittance Method is by far the best choice. Ecotect analysis Figures 10 and 11 indicate predicted energy loads on a monthly basis for the Student Health Centre and Mundy School. The results show that both buildings


929

have a low energy requirement which will be also compared against the energy benchmarks set by the CIBSE Guide F (2004).

Figure 10: Health Centre monthly energy performance.

Figure 11: Mundy School monthly energy performance.

930

HANDBOOK OF RESEARCH IN MASS CUSTOMIZATION & PERSONALIZATION Table 1: Monthly heating/cooling loads.

Month

Health Centre

Mundy School

Heating (kWh)

Cooling (kWh)

Heating (kWh)

Cooling (kWh)

Jan

43050.17

0

21002.35

0

Feb

32094.08

0

15242.61

0

Mar

34206.21

0

16348.42

0

Apr

17484.29

0

7862.565

0

May

17287.6

0

7755.471

0

Jun

10913.78

0

4624.892

0

Jul

4396.771

0

1140.225

0

Aug

5606.814

0

1851.69

0

Sep

9316.814

0

3511.131

0

Oct

17459.82

0

7764.804

0

Nov

24966.15

0

11820.45

0

Dec

30178.29

0

14669.22

0

Total

246960.8

0

113593.8

0

Per m²

161.653

0

97.388

0

Table 1 shows predicted monthly heating/cooling loads in kWh with max heating of 116.863 kWh and no cooling for the Health Centre and max heating of 75.009 kWh and no cooling for Mundy School. The next table (Table 2) displays the energy benchmarks for both building types based on typical and same purpose built buildings, and the results show that both buildings are performing better than good practice benchmarks. However, it should be born in mind that the results are determined based on a certain period occupancy and an air change rate, and therefore the values in real time are more likely to show higher values, which would still be in the range between typical and good practices. In the case studies both building fabrics were assumed to have the minimum uvalues required by the 2002 building regulations as indicated in Table 3, which was at the time in affect when the buildings were built (ADL2 2002). This will make the comparisons between other aspects of assessment more convenient and

931


accurate; aspects such as examining the environmental impact and the energy performance based on methods of construction used. Table 2: Comparison against the energy benchmarks.

Building

Predicted loads (kWh/m2)

Benchmark for typical practice (kWh/m2)

Benchmark for good practice (kWh/m2)

Health Centre

162

270

174

Mundy School

98

164

113

Table 3: Minimum u-values (ADLA 2002). Building element

u-values (W/m2K)

Walls

0.35

Windows

2.2

Floor

0.25

Roof

0.25

Environmental impact analysis Another studied aspect reflects the buildings' environmental impact. This is usually personified by how a combination of embodied and operational energy stands against the standards. Again, the higher the environmental impact, the less the building contributes in a making a sustainable environment. The tool: Envest – opportunities and limitations Envest is a life cycle analysis tool that measures embodied and operational energy, and the environmental impact of design strategies (Envest 2008). In Envest, very complex processes of environmental impact of designs are simplified by allowing both environmental and financial tradeoffs to be made explicit in the design process. It provides the client with optimization of the concept for the best value according to their own priorities. Envest identifies building elements with the greatest influence on the building’s environmental impact as well as the whole life cost by showing the effects of

932


selecting different materials. The environmental impacts of construction encompass a wide range of issues, including climate change, mineral extraction, ozone depletion and waste generation. In Envest, assessing such different issues in combination requires subjective judgements about their relative importance. Therefore, Building Research Establishment (BRE) has developed Ecopoints, which are normalised environmental impacts in order to enable such assessments to be made on a single score scale (Envest 2008, BRE 1999). Measuring each environmental issue using its own unit (e.g. measuring mineral extraction with using tonnes of mineral extracted and climate change in mass of Carbon Dioxide equivalent) can be hard for making useful comparisons; however, by comparing each environmental impact to a norm, each impact can be measured on the same scale. Envest was developed with a methodology based on normalised data (the impacts of a typical UK citizen) for the UK and therefore this can be seen as a limitation for assessment of buildings outside the UK. Envest analysis One of the issues facing sustainable construction is to ensure that the embodied and operation energy usage is kept to the lowest possible amount. In some ways it is difficult to accurately assess this issue because it very much depends on the exact type and specification of the materials and how often they are replaced; however using Envest, we were able to carry out predictions for both of the projects. The environmental impacts are determined by the input building design data e.g. height, window area; the detailing specifications e.g. materials, construction components i.e. external walls, roof covering; mechanical features e.g. services and their operation. The Impacts include Environmental Impact of Construction, Life Cycle Assessment (LCA) as well as Climate Change, Acid Deposition, Water Extraction and Water Pollution. The environmental impact takes into consideration everything that occurs during or is required in the course of construction from material extraction, processing, component assembly, transport and construction, to maintenance and disposal of construction products that have an environmental impact over their entire life cycle. Life cycle assessment is a method to measure and evaluate the environmental impacts associated with a product system or activity by describing and assessing the energy and materials used and released to the environment over the life cycle of a building, and are also carried out in the case studies. The summary of the results of the environmental impact analyses carried out for the two building including: Embodied and Operational Energies (Figure 12), Embodied Environmental Breakdown (Figure 13), and Operational Environmental Breakdown (Figure 14).


Figure 12: Embodied vs. operational energy: Comparison.

Figure 13: Embodied environmental breakdown: Comparison.

Figure 14: Operational environmental breakdown: Comparison.

933

934


Table 4 indicates the impact of the embodied and operational energies of the two buildings. Normalised environmental impacts of the two buildings using the BRE’s Ecopoints are compared in Figure 15 and Table 5. Table 4: Embodied and operational environmental impacts: Health Centre vs. Mundy School. Embodied Environmental Impact

Impacts on …

Operational Environmental Impacts

Health Centre

Mundy School

Health Centre

Mundy School

726

594

2324

3619

Acid Deposition (tonnes SO2 eq.)

6

4

20

33

Ozone Depletion (kg CFC11 eq.)

0

0

0

0

Human Toxicity Air (kg tox.)

4055

3780

23646

38654

Ozone Creation (kg ethene eq.)

548

548

146

224

Human Toxicity Water (kg tox.)

0

0

0

0

Eco Toxicity Water (m3 tox.)

5055614

4324991

0

0

Eutrophication (kg PO4 eq.)

220

211

771

1224

Fossil Fuel Depletion (tonnes of oil eq.)

206

176

729

1102

Minerals Extraction (tonnes)

1231

1218

0

0

Water Extraction (m3)

4500

5990

31845

19702

Waste Disposal (tonnes)

735

769

0

0

Climate Change (tonnes CO2 eq. (100yr))

Implications and interpretation of results Without an understanding of how the occupants are actually using any of the two buildings, particularly with respect to heating controls or ventilation strategies (i.e. window opening patterns and/or air conditioning), making exact assumptions regarding the amount of energy used for heating or cooling is a bit difficult. The computer simulations are determining the most likely energy loads and environmental impacts based on the input data for the case studies and they may be slightly underestimated by the calculations, however, without carrying out a post occupancy monitoring program in these buildings, it is not possible to check how reliable these results are. Therefore the results in this case studies should be taken

935


as an indicative summary and not a representation of reality for the two chosen projects. As the buildings are built to the 2002 Building Regulations standards, the two selected projects are shown with minimum u-values in regards to building fabric and air tightness values. This gave the heating loads approaching to 162 kWh in Healthcare building and 98 kWh in School building per square metre which are relatively optimistic for the given assumptions for operation schedules.

Figure 15: Ecopoints environmental breakdown: Comparison. Table 5: Ecopoints environmental impacts: Health Centre vs. Mundy School. Impacts on

Health Centre

Mundy School

3050

4213

Acid Deposition (tonnes SO2 eq.)

26

37

Ozone Depletion (kg CFC11 eq.)

0

0

27700

42434

Ozone Creation (kg ethene eq.)

694

773

Human Toxicity Water (kg tox.)

0

0

Eco Toxicity Water (m3 tox.)

5055614

4324991

Eutrophication (kg PO4 eq.)

991

1435

Fossil Fuel Depletion (tonnes of oil eq.)

935

1278

Minerals Extraction (tonnes)

1231

1218

Water Extraction (m3)

36345

25692

735

769

tonnes CO2 eq. (100yr)

Human Toxicity Air (kg tox.)

Waste Disposal (tonnes)

936


Experience indicates that air change rate values between 0.5 and 1 are more likely to be the case for a real building and therefore 0.5 air change per hour is the value used as the maximum attainable in the computer simulations (CIBSE 2004). It must be noted that increasing the air change rate to over 1 will have a very significant effect on the heating loads which may result in them going beyond the energy benchmarks set for a typical building built with the same purpose. The energy performance results were compared against the energy benchmarks and this indicated that the chosen case studies are performing in good practice standards. As a result, this study can suggest that the design strategies and construction methods used in both projects with respect to the use of natural light and natural ventilation had a positive impact on buildings' energy use by reducing overall energy loads. However, in a more careful consideration the energy performance of Mundy School is 13% better than the benchmark for the good practice where the energy consumption of the Health Centre is just under 7% better than the benchmark for good practice. Both energy performances are almost 40% better than what it should be in benchmark projects for typical practice. In terms of environmental impact and life cycle analyses both projects have shown similar embodied and operational environmental impacts. In comparison, the school building had shown lower impacts on the environment with respect to impacts including environmental impacts of construction, the life cycle impact, climate change, etc. As a rough conclusion and taken into account that the Health Centre was designed having the environmental concerns on the design agenda where in case of the Mundy School the only concern has been to develop a specific window frame for the schools, the novel method of construction employed in Mundy school, although with a marginal difference shows to be slightly better than the Health Centre. However, the fact that there is still scope for great improvements if a fully customized approach to the DFI processes in design and construction is applied, this study shows that, customization can potentially help achieve more sustainable construction and consequently a more sustainable environment. Conclusions and Further Research The study presented in this paper provides some evidence to support the fact that if a more customizable approach to design and construction is taken, one can expect realistic environmental benefits and lower impacts. The research investigated some of these aspects and respectively addressed the areas which can be improved further. Bearing in mind that customization in design and construction


937

in the building industry is not fully customer-driven; this research has led to conclude that there are great opportunities lying ahead. Namely, if customization is to be applied as method more broadly it would be realistic to target different aspects of sustainability. If customization is applied to design and fabrication it would affect building performance, decrease the maintenance overheads, most likely reduce the human error, eliminate waste to some extent, ease dismantling/destruction and make reuse/recyclability easier, all of which would increase the life cycle of the building. Although this seems straight forward as a strategy, the authors of this paper are fully aware that for this to happen and become everyday practice not only is it needed to benefit from a very strong ICT infrastructure but also to have a new mindset in the building industry; the mindset which very much resembles that of manufacture industries“. It is trusted that a customizable approach to the DFI in the building and construction industry can only be initiated if a comprehensive understanding of the Zeitgeist (the spirit of time) and its changing factors is appreciated. Changes and actions must happen broadly and across the whole building sector. As stated earlier the research showed that there are some benefits in the application of customization strategies in favour of a more sustainable environment. However, more research and careful monitoring are still needed to be carried out to investigate all potential benefits particularly those which are embedded in the construction process. Here are some suggestions for further work in this area:

A comparative POM (Post Occupancy Monitoring) on the two projects to help this study to achieve more accurate results over a period of their service life.

An in-depth study on what would be more sustainable choice: renovating an existing and non-efficient building or knocking it down and building a new one applying tight sustainability measures.

An in-depth study of the scale of customized building production needed to break even and start generating profit because of the increased design time.

A more specific detailed study on the construction methods independent of the building performance to investigate how the potential higher costs of a customization driven MMCs would be reconciled by the potential savings in the POM period of the building also looking at the potential for its deconstruction, reuse and recycling.

References ADL2 (2002). The Building Regulations 2000. Birkeland, J. (2005). Design for Sustainability. London, Earthscan.

938


BRE (1999). Ecopoints: A Single Score Environmental Assessment. Brown, L. R., C. Flavin, et al. (1996). Vital Signs 1996. New York, W.W. Norton and Co. Brundtland, G. H. (1987). Report of the World Commission on Environment and Development: Our Common Future Geneva, Switzerland UN General Assembly. CIBSE (1998). Building Energy and Environmental Modeling. CIBSE (2004). Guide F: Energy Efficiency in Buildings. CLASP (2006). Mundy Junior School, Heanor. Derbyshire County Council. (Accessed on 1 April 2008) www.derbyshire.gov.uk. Ecotect Software (Accessed on 26 March 2008) www.ecotect.com. Williams, K. and E. Burton, et al. (2000). Achieving Sustainable Urban Form. London: Spon Press. Edwards, B. (2004). Sustainability and Education in the Built Environment. The Sustainability Curriculum: The Challenge for Higher Education. J. Blewett and C. Cullingford. London, Earthscan. Edwards, B. (2005). Rough Guide to Sustainability. London, RIBA Enterprises Ltd. Egan, S. J. (1998). Rethinking Construction. London, the Construction Task Force. Envest Software: (Accessed on 26 March 2008) envestv2.bre.co.uk. Foster and Partners, N. (1999). (Accessed on 16 April, 2008), scom.hud.ac.ukscomjm4/ mmport/sysmod/page7.htm. Geyer, A., F. Scapolo, et al. (2003). The Future of Manufacturing in Europe 2015-2020; The Challenge for Sustainability:Scenario Report (Summary), Institute for Prospective Technological Studies, European Commission (Joint Research Centre). Kieran, S. (2004). Refabricating Architecture / Stephen Kieran, James Timberlake. London; New York, McGraw-Hill. Latham, S. M. (1994). Constructing the team: final report: joint review of procurement and contractual arrangements in the United Kingdom construction industry / by Sir Michael Latham. London, H.M.S.O. Miles, I. P., M. A. Weber et al. (2003). The Future of Manufacturing in Europe 2015-2020; The Challenge for Sustainability. Piroozfar, A. E. (2005). Mass customization: applicability and complications in building industry. 3rd Interdisciplinary World Congress on Mass Customization and Personalization MCPC2005, Hong Kong, China. Pout, C. (1994). Relating CO2 Emissions to End-Uses of Energyin the UK. 1st International Conference on Buildings and Environment CIB and BRE, Watford. Roodman, D. M. and N. Lenssen (1995). A Building Revolution: How Ecology and Health Concerns Are Transforming Construction. Worldwatch Paper. Washington, DC. Smith, M., J. Whitelegg et al. (1998). Greening the Built Environment. London, Earthscan. Sustainable Construction Task Group (2003). Better Building Summit Issues Paper, ODPM, DTI, DEFRA. The University of Sheffield. "(Accessed on 1 April 2008) www.shef.ac.uk." Tucker, S. N. (1994). Energy Embodied in Construction and Refurbishment of Buildings. 1st International Conference on Buildings and Environment CIB and BRE, Watford. Vale, R. and B. Vale (1994). Towards a Green Architecture. London, RIBA Publications. Zeiher, L. C. (1996). The Ecology of Architecture. New York, Whitney Library of Design.


939

Author Biographies Dr. Amir E. Piroozfar (Poorang) has recently joined the School of Environment and Technology (SET), University of Brighton as a senior lecturer and researcher. Prior to taking up the post in 2009, he has been researching, teaching and practicing both as an architect and an urban designer in different countries for nearly 15 years. He finished his PhD on "Application of Mass customization on Design, Fabrication and Implementation of Building Envelopes" in the School of Architecture, University of Sheffield in 2008. His research interests include Mass customization in Building Industry, Digital Fabrication and Digital Craftsmanship, Building Envelopes (Theory and Practice), BIM applications, VR (Virtual Reality) and AR (Augmented Reality), Urban Façades (Theory and Practice), Urban Morphology and Urban Evolution, Sustainability with special reference to the Third World countries (Sustainable Environment, Sustainable Development, Sustainable Design), and Fuzzy Logic (the application on qualitative data assessment in built environment). Contact:[email protected] Prof. Dr. Olga Popovic Larsen leads the Structures in Architecture research and education group at the Royal Danish Academy of Fine Arts School of Architecture in Copenhagen. Before coming to Denmark for 13 years she was at The University of Sheffield School of Architecture in UK where as an Associate Professor she was researching and lecturing. Her research is into fields that bridge Architecture and Structural Engineering, such as Structural Morphology and Advanced Structural Systems including Reciprocal Frames and Tensegrities. Special interests include Sustainable aspects of Mass customization and Design for Disassembly and Adaptability. Contact: [email protected] Dr. Hasim Altan is a Lecturer in the School of Architecture and Director of the Building Energy Analysis Unit (BEAU) Research Centre in the School. He has been promoting Energy Efficiency in Buildings and providing Environmental Design and Sustainability advice and resources to the building industry in related areas as part of Knowledge Transfer Partnership. Dr. Altan’s previous research focused on the drivers and barriers to improving energy efficiency and reducing carbon dioxide emissions in the private housing sector, where he investigated energy efficiency standards of 250 homes in Sheffield including energy performance ratings and carbon dioxide emissions. He has worked extensively on energy efficiency strategies for buildings and is also part of CaRB project funded by the EPSRC and Carbon Trust aimed at reducing carbon emissions from the UK building stock. His objective, as an architect-lecturer, is to contribute to the growth of knowledge in Sustainable Architecture, Energy-Efficient Building Design. Contact: [email protected]


4

Applications of MCP in Various Contexts

941


4.1

The State of the Art of Mass Customization Practices in Finnish Technology Industries: Results from a Multiple-Case Study Marko Mäkipää Department of Computer Sciences, University of Tampere, Finland Lea Ahoniemi Department of Computer Sciences, University of Tampere, Finland

Markus Mertanen Institute of Industrial Management at Tampere, University of Technology, Finland Matti Sievänen Institute of Industrial Management at Tampere, University of Technology, Finland Linnea Peltonen Department of Computer Sciences, University of Tampere, Finland Mikko Ruohonen Department of Computer Sciences, University of Tampere, Finland

The mass customization literature has increased a great deal in recent years and today covers a wide range of topics from customer interaction strategies to product development principles. However, the extant literature is biased; it seems to concentrate more on business-to-consumer commerce and theoretical research approaches. This paper seeks to tackle this research gap by presenting the results of a business-to-business multiple-case study conducted in 37 Finnish companies. According to the results, the concept of mass customization was not always identified by companies, but mass customization strategies and practices are widely used in Finnish technology industries. In the current state, the product qualities and production processes are in general managed well, but there are numerous challenges, especially in cross-functional cooperation, sales configurator deployment and the integration of different information systems. Derived from case interviews and insights gained, a comprehensive model of mass customization (McMountain) is suggested to assist companies to better succeed in their mass customization development projects.

943

944


Mass Customization in the Business-To-Business Context Mass customization was envisioned as a manufacturing method of the future as early as 1970, and this phenomenon was named in 1987. However, only after Pine wrote his best-selling book on this subject in 1993 did mass customization strategies begin to penetrate corporate business strategies on a large scale. One of the enablers behind this evolution was the strong development in the area of information and communications technologies that to a great extent enabled the cost-effective adoption of various customer-focused mass customization strategies. It is even considered that the information revolution has actually created the heterogeneity of demand and ever smaller customer segments. In business-tobusiness (B2B) markets, the current management trends such as specialization of companies and concentration on core competencies have increased the need for specialized tools and machines, increasing the need for the customization of capital goods. The importance of mass customization has even been characterized as a revolutionary force, which is similar to the movement from craft-type production strategies to mass production manufacturing at the dawn of the 1900s. Whether true or not, the significance of mass customization should not be underestimated, since as an undergoing change it will cause increasingly large impacts on the competitive abilities of most industrial areas. This development provides also new opportunities for countries with high production costs as customer-tailored products create added value. In the B2B setting, the primary factor for purchasing decisions is not merely the price but also the delivery time, unique product qualities and MRO services. By using mass customization strategies, added value can be created both cost-effectively and with a rapid delivery time. In addition, the adoption of mass customization strategies often requires a high-technology-based manufacturing environment. This gives a certain advantage to advanced industrial settings such as Finnish technology industries. This paper attempts to tackle the two main shortcomings in the extant mass customization literature, i.e., too few empirical studies and lack of business-tobusiness focus. A multiple case study of 37 companies was conducted as a part of a three-year research and development project, concentrating on ICT-enabled mass customization opportunities in Finnish technology industries. Finnish technology industries are characterized by short production series, extensive product customization and advanced production technologies. The industries are also characterized by high labor and production costs, small internal market size and remote distance from large market areas. These features make Finnish technology industries a challenging operating environment where constant


945

development of new value (creation and capture) systems is needed. Therefore, it is also an interesting context of research for B2B mass customization possibilities and its potential in highly developed industrial countries. In this study, the mass customization concept was understood in a specific meaning: customers are involved in specifying the product configuration, and the "mass" in mass customization is obtained by standardized components and product modularity. The rest of the paper is divided into sections as follows. In Section 2, the research setting is described. In Section 3, the overall state and some of the most interesting characteristics of mass customization practices in Finnish technology industries are described. In Section 4, key development areas are described, and a comprehensive mass customization strategy is suggested, as identified from company interviews. Section 5 concludes the paper with discussion, conclusions and future research recommendations. Research Setting Research context The multiple-case study of Finnish technology industries reported in this paper was conducted as a part of a both publicly and privately funded research and development project. The research project concentrated especially on mass customization opportunities utilizing information and communication technologies in Finnish technology industries. This paper describes the general results from the first phase of the research project that included a multiple-case study of 37 companies in order to outline the state of mass customization practices in Finnish technology industries. The Finnish technology industry is an extremely important business research context from the national point of view. Technology industries and their competitive ability have significant impact on the entire national economy of Finland, both directly and indirectly. Technology industries are not only Finland’s largest industrial sector, accounting for 59% of export income, but they also count for no less than 84% of all research and development investment in Finland. The objective of the study was to outline the main development paths of Finnish technology industries, focusing in to chart the state of mass customization practices in each company. The purpose of this interview study was also to use collected information in identifying general development challenges, problem areas and development potential. An additional aim was to create a roadmap for the development of mass customization strategies in technology industries.

946


Research method The multiple-case study method was selected because it has many advantages in this kind of research context. One of the reasons was that mass customization is still an emerging concept unfamiliar to many companies. A survey as a research method would have produced very distinct results difficult to interpret; in this kind of explorative research setting, only a case study method can produce sensible results. This is because in the interview situation a researcher can interact with interviewees about the subject area and a common understanding of the concepts used can be achieved. The multiple-case study method was required to be able to extend our knowledge among individual cases and to create an outline of the situation in technology industries. Furthermore, results from a multiple-case study are considered to be more convincing and theoretically more generalizable than the ones from a single-case study. Case companies were selected by using theoretical sampling. According to Eisenhardt, theoretical sampling can be used when a researcher wants to replicate a study, expand created theory or fill theoretical categories or give examples of polarized types. The companies were selected to fill different categories of product types and production volumes. For each selected company, a counterpart with similar product/volume features was selected to find out possible differences and similarities in mass customization practices. The empirical data for this study were collected from 37 Finnish companies. The target number of interviewed companies was set to 30-40 at the beginning of the study, and the saturation level where no new findings were found from new interviews was reached after a little over 30 cases. The rest of the scheduled interviews were conducted, and a total of 37 case companies were interviewed. The study is composed of company visits, related documents and in-depth interviews of 63 directors and managers altogether. Company interviews were conducted between October 2006 and April 2007. Company representatives were typically production managers, product development managers, product managers or sales managers; in most cases, more than one of the before-mentioned managers took part in the same interview. The interviewees were selected by the companies internally after phone or email contact by researchers. Both large enterprises and SMEs were included in the study (Table 1) as well as both brand owners and sub-contractors. The volume of production varied from 10 units to 8,000,000 units per year. Some companies applied built-to-order production while some manufactured finished or semi-finished products to stock. Also, various industries were present, although the machine construction industry was somewhat overweighted (Table 2).

947

VOLUME 2: APPLICATIONS AND CASES Table 1: Size distribution of interviewed companies. Size

Personnel

Turnover

Number of companies

Percentage

Small

10 - 49

2 - 14 M€

5

13,5%

Medium

50 - 249

13 - 150 M€

13

35,1%

250 +

55 - 1 000+ M€

19

51,4%

av. 417

av. 145,8 M€

tot. 37

100%

Large Average / total

Table 2: Represented industries. Industry

Number of companies

Percentage

Machine products

15

40,5%

Electrotechnical products

5

13,5%

Metal products / metal refinement

4

10,8%

Plastic and rubber products

4

10,8%

Machine product designing

4

10,8%

Wood products

3

8,1%

Other manufacturing

2

5,4%

Total

37

100,0%

The process of conducting interviews After initial contact and agreement to an interview were made, an outline of topics was sent via email for the interviewees for their mental preparation. The outline of topics was carefully planned by the research team to cover the most important topics of mass customization. The objective was to be able to extrapolate a blueprint of mass customization practices and variations of practices in different industrial environments based on the collected data. In the actual interview session, the outline was not followed slavishly; the aim was to let the discussion roll on its own to get the richest description possible and to write down companyspecific development stories. Therefore, researchers guided the discussions only in order to cover all the topics listed in the interview outline. The topics listed and sent beforehand to interviewees included structured, semi-structured and openended type of questions.

948


In practice, all of the interviews were conducted during company visits. Most of the interviewees were from middle management, although members of top management and product/process owners were also present in some interviews. Typically, 1 or 2 managers were present in the interview situation per company. Open yet confidential discussions were possible because among the interviewees existed a strong trust for researcher not to misuse the interview material, which was attributable to the good reputation of Finnish research institutions among companies. In all interviews, a minimum of two researchers were present to ensure the validity of the interpretations. In addition, a company-specific interview report was sent to each interviewee to double-check for misinterpretations and correction of clear mistakes. Also, total anonymity was guaranteed for companies; only aggregate results would be published straightforwardly. Table 3: Nature of operations of interviewed companies. Role of operations

Number of companies

Percentage

Design

4

10,8%

Manufacturing

33

89,2%

Component manufacturing

6

16,2%

Assembly

7

18,9%

Versatile manufacturing

20

54,1%

Total

37

100,0%

The State of Mass Customization Practices A characteristic of the interviewed companies was their clear orientation in customer-specific production (26 out of 37 classified). The custom-made orientation of the study population was not known beforehand when the research group selected the companies asked to join the study. Neither is it known how well the study population represents the total population of Finnish business-tobusiness manufacturers. However, it must be acknowledged that the practices of custom-made orientation varied. Most of the companies analyzed in this paper utilized at least some mass customization principles, but some could better be characterized as engineering-to-order manufacturers. The unit price was reversely dependent on the scale of manufacturing volume, although some variance appeared. Also, most of the studied companies were operating their own brand and controlled all phases of their total manufacturing process. Only a minority of


949

the interviewed companies classified themselves as component suppliers. Four companies had their core business in design operations (Table 3). Perceived changes in business environment There are some common underlying assumptions concerning changes in the business environment in the mass customization research literature. These assumptions, such as increasing demand for customized goods, continuously changing customer needs and shorter product life-cycles, are frequently repeated in mass customization articles and even supported by empirical studies. During the first company interviews, these same issues started to emerge as factors influencing the implementation of mass customization strategies. After a number of interviews, the research team decided to concentrate more on this particular issue and added some structured questions about business environment changes to interview outline. The formulation of these structured questions followed one earlier comparable study by Åhlström and Westbrook. According to the collected data, most of the companies agreed with common perceptions concerning changes in the business environment. The power of customers and their changing needs were considered major drivers in the business environment. Most of the companies felt that there is more demand for customized products than before and companies were willing to answer this demand. Even though most of the interviewed companies had already customized their products extensively, a trend toward even more customizable products with faster delivery times and more cost-efficient production was evident. Also, some companies thought that with the implementation of mass customization, compared to the earlier non-systematic customization approach, they were able to customize their products to a greater extent due to better management of product structures and product information. Companies were also asked whether or not the life-cycle of their products had shortened during the last five years. Surprisingly, and as a contradiction to earlier studies, the answers were distributed almost half and half; just a tight majority disagreed with the research question and a minority agreed (Table 4). This result was seen as a derivative of the maturity of product modularity among companies that enabled them to introduce product updates more easily and frequently. Companies saw that product modularity and constant development blur the boundaries of product generations and actually decrease the need for totally new product launches.

950

HANDBOOK OF RESEARCH IN MASS CUSTOMIZATION & PERSONALIZATION Table 4: Perceived changes in the business environment (n=32).

Research questions

Yes

No

Don’t know

Are customers more demanding than five years ago?

27

3

2

Are customer needs changing faster than five years ago?

23

5

4

Is demand increasing for non-standard products?

20

9

3

Do you plan to increase the degree of customization?

19

9

4

Is the market lifetime of your products less than it was five years ago?

15

16

1

Observed benefits from implementation of mass customization The interviewed companies were asked what might be their major observed (for mass customizers) or expected (for those planning to implement mass customization) advantages from applying mass customization. The answers were classified and grouped as seven benefit themes (Table 5). The most important reason to apply mass customization was to shorten both throughput and delivery times. A very tight competitive situation together with a geographically long distance from the main market areas was mentioned frequently among the main driving factors in utilizing mass customization strategies. The distance between Finland’s geographic position in northern Europe and the large-scale markets in Europe and North America as well as the growing markets of Southeast Asia and South America gives competitors a few days head start in delivery times. It is an issue that is strongly pushing Finnish companies into modern solutions. More than half of the interviewees pointed out an opportunity to achieve a shorter delivery and throughput time in their manufacturing process when they were asked about their companies' keenest interest in mass customization utilization (Table 5). The second most expected advantage from mass customization was rising from a need to control and manage a great number of custom variations. About one third of the interviewees mentioned this factor. This was typical especially for companies producing capital goods with MRO services for an entire product lifecycle lasting even for decades. Other objectives of mass customization utilization were decreased level of unit costs, increased flexibility and efficiency of production as well as achieving better operational control. Interestingly, a need to increase customer intimacy didn't appear at the top of the list, perhaps as a result of the long production tradition close to customers. Moving into mass customiza-

951


tion meant for many companies a step away from customer-driven agile operations toward customer-focused stable operations. Table 5: Pursued benefits from mass customization, multiple selection (n=37). Pursued benefits from mass customization Shortening the delivery/throughput time

19

Variation management

11

Cost reduction

10

Flexibility/efficiency of production

9

Operational control

9

Customer intimacy

7

Quality improvements

7

Typical challenges in mass customization Companies had met most of their practical problems in variation management (Table 6). Companies wanted to be customer driven and to offer customers everything they wanted, which led to ever-new product variants. The implementation of mass customization had made it easy to make small customer-specific changes in order engineering. The problem was that sometimes these small changes in product qualities caused major changes in production processes. Also, too many new components are needed and created for IT systems only for minor customer changes, not to mention the new supply chain relationships required to be able to produce these new variants. Table 6: Main challenges considering mass customization utilization (n=37). Main challenges in mass customization Variations management / amount of variation

13

Product change management

11

Communication / IT problems

7

Sales configuration

7

Modularization of product structure

6

952


Another big issue seemed to be change management. When and in which order should product changes be introduced to sales departments, to production and to suppliers? Sales representatives need to at least be aware of the new version publication schedule in order to sell all the old products before new arrivals. The problem is that global component stocks and production lines run in different phases; both the producers and suppliers need to know when to change their production to the new components/product configurations to ensure a seamless change to new models. Some other challenges in implementing mass customization strategy mentioned were mainly related to information technology, such as isolation of different information systems and lack of integration. In many cases, there was too many manual information operations conducted between different systems. Overall, the integration of different information systems and some special applications, such as sales configurators, need much more development. Furthermore, some companies had difficulties in defining the modular structure of their products. Trials had led to errors when it was noticed in practice that, e.g., the designed modular structure forced double structures or it was unsuitable for production. Level of variety management practices Blecker and Abdelkafi define seven sequential strategies of complexity reduction and variety management for mass customization: component families, component commonality and process commonality as required steps before product and process modularity, then product platforms and finally delayed differentiation. The first three steps were largely used in the companies studied, and the following steps of product and process modularity were utilized only somewhat. Product platforms and delayed differentiation were scarcely used by companies. Compared to the craft customization approach, the interviewed companies had a clear goal of standardization of components and modules but, at the same time, toward better management of product structures and product data (Table 7). However, the importance of delayed differentiation remains partly unclear. Many of the interviewed companies had succeeded in shortening their throughput time enough to make delayed differentiation unnecessary. The situation became problematic only in cases when the variation point had moved outside their own production lines to suppliers. However, a few case companies were able to handle even this kind of situation exemplarily with deep cooperation with their partners.

953

VOLUME 2: APPLICATIONS AND CASES Table 6: Main challenges considering mass customization utilization (n=37). Component families

Component commonality

Process commonality

Yes

24

21

22

Partly

7

10

8

No

6

6

7

Process modularity

Product platform

Delayed differentiation

Yes

13

6

4

Partly

9

8

12

No

15

23

21

Utilization of sales configurators The utilization of sales configurators varied significantly among the interviewed companies. Sales configurators are used both to demonstrate product qualities for customers and in order entry to collect all needed product attributes systematically and in a digital form right from the start. If used right, configurators enable flawless order information flow from sales to production and even forward in the supply chain. However, many companies had difficulties establishing the discipline required in order entry; salespeople added different remarks in the open text fields that affected the product structure. These remarks were usually noticed way too late when order was already going into production; then it was noticed that the order required some component that needed to be ordered from the supplier, causing lengthy delays. Nearly half of the interviewed companies were using sales configurators to process customer orders (Figure 1). The main factors influencing the possibilities of using configurators seem to have a close connection to the product and process characteristics and production volume. The 16 companies that had never utilized or even planned to utilize a sales configurator produce mostly non-modular simple products. Those products are not easily configurated with selectable distinct options. Likewise, if a production mode still falls into the craft customization category, utilization of configurators is rare. Interestingly, only one of the studied companies had given up on their configurator after its test phase. The other companies (16 out of 37), that have a configurator system in use are still utilizing it at least as one of the several ordering practices.

954


Figure 1: The utilization of sales configurators in companies (n=37).

A typical product configuration process was conducted by a salesperson together with a customer, never by a customer on his own. However, the role of salespersons varied from simple entering of the customer requirements into the sales configurator to creative problem solving of a product expert. The characteristics of the configuration process depended mainly on the complexity of the product and on the rules of configurations: how strictly it is defined. The biggest problem (8 out of 16 users) with configurators was that the rules were not tightly followed or the rules were defined too loosely. Especially, the use of free input fields of configurations was problematic and caused most of the problems. Many times, the fields were misused to convey configuration data that should have been expressed elsewhere. Or, a free text area was used to request some additional customization outside the solution space; a situation where order should have been directed to order engineering for determination of price and delivery time. Only five companies had succeeded in implementing sales configurator systems with clearly defined rules. Partly, this was also a question of sales force discipline. A well-defined configurator system was seen to have some clear advantages by companies using it. The product definition process is seen to be more fluid if the salesperson and customer are able to discuss product characteristics through a limited number of options. Customer requirements can be matched with different product features and accessory options, and the order can be placed into production according to the sales configurator information. Customer requirements are smoothly translated to the product structure, production order and, with a good IT-integration, also to purchase orders. The key issue of system success is the

955


management of product data and a well-defined product structure, including module and component properties as well as their interfaces. A Roadmap towards Comprehensive Mass Customization The interview data suggest that there are some common features in mass customization development approaches among the companies operating in B2B technology industries. The primary observation of different company interviews shows a close connection between a company’s industrial history and its mass customization practices. Thus, this study confirms the earlier results (e.g., Duray et al.) that the key question in moving ahead in mass customization seems to depend on whether a company has its roots in customer-driven individual production (i.e., engineering-to-order; ETO) or in mass production (i.e., make-tostock; MTS).

Flexibility of production processes and machinery

Management of change, commitment, organizational learning, creativity, etc.

Ut i an liza d ti bu rep on o De sin or f m ve es tin lop s i g s onit me o nt ell yste ring nt ig o en ms, to f m ce ols an ag em en t

sto Cu

n, tio mer ac er usto nt ri fc o me nt tion y sto me ma ac Cu e or g f im na in nt ri ma me

COMPREHENSIVE MASS CUSTOMIZATION

n tio ra pe oo of kc t ics n or t tw me is ge log rks Ne na g, o M a rcin etw n u so and

Pr od uc C td mo om ev d u po elo lar ne pm ity nt en , p co t ro mm du ct ona pl lit at y, fo rm s

Production development

Organizational and other factors

Figure 2: McMountain – A comprehensive model of mass customization development.

However, earlier studies concentrated only on certain individual aspects of mass customization development, and a comprehensive view has been missing. The same situation can be found in practice; only a few of the 63 interviewed company managers had constructed themselves a comprehensive view of business opportu-

956


nities that advantageous mass customization utilization can offer. Neither was realized all the requirements needed to implement mass customization successfully. The lack of a comprehensive approach to mass customization development both in literature and practice suggests that more attention should be directed to this issue. In Figure 2, the combined results and experiences of this interview study are transformed into a model of "McMountain," i.e., a comprehensive view of mass customization development. Comprehensive mass customization can be seen as a high mountain situated on flat surroundings. By climbing up to the hilltop, you can see much further ("better business opportunities"), but on your way up, you have to make a full circle around mountain and even to find your paths on more unfamiliar hillsides. In the development process of mass customization strategy, the "hillsides" are named after identified common development areas and themes identified from the interview data collected from 37 companies. Customer intimacy: Customer interaction and management of customer information For mass producers, this means paying closer attention to customer needs and customization of products. But how about for engineering-to-order (ETO) companies that constituted the majority of study population; what customer intimacy means for them? Interestingly, ETO companies perceived that moving on to mass customization can actually enhance customer intimacy by at least three ways: 1) quicker delivery time, 2) more cost-efficient solutions and 3) better quality of customized products. Some companies even argued that moving on to mass customization had enabled them to customize their products more profoundly due to explicitly managed product structures. Moving on to mass customization requires multi-faceted development and especially better management of information processes related to all operations. In the customer interface, this can mean development of customer relationship management systems (CRM), a sales configurator and other sales tools. Systematic management of product information and its representation for customer in a structured and accessible way enhance the customer interaction process and communication of expertise. In the sales process, the customer expectations are created; a fluid and appealing sales process is an important part of the customer experience. Fulfilling the created customer expectations is the other side of the same coin.


957

Product development: Component commonality, modularity, product platforms Product modularity is considered to be a key enabler of mass customization. Modularity allows production of modules and components in volume while offering a greater range of end products. Modularity is seen as an enabler of lowcost customization. While there were also some other approaches in creating the customizability of products, modularity was considered to be the most important enabler among the interviewed companies. It was also considered a major obstacle in the path toward mass customization. Product designing requires time, and therefore, a natural way to introduce modular products would be in the same cycle as new product development and new product family launches. However, new product families are launched quite seldom, which suggests that faster introduction of mass customization should involve modularization of existing products. In either case, product design should be paid attention to. Some companies had have major drawbacks in their mass customization development due to fact that the product was unsuitable for customization. Flexible production processes alone were found not to be enough for successful implementation of mass customization operations. There are several approaches to the modularization of products, and the most obvious modularization criterion is not always the smartest one. And, many times, the product cannot be totally modular – in addition to mechanical interfaces, there are often electrical and hydraulic connections and embedded information technology. Even though most companies had modularized their products according to a structural or functional way, other approaches, such as modularization according to production phases, supplier capabilities, future product development focus areas, product semblance or maintenance needs, were also at least considered. Production development: Flexibility of production processes and machinery Mass customization generally calls for the postponement principle in production that is to delay some of the value-adding activities until a customer order arrives. Some components and modules, common for all or most of the orders, can be produced beforehand to accelerate the throughput times. On the other hand, the fluctuation in customer requirements assumes that at least assembly from readymade modules is done according to the customer order. The point at which the customer order penetrates the production process is called a decoupling point. This point can be located inside a company’s own production line or even somewhere from the supply chain. The location of the decoupling point largely defines the customizability of the product on the one hand, and the

958


length of the delivery time on the other hand. In order to achieve both efficiency and effectiveness of production, as required in mass customization, companies need to combine lean and agile approaches. The push sub-system of standard components and modules should be as lean as possible, and the pull system of assembly and customer specific modules should be agile to enable responsiveness to customer needs. The aim is to be lean up to the decoupling point and to be agile beyond it. Components and modules in the pull sub-system were usually produced in production cells near the assembly line. This way, the production line can be replenished with customer-specific components and modules as efficiently as possible, enabling short delivery times. Assembly lines were typically phased unless the difficulty in moving products around forced slot assembly. The increase in production volumes has forced many companies to build multiple parallel production lines. From the point of view of production flexibility, the specialization of these production lines is not very advantageous. Production lines concentrating on too narrow a product segment decrease the flexibility of production – which is needed in mass customization. Network cooperation: Management of sourcing, logistics and networks A characteristic of the interviewed companies was their high dependency on their suppliers. In many cases, as much as 70-80% of the production costs were generated outside the company. When companies are pursuing operational gains, the development of procurement and supplier cooperation has become an ever more important subject. The competitive situation puts pressure to decrease cost and to shorten delivery times, especially for companies with a background in engineering-to-order production. Many of these companies have embraced mass customization as a solution. By modularizing the product into smaller units, companies are able to outsource component production to cost-efficient suppliers. For standard components, sourcing can be done according to forecast, e.g., from cost-competitive countries. In this case, the balance between delivery times (time required for supplier production and logistic) and stock levels needs to be created carefully. For customer-specific components, rapid responsiveness is more important to be able to keep the delivery times rational. This can mean suppliers located geographically close to production lines or the use of a fast logistical method, e.g., air cargo. In either case, deep supplier cooperation is needed to ensure a fluid supply of customer-specific components. In technologically advanced environments, the competitive ability was not based solely on the head buying company’s capabilities but more on network strategy


959

and its agility in responding to the changing needs of customers. The supplier cooperation was no longer simply buying components and making yearly agreements, but more and more collaborative process development and even joint product development. Head buying companies had in many cases only a limited number of partners (both module suppliers and engineering offices) with whom the cooperation was deep and wide. Development of management tools: Utilization of monitoring and reporting systems, business intelligence Mass customization is a complex system that cannot be transformed to a lowcomplexity level – rather this complexity should and can be managed. Continuous monitoring of operations and key indicators is needed to be able to quickly respond to changing customer needs. The unforeseeable nature of customer order configurations in mass customization puts pressure on responsiveness. Information systems development was considered essential for management of complexity of mass customization. Enterprise resource planning (ERP) systems have to be tuned to handle individual orders in production, sales tools might require some sort of configurator to support product definition process and product data management (PDM) systems were often responsible for management of product information and for production configuration. Mass customization was considered to require systematic and well-defined information management, especially from the product information perspective. This requirement for systematic management of product structures is important for costs calculation as well. Without a detailed product structure corresponding to reality, it is difficult to calculate cost and profitability structures for customized products. The realistic unit cost structure requires constant monitoring of the working hours per product, machine hours, raw materials, components and their up-to-date costs, not to mention a fair allocation of overhead. Many companies were unsure how they should allocate overhead between different products or product groups. In fact, problems in product structure and deficiency in monitoring activities resulted in ignorance of the true costs and profitability of different products or product groups. This is an especially important issue in technology industries where the time from order to production and to delivery can typically reach from months to even years. The price of raw materials, such as metals, can change significantly during that time.

960


Organizational and other factors: Management of change, organizational commitment, creativity etc In the end, neither nice strategy work, nor new technology nor even the best intentions are enough. The people working for the company need to be receptive and concentrate on learning the new working methods and to develop new capabilities. The management of the change process to mass customization requires commitment, complexity management and education of employees. If an organization doesn't stand behind the intended changes, it is very difficult to realize the changes. Many of the very advanced mass customizers had moved on mass customization because of necessity. Problems in variations management, doubled production volumes, a threat of bankruptcy and other compelling reasons seemed to facilitate the change willingness. In a situation where everybody understands that there is no alternative for change, it is much easier to introduce the change. On the other hand, a change in a situation where a "company prepares itself to future changes in business environment" is much more difficult to introduce, especially because of old attitudes: if it works, don't touch it! Yet, the change process to mass customization might take up to 5-10 years, which means that the change initiative has to be made a way before competitiveness has been lost. Comprehensive mass customization Our multiple-case study shows that most of the companies in the Finnish technology industries are already utilizing different parts of mass customization, but only a few advanced companies have embraced the comprehensively different hillsides of McMountain. There is still a lot of potential in developing competitiveness in companies with comprehensive mass customization. Production and engineering orientation was visible, but more and more managers had started to create a more extensive conception of the total business model and factors contributing to success. For example, the role of services was highlighted in many companies, and service business had in fact increased its portion of total revenues. Also, some experienced managers had started to highlight the importance of "soft" issues, such as talking about feelings during the change; something they thought in the beginning of the change process wouldn't fit in the "hard" engineering community of their workplace. Companies need to develop versatile mass customization capabilities before achieving the ability to launch successful mass customization operations. For example, a change process originating from production might trigger the development of many other functions as well, as the next minicase shows.


961

"Our problem was in forecasting the needed product variations. Our customers need fast delivery times since their business is quite fluctuating and unforeseeable, and we tried to answer that challenge by producing products to stock. However, quite rarely our customers were able to find a suitable product from stock, and expensive postproduction customization was quite common. Often, this led to unsatisfied customers. The change initiative came from production. The main objectives were efficient variation management and reducing the cycle times of production to enable configure-to-order production mode. A thorough preliminary study revealed that development of modular product architecture, accelerating the inbound logistics and creation of fast production system was required for achieving the goals. Before starting the project, many companies producing variable products were benchmarked. Existing products were roughly modularized to enable configuration, and at the same time, a new product platform was created for new customizable and modular products. Postponement variation as well as component commonality and module commonality were the central objectives. All products were modeled to sales configurator, which became the only ordering method. Mass customization was the only alternative for increasing production volumes while managing growth. The fastest time from entering the order in the configurator to production order takes now only 15 minutes. The previous 7-8 week throughput time has now been reduced to about 4 weeks’ delivery time. New product development is much faster since not all modules require changes at the same time. Also, outsourcing sub-assemblies and components is much easier since the product structure is clearly described." Conclusions One of the interesting findings was that although the majority of the interviewed companies utilize some mass customization solution in their operations, the concept of mass customization itself was not yet very well-known. The interviewed managers were more familiar with terms like modularity, customer specific production and the like. Also, the development of mass customization practices had typically started as a production or product development project, and the management of wholeness seems to be left behind. The need for mass customization had mainly risen from difficulties in managing numerous product variations, from the pressure to shorten delivery times and decrease the costs of end production all at the same time. Companies felt that they had to develop their operational performance faster than their competitors due to tight competitive situations, even though the interviewed companies were

962


generally very successful and confronted the problem of managing the increased production volumes. Contradictory to most of the mass customization literature, companies in Finnish technology industries mainly approached mass customization from custom-oriented manufacturing premises instead of mass production. Therefore, the main benefits were also seen from a different angle. Gaining customer intimacy was not considered the main driver toward mass customization, although, and ironically, in some companies mass customization was considered providing deeper customization possibilities than custom manufacturing. The most often-mentioned continuing challenges in mass customization practices were found in the areas of cross-functional cooperation and mastering the product data automatically through different information systems. Many companies considered that different information systems are too isolated and are not able to communicate with each other well enough, e.g., from engineering systems (CAD) and sales systems to production. Many companies had also implemented a product data management (PDM) system in order to master the product data and to create an online link between different systems. Also, the challenge of building an effective sales configurator (for sales representatives) was considered challenging. This study charted the state of mass customization practices in Finnish technology industries. As a main result and a compilation of company experiences, a comprehensive model of mass customization development, the McMountain, was suggested. Future research is encouraged to explore different hillsides of the McMountain in more depth and to figure out how different hillsides and the whole McMountain appear in different industries. Also, more empirical research concentrating on B2B commerce is called for to spread the experiences and provide material for subsequent theorizing.

References Åhlström, P. and R. Westbrook (1999). Implications of Mass Customisation for Operations Management: An Exploratory Survey, International Journal of Operations and Production Management. 19(3): 262–274. Blecker, T and N. Abdelkafi (2006). Complexity and Variety in Mass Customization. Management Decision, 44. Christopher, M. (2003). Supply chain structires. In: Gower Handbook of Supply Chain Management, 5th Edition, ed. by J. L. Gattorna, Gower Publishing: 283–295. Christopher, M. and D.R. Towill (2000). Supply chain migration. International Journal of Supply Chain. Management. 5(4): 206–213. Cunningham, J.B. (1997). Case study principles for different types of cases, Quality & Quantity. 31: 401– 423.


963

Da Silveira, G., D. Borenstein and F.S. Fogliatto (2001). International Journal of Production Economics, Mass customization: Literature review and research directions. International Journal of Production Economics. 72: 1–13. Davis, S.M. (1987). Future Perfect, Addison-Wesley, Reading. Duray, R, P.T. Ward, G.W. Milligan and W.L. Berry (2000). Approaches to mass customization: configurations and empirical validation. Journal of Operations Management. 18(6): 605–25. Duray, R. (2002). Mass Customization origins: mass or custom manufacturing? International Journal of Operations and Production Management. 22(3): 314–328. Eisenhardt, K.M. (1989). Building Theories from Case Study Research. Academy of Management Review. 14(4): 532–550. Lau, R.S.M. (1995). Mass Customization. Industrial Management. 37(5): 18–19. Piller, F.T. and K.M. Moeslein (2002). Mass Customization. Proceedings of the ANZAM / IFSAM VIth World Congress, Gold Coast, Australia. Pine, B.J. (1993). Mass Customization: The New Frontier in Business Competition, Harvard Business School Press: Boston. Prahalad C.K. and G. Hamel, The Core Competence of the Corporation. Harvard Business Review. 68(3): 79–90. Salvador, F. and C. Forza (2004). Configuring products to address the customization-responsiveness squeeze: A survey of management issues and opportunities. International Journal of Production Economics. 91(3): 273–291. Soberman D.A.. (1999). It's a whole new ball-game. 17(3): 290–295. Tofler, A. (1970). Future Shock, Random House: New York. Yin, R.K. (1984). Case Study Research: Design and methods, Sage: New York.

Author Biographies Marko Mäkipää is a researcher in Center for Research on Information, Customer and Innovation Management (CIRCMI) and a Ph.D. student for Information Systems Science at the University of Tampere. He holds a master degree in Computer Science from University of Tampere with an emphasis on Information Systems. He is currently writing his Ph.D. thesis on the management of IT in inter-organizational networks. He has participated in several projects related to increasing IT related collaboration in manufacturing networks. He has also investigated the adoption of mass customization in manufacturing industries. He has published articles in e.g. International Journal of Mass Customization, International Journal of Enterprise Network Management and The European Retail Digest. Contact: www.uta.fi/~marko.makipaa | [email protected] Dr. Lea Ahoniemi is the Research Program Director at Center for Research on Information, Customer, and Innovation Management (CIRCMI) at Department of Computer Sciences at University of Tampere, Finland. She has been working as a senior business researcher at University of Tampere since 2000. Ahoniemi is also nominated on 2008 as the Adjunct Professor of Institute of Leadership and Management Sciences at the National

964


Military University. Ahoniemi has conducted numerous research projects on the field of competitiveness abilities of private companies like their mass customization and networking capabilities. One of major themes has been a connection between performance and measuring in HR-policy and work attitudes. Ahoniemi has also studied public sector organizations, their processes and economical efficiency and their cooperation opportunities with the private sector companies. Contact: www.uta.fi/circmi/ | [email protected] Markus Mertanen, Tampere University of Technology: Markus Mertanen worked as a researcher at Cost Management Center in Tampere University of Technology. He reviewed the mass customization practices in Finnish technology industries in his M.Sc. thesis. Currently Mertanen is a Ph.D. student at TUT. Dr. Matti Sievänen works as a senior researcher at Cost Management Center (CMC). Sievänen is one of the founding members of CMC, which is the leading research unit in Finland in the field of management accounting. Sievänen’s research interest lies in combining management accounting tools with mass customization and product variety management. He has published several articles and holds over ten years experience of research projects with industrial companies. He is currently working in a project of masscustomized services. Linnea Peltonen, M.Sc. in Industrial Management, is a mass customization professional having experience both from practitioner and academic side. She has graduated from Industrial Engineering and Management at Tampere University of Technology in 2004. Her research interests are related to mass customization and organizational change management. Before entering her position as Business Intelligence Analyst at Metso in 2008, she worked for Tampere University of Technology (2002-2005), Marimekko Corporation (2005-2006) and University of Tampere (2006-2008). She has published numerous mass customization related articles, for example in International Journal of Mass Customization. Mikko Ruohonen is professor of business administration & information systems at the University of Tampere and director of the Center for Research on Information, Customer and Innovation Management (CIRCMI). He holds professorship at the University of Tampere and docentship at the Turku School of Economics. He has published up to 100 articles, reports and columns, four textbooks and several large research reports. International Federation of Information Processing (IFIP) granted him IFIP Silver Core Award year 2007. Silver Core is conferred on those who have served IFIP as General Assembly (GA) members, committee officers, members of IFIP Congress Program Committees, and editors of proceedings of IFIP conferences. Contact: www.circmi.fi | [email protected]

4.2

Opportunities and Challenges of Furniture Manufacturers Implementing Mass Customization Torsten Lihra Forintek Canada Corp., Canada Urs Buehlmann Virginia Tech, USA Robert Beauregard Faculté de Foresterie et de Géomatique, Université Laval, Canada

Globalization and other inherent factors have created challenges to the North American furniture industry. Imports from low production cost countries did increase greatly over the last five years. At the same time, the U.S. and Canadian furniture manufacturers lost market shares. Mass customization (MC) of furniture products is considered to be a potential strategy to regain competitiveness for North American manufacturers. This study sheds light on North American furniture manufacturers' perception of MC, its potential and challenges to implement it. A survey of furniture manufacturers in Canada, the USA and Germany showed that developing modularity and agility, integrating the supply chain and pursuing a competitive cost structure are critical elements of a furniture customization system. Manufacturer perceived that having end users designing products the true limit of MC. Assisting end users through the personalization process and suggesting a limited number of pre-packaged options should be the favored approach.

Introduction U.S. furniture sales totaled approximately $120 billion in 2004 (U.S. Census Bureau 2007), making furniture the second largest personal consumption expenditure for durable goods in the U.S. after homes but ahead of new cars (Toosi 2002). Competition in furniture production from low labor and production cost countries, which have dramatically increased their exports to the U.S. over the past few years, has led to serious downsizing in the U.S. furniture industry (United States International Trade Comission 2007; Buehlmann et al. 2004; Hilsenrath and Wonacott 2002). Mass customization (MC), for a variety of reasons, is considered a promising strategy for domestic manufacturers to successfully compete in the future (Buehlmann 2004; Schuler and Buehlmann 965

966


2003). While standardized, mass-produced furniture is made more cheaply at similar or better quality in low production cost offshore countries, customized furniture gives producers close to customers a sustainable competitive advantage by allowing more direct, better quality communication with the customer, incurring lower transportation costs, and by a decreased risk of supply chain disruptions. Therefore, the domestic furniture industry should be aggressive in exploring and implementing opportunities to manufacture mass customized furniture. Stanley M. Davis in 1987 introduced the term « mass customization » in his book Future Perfect (Davis 1987). He described mass customization (MC) as a trend towards the production and distribution of individually customized goods and services for mass markets. Pine (1993) considers MC as the historical successor of mass production while Kotha (1995) sees it as a system that may co-exist with mass production. For practical reasons, MC is closely linked to Just-in-Time (JIT) production, as customized products can be sold only to the particular customer who ordered it. The concept of product platforms (Van Vuuren and Halman 2001) is also strongly related to mass customization. Van Vuuren and Halman (2001) point out that the underlying logic of a product platform consists of three aspects: 1. modularity, 2. standard interfaces for assembling and 3. design standards that the modules conform to. Product platforms may be considered as enablers for product configuration. The objective of this study was to assess the furniture manufacturers' opinions towards mass customization and its potential as a profitable business strategy (Buehlmann and Schuler 2008; Oh et al. 2008; Lihra et al. 2008; Schairer and Buehlmann 2007; Buehlmann et al. 2004; Schuler and Buehlmann 2003). Methodology To collect data, an experience survey has been conducted. An experience survey seeks to obtain insight into the relationship between variables rather than to get a simple consensus as to best practices (Malhotra 2004). Experience surveys are qualitative exploratory research being characterized by an unstructured methodology based on small samples that provides better understanding of the problem setting. To shed light on MC as perceived by furniture manufacturers, 23 in-depth interviews in the USA, Canada, and Germany have been conducted in the Fall of 2005. To perform the survey, individual in-depth interviews (IDI) with furniture industry specialists were performed in Canada, the USA, and Europe. A total of 23 specialists from 22 different companies were interviewed. Sample characteristics are summarized in Table 1 to 4. Table 1 presents the company distribution by

967


country. The majority (13) of the 22 companies included in the study were located in Eastern Canada. Four companies were located in North Carolina, USA, and five in northern Germany. Emphasis was put on the household furniture sector (Table 2) as the household furniture industry is the industry suffering the most from imported furniture and is, thus, most in need to find new ways to compete successfully. Five interviews were conducted with people from furniture industry related companies (component manufacturers, software developers, and a furniture manufacturer association). These respondents were included in the survey to get a more accurate picture of the impact of MC on the supply chain. Table 1: Company sample distribution by country. Country

No. of companies

Percent

Canada

13

60

USA

4

18

Germany

5

22

Total

22

100

Table 2: Company sample distribution by industry sector. Industry sector

No. of companies

Percent

Household furniture

10

45

Office furniture

2

9

Upholstered furniture

2

9

Kitchen cabinets

3

14

Furniture components

2

9

Software

2

9

Furniture manufacturer association

1

5

Total

22

100

Table 3 presents the company size expressed by annual sales volume. The furniture industry consists mainly of small and medium size enterprises (SME). There is no standard definition of SMEs and classifications vary across different

968


countries. For the purpose of the present study, SME is defined as companies having an annual sales volume of less than 50 Million $US. Fourteen out of the 22 companies included in the survey fit that definition. The position of the respondents within their companies is presented in Table 4. In one company, the owner and the plant manager were interviewed. All respondents were part of the middle or upper management, with 40 percent being company owners. Table 3: Company sample distribution by annual sales volume. Annual sales

No. of companies

Percent

up to 10 Million $US

2

9

>10 - 25 Million $US

8

36


4

18


1

5


5

23

more than 250 Million $US

2

9

Total

22

100

Table 4: Respondents job title. Job title

Frequency

Percent

Company owner

9

40

General manager

5

22

Plant manager

4

17

R&D manager

2

9

Quality control manager

1

4

Sales manager

1

4

Marketing manager

1

4

Total

23

100

A set of 26 open-ended questions were used as a survey framework. Interviews were conducted at company locations with the exception of one interview conducted at a wood machine show in Germany. Interview durations varied from


969

30 to 70 minutes. With the permission of the respondents, all interviews were recorded by a digital voice recorder and transferred to a computer. Additional notes on interview circumstances and any other particularities were written in a survey logbook. Answers were transcribed to a text file and coded. The coded answers enabled to build a database using SPSS 11.5 software (SPSS 2002). The qualitative nature of the survey did not permit statistical data analysis but the database facilitated interpretation of the information gathered. Results of the Furniture Manufacturer Survey Level and type of MC offered A set of questions addressed the level and type of customization offered by the furniture industry. One somewhat surprising finding of the survey was that generally, end users did not ask for customization. All respondents mentioned that the initial step towards customization came from manufacturers. Furniture manufacturers first offered product customization and assessed customers' perception through sales statistics. After this initial step, customers might have asked for further customization. This "trial and error" approach may be explained by a lack of direct communication between manufacturers and end users. Seven out of the 19 furniture or component manufacturers had no contact with end users. Seven companies got end user feedback through their retailers or sales representatives and five companies conducted market studies. Conducting market studies was not related to company size. Most of the time, manufacturers appear to rely on retailers' willingness to share information about end user needs. Respondents mentioned that the most valuable feedback came from specialized independent furniture stores. These stores seem to have a closer relationship with end users and manufacturers. Asked about the type of customization end users appreciate most, all respondents mentioned color and finishing options first, followed by options on furniture dimensions. Furniture configuration and choices on hardware were also perceived as valuable options. All manufacturers agreed that offering more customization to end users should have a positive impact on their businesses. Industry rankings in regards to MC adaptation The furniture industry leaders interviewed were asked to rank the four industry sectors researched (kitchen cabinet, office furniture, upholstered furniture and household furniture) in regards to their respective level of MC adaptation on a seven point Lickert scale (1=very low customization level, 7=very high customi-

970


zation level). The results presented in Table 5 confirm the hypothesis that the kitchen cabinet industry offers the highest level of customization (6.35). The small standard deviation (0.65) highlights that respondents had little disagreement about the leading position of the kitchen cabinet industry sector in regards to the level of MC adaptation. With an average score of 3.09, the household furniture sector obtained the lowest MC adaptation score of all four sectors. The standard deviation reveals that there was little agreement about the level of MC adaptation of household furniture manufacturers. As can be observed in the marketplace, some companies in the household furniture industry sector offer a high level of customization to their end users. Table 5: Furniture industry sector ranking in regard to customization offering. n

Industry sector

Score

Std. Dev.

23

Kitchen Cabinets

6.35

0.65

22

Office Furniture

4.86

1.32

21

Upholstered Furniture

3.33

1.32

22

Household Furniture

3.09

1.45

The respondents ranked the office furniture industry second (4.86) as to the level of MC adaptation. Distribution channels of professional office furniture manufacturers were described as different from the other sectors. Offices are planned by designers, decorators or architects. End users – people who will work in these offices – are generally not consulted during the planning period. The upholstered furniture industry was rated third in regards to the level of MC adaptation (3.33), its score being close to the score of the household furniture industry (3.09). Also, the end users' buying process for upholstered furniture was described as similar to the one for household furniture. No correlation was found between the variation of the scores, company size and country of location. Barriers to MC adaptation Interestingly, respondents perceived retailers as a barrier in their effort to offer more customization to end-users. The experience of manufacturers revealed that retailers were not convinced that MC would have a positive impact on their profits. In-store end user assistance to customized products increases the time per sale ratio and more product knowledge is needed to explain product options to customers. Respondents mentioned that retailers often choose one particular


971

product configuration they offer in their stores without giving end users the opportunity to customize the product, even though manufacturers may offer such options. However, one household furniture manufacturer offering a variety of product configurations pointed out that a test conducted in a furniture store showed that only 40 percent of the products were sold as presented in the store, while 60 percent of their products sold in that store were configured differently by end users. To overcome the retailers' reluctance to offer more customized products, it may be suggested that MC furniture should not be displayed together with standard furniture. Specially trained sales staff could sell MC furniture in a separated part of a store and offer assistance to end-users efficiently, allowing stores to target the necessary training of their sales force. Such a strategy also would lower the customers' ability for direct price comparisons. Customization offered by manufacturers to retailers was generally limited to product exclusivity and customized packaging. Exclusive products, (exclusive furniture collections or exclusive items in non-exclusive collections) were offered to major accounts. Some manufacturers customized packaging by printing the retailer’s logo on the boxes or by adding special padding to protect furniture from transport damage. Twenty-one out of the 23 respondents (91 percent) were convinced that customized products could be sold at higher prices when compared to standard products. The price premium was expected to depend on the level of customization. Most respondents shared the opinion that the maximum price premium could not exceed 20 percent. In regard to lead-time, all respondents mentioned that order fulfillment time had to be decreased significantly for both, standard or customized products. Respondents did not expect end users to accept longer lead-time for customized products when compared to standard products. The majority of the respondents shared the opinion that the end users' design ability was the true limit to customization. A pre-established set of options offered by the manufacturer to end users was perceived as the most promising way to offer customization. All respondents assigned a key role to retailers in regard to MC. General opinion of the respondents was that qualified sales staff should assist end users through the customization process. A set of questions addressed the challenges of MC implementation as perceived by the respondents. These challenges were perceived differently by respondents working for companies just starting the MC-implementation process and respondents with companies offering MC products for an extended period of time. Companies that started to change from mass production to MC perceived industrial engineering as the biggest challenge. Most of the difficulties were

972


related to batch size reduction (eg., machine set-up time, labor flexibility and work cell integration). Companies already practicing MC, perceived maintaining the right company culture as the biggest challenge. Respondents mentioned the difficulties to maintain the companies' commitment to MC alive, especially on the shop floor and the middle management level. The rather long presence of a mass production environment had a lasting impact on people’s perception of productivity and performance. Indicators were related to volume and economy of scale. The change to MC with indicators like unit production flow, economy of scope and flexibility required continuous training of shop floor workers and staff. Twentyone out of the 23 respondents agreed that implementation of lean manufacturing concepts were necessary to support MC. Also, labor skill requirements were described to be different in an MC environment. A higher degree of technical competence and flexibility were required. Unionized companies had a particular challenge to implement workforce flexibility. The high number of job classes made a rotation of shop floor workers difficult or impossible. To support MC implementation, unions had to be integrated very early in the change process. At the foremen level, more administration and planning skills were necessary to adapt to fast changing production requirements. All respondents mentioned that MC had an important impact on the supply chain. This impact was characterized by smaller order quantities and shorter lead times. To support MC, suppliers had to be carefully selected in regard to their flexibility and capacity to adapt to a changing demand. Respondents from supplying companies mentioned their need to get more accurate information on sales forecasts from the furniture manufacturers to plan production. That pointed to a need for powerful, flexible, and userfriendly information technology (IT). IT was mentioned to be one of the most important enablers for MC by all of the respondents. Twenty-two of the 23 respondents shared the opinion that an investment in IT is necessary to implement MC compared to only 15 respondents mentioning that investment in production technology is required. No specific business model was perceived as particularly adequate for MC. All respondents agreed that customer orientation, supply chain management and marketing do gain in importance when introducing MC. Both, vertical integration and network structures were identified as possible approaches to MC. One respondent pointed out that the managerial competences were an important asset to successful implementation of MC. Future industry competitiveness Respondents expressed their opinion on how they thought the furniture industry may stay competitive in their respective countries. It is interesting to point out that


973

none of the German companies mentioned cost reduction as an important factor. The German companies clearly focused on increasing end users' perceived value. MC, branding, cluster development, service improvement and targeting of niche markets were embraced by those companies. Respondents from North American companies mentioned production cost reduction, MC and faster shipment first followed by service improvement, targeting niche markets and global sourcing. The low weight assigned to cost reduction by the German respondents may be explained by their definition of business strategy which was related to the development of a companies' ability to differentiate themselves in the market place. Low production cost was not expected to become a competitive advantage for the German companies and was therefore not part of their business strategy. This does not mean that cost reduction was not part of their daily challenges. Respondents generally seemed to accept that retailers keep control on the relationship with the end users. Under that condition, it was surprising that only two respondents mentioned an improved and tighter manufacturer – retailer relationship as an important asset for future competitiveness. One respondent pointed out that the furniture-buying experience of end users has to be improved: "Buying furniture today is generally a boring undertaking." When people will say "Let’s have fun today – let’s see some furniture!" the industry will have made a big step forward. Offering babysitter services and play sectors for kids, addition of coffee bars to sit down and read the catalogues, distractions of all kind, attractive presentation of furniture, customer service, customization of furniture, fast shipment … all these concepts may create a pleasant and comfortable environment that impacts positively end user buying behavior. Conclusions Exposed to global competition and increasing imports from low labor and production cost offshore countries, North American furniture manufacturers have to revise their business model to stay competitive. Mass customization (MC) might be a potential strategy to increase the value offered to customers by North American furniture manufacturers, thus strengthening their market position. A survey of 19 furniture manufacturers and three furniture industry related companies was conducted to assess the state of and the potential for MC of furniture as perceived by those manufacturers. Respondents ranked the kitchen cabinet industry first and the household furniture industry last in regard to the level of MC adaptation. Implementation of MC was perceived as an important challenge. Changes in the production system to become more lean and flexible, changes of the company culture that was used for mass production, and supply chain

974


management were pointed out by the respondents to be the most important challenges faced by industry participants when implementing MC. Manufacturers perceived end users' design capacities as the true limit to customization. Consequently, offering the highest number of options directly to end-users should not be the target. Assisting end users through the customization process and suggesting a limited number of pre-ordained options should be the favoured approach, according to findings of this study. Furniture manufacturers seemed to be disconnected from end users. They obtained little or no direct feedback about end users' needs. Retailers control the relationship with end users and manufacturers rely on retailer information to learn about end users' needs. Manufacturers perceived retailers as a barrier to MC. Retailers seemed hard pressed to see benefits from offering customization to end-users. Traditional distribution channels may have to change or alternative distribution channels may be necessary to bring MC furniture to the market. The Internet offers opportunities to create a direct contact between manufacturers and end users and represents a tool to customize products. Direct sales of MC furniture through the Internet are expected to increase in the future. New retail store concepts adapted to MC furniture may be another alternative to sell customized products. This situation leads to research topics that should be addressed in the future such as: What levels of customization are valued by end users? How should the end user customize furniture? What are the necessary conditions for retailers to support MC? Further research on those and other MC-related topics will help the North American furniture industry to develop and retain a competitive advantage and to regain lost market share.

References Buehlmann, U. (2004). Furniture manufacturing revisited, Editorial of the Ims Smart-fm Newsletter no 6, Aidima, Valencia, Spain. Buehlmann, U. and A. Schuler (2008). The U.S. household furniture manufacturing industry in 2008 – status and opportunities. Forest Products Journal. Buehlmann, U., A. Schuler and D. Merz (2004). Reinventing the U.S. furniture industry: Facts and ideas. Proceedings of the Industry Focus Day keynote presentation at the 58th Annual Meeting of the Forest Products Society. Grand Rapids, MI. June 28, 2004. Davis, S.M. (1987). Future Perfect Addison Wesley Publishing Company. Reading, MA. Hilsenrath, J.E. and P. Wonacott (2002). Imports hammer furniture makers, Wall Street Journal, Sept. 20: A2-4. Kotha, S. (1995) Mass Customization: Implementing the Emerging Paradigm for Competitive Advantage, Strategic Management Journal. 16: 21–42.


975

Lihra, T., U. Buehlmann and R. Beauregard (2008). Mass customization of wood furniture as a competitive strategy. International Journal of Mass Customization. 2(3/4): 200–215. Malhotra, N.K. (2004). Marketing research: An applied orientation. Pearson Education Inc., SaddleRiver, New Jersey. Oh, H., S-Y. Yoon and C-R. Shyu (2008). How can virtual reality reshape furniture retailing? Clothing &Textiles Research Journal. 26(2): 143–163. Pine, J.B. (1993). Mass customization: The new frontier in business competition, Harvard Business School Press. Boston. Schairer, S. and U. Buehlmann (2007). Management Innovation the key element for Mass Customization. Proceedings of the 2007 World Conference on Mass Customization and Personalization (MCPC). MIT, Boston, MA. October 2007. Schuler, A. and U. Buehlmann (2003) Identifying future competitive business strategies for the US residential wood furniture industry: Benchmarking and paradigm shifts, USDA Forest Service General Technical Report GTR-NE-30415. SPSS (2002) SPSS Users' Manual, SPSS, Chicago, IL. Toosi, M. (2002). Consumer spending: An engine for U.S. job growth. U.S. Bureau of Labor Statistics. Monthly Labor Review, (November), 12–22. United States International Trade Comission (2007), www.dataweb.usitc.gov/scripts/user_set.asp. February 2, 2007. Van Vuuren, W. and I.M. Halman (2001) Platform-driven development of product families: Linking theory with practice, Paper presented at the conference The Future of Innovation Studies. Eindhoven University of Technology. Eindhoven, Netherlands.

Author Biographies Torsten Lihra is leader of the furniture group at FPInnovations - Forintek Division., Canada’s wood products research institute. He played a key role in creating the Partenariat de recherche sur l'industrie du meuble (PARIM) (Research partnership for the furniture industry), a major initiative for conducting R&D dedicated to the Canadian furniture industry. His research work is focused on marketing and management issues related to the furniture industry. Holding a Master’s degree in Wood Science, Mr. Lihra is presently finishing a Ph.D. project on mass customization concepts for furniture manufacturers. Through his involvement in organising numerous seminars, his knowledge of secondary wood processing and his close alliances with international partners, Mr. Lihra is recognised as a key expert in the Canadian value-added wood products industry. Contact: www.fpinnovations.ca | [email protected] Urs Buehlmann leads the Manufacturing Systems Group at the Department of Wood Science and Forest Products at Virginia Tech. He is a member of the Sloan Foundation Forest Industries Center, an Adjunct Professor at Université Laval and a member of the board of WoodLINKS USA. Urs moved to Virginia Tech in 2007 from Enkeboll Designs, where he was General Manager. His research focuses on manufacturing systems engineering, lean manufacturing, business benchmarking and competitive strategy. Contact: www.woodscience.vt.edu | [email protected]

976


Dr Robert Beauregard is dean of the Faculté de foresterie et de géomatique from the Université Laval. He was between 2003 and 2008 holder of the Industrial Research Chair on Engineered Wood Products for Structural and Appearance applications (CIBISA). His area of expertise is the modeling of manufacturing systems for the forest industries. He develops comprehensive approaches to the design of business models taking into account the interactions between the wood resource, process development and innovative products for better business performance. In 1995-97, he was research scientist with the New Zealand Forest Research Institute. From 1997 to 2000, he has been at the Eastern Laboratory of Forintek Canada Corp. where he was instrumental in the creation of the Department for Value Added Wood Products.

4.3

Mass Customization in the Ophthalmic Lens Industry: Progressive Addition Lenses for Your Visual Map Begoña Mateo Institute of Biomechanics of Valencia, Polytechnic University of Valencia, Spain Rosa Porcar-Seder Institute of Biomechanics of Valencia, Polytechnic University of Valencia, Spain José Salvador Solaz Institute of Biomechanics of Valencia, Polytechnic University of Valencia, Spain José David Garrido-Jaén Institute of Biomechanics of Valencia, Polytechnic University of Valencia, Spain Juan Carlos Dürsteler Indústrias de Óptica, S.A. (INDO), Spain Antonia Giménez Indústrias de Óptica, S.A. (INDO), Spain Carmen Prieto Indústrias de Óptica, S.A. (INDO), Spain

The paper describes the Progressive Addition Lenses (PAL) personalization system obtained as a result of a joint R&D project conducted by Indústrias de Óptica, S.A. (INDO) and Institute of Biomechanics of Valencia (IBV). Traditionally, users have been asked to adapt to progressive lenses that are designed to fit an average wearer. INDO proposal is that a customized progressive lens that mimics the natural vision can be obtained by measuring the visual strategy of each individual user, defined as the coordination of eyes and head movements. The result is EyeMADE "made by your own eyes" and represents a major scientific advance and has positioned INDO at the head of the progressive lens field. The advantages of EyeMADE over conventional lenses are clear since it optimizes the visual comfort and allows for an easier adaptation process. A simple, robust and reliable system known as VisualMap DEVELOPER was developed in laboratory conditions to measure the visual strategy. Research showed that this simplified technique of measuring the visual strategy was equivalent to the more complex laboratory measurements. The individual visual strategy obtained with the VisualMap is then used by

977

978


INDO to manufacture the personalized EyeMADE PAL using the latest ophthalmic lens technology using computer numerically controlled systems.

The Optical Industry The ophthalmic lens industry has shown a worldwide change over the last few years. This is mainly due to the conversion from glass to plastic lenses and also to the growth of high value products such as progressive lenses. Nevertheless, the ophthalmic lens market in western countries may be described as relatively stagnant. Thus, the growth in volume has been kept relatively stable despite an increase in value. In general, growth in volume may be determined by the global demand for corrective lenses. Growth in value is the result of advances in the development and innovation of the optical surface of corrective lenses, materials and lens treatments. Nowadays, there is an increase in the demand of sophisticated lenses that provide a value-added product. Progressive addition lenses (PALs) are the most complex lenses on the market. They are intended to solve the problems caused by presbyopia, a physiological aging condition that affects everybody starting at approximately 45 years of age. PALs have a distance section towards the top of the lens and a reading section towards the bottom nasal area of the lens. A progression corridor connects both sections of the lens. Hard progressive lenses appeared first in the market. They had a large distance and near zone but bother users with an abrupt peripheral blur. More recent designs, named soft progressive lenses, are aimed at reducing the harshness of hard lens designs by decreasing the amount of blur in the peripheral blending zones. Over the last five years, the production of ophthalmic lenses has changed dramatically due to the advances in lens technology. In particular, computer numerically controlled (CNC) free-form grinding technology has allowed to produce virtually any possible lens design enabling the individualized production of complex lens designs as well as the customization of designs based upon the visual needs of the individual wearer. Nearly all the subjects older than 50 years require some sort of optical correction due to the lack of accommodation at near (i.e. presbyopia). This combined with the fact that our population is aging may result in a volume growth in the ophthalmic market in the coming years. The breakdown in lens products and lens value (Figure 1) in the Spanish ophthalmic market in 2005 is given as a representative example of the trends in the ophthalmic lens market. It is worth noting that the progressive lenses represent the highest lens value despite their lower percentage in volume.

979

VOLUME 2: APPLICATIONS AND CASES Bifocal 4% Progressive 19%

Bifocal 6%

Single vision 45%

c

Progressive 49%

Single vision 77%

a

b

Figure 1: Breakdown of lens market (percentages) according to lens products (a) and lens value (b).

Manufacturers are constantly working to improve their lens designs to maintain their position in a highly competitive market. Mass customization has allowed the development of customized progressive lenses either task specific (e.g. lenses adapted to computer use) or frame specific (e.g. lenses adapted to small, fashionable frames). In doing so, the obtained lens presents functional benefits to the users. For example, a secretary working in an office and complaining of poor vision with the use of traditional progressive lenses may benefit from wearing progressive lenses specifically designed for office use only. In addition, the mass customization industry has the potential to increase the progressive market volume without increasing the number of wearers since every wearer may have both a general and a specific-purpose lens. Finally, another relevant indicator in the ophthalmic lens market is lens replacement. Nowadays, lenses are replaced every three to four years. The relatively long replacement time limits the ophthalmic lens growth but it also enables potential growths in the market if a reduction in the lens replacement takes place. Efforts are continuously made by manufacturers and opticians to encourage users to renew their spectacles more frequently even if a change in prescription is not required. Spectacles are no longer considered to be an unfashionable accessory but they are increasingly used to make a fashion statement. At the moment, INDO is the leading Group in the Spanish optical market. INDO has three main business areas: lenses, frames and capital equipment. In other words, the INDO Group manufacturers and distributes lenses and frames,

980


distributes optical and ophthalmological instruments and machinery and finally designs the interior spaces of opticians and ophthalmologic practices. The leading position of INDO in the Spanish market is the result of over 65 years of experience in the field of ophthalmic optics. During the last five years, INDO has been working towards the development of some technology that has permitted to create a personalized progressive lens for each individual user according to their visual strategy. The result is EyeMade, a state-of-the-art development in personalized progressive lenses based on VisualMap DEVELOPER (Figure 2). EyeMADE acts as an acronym of "made by your own eyes" and was launched in September 2005. This major scientific advance has positioned INDO at the head of the progressive lens market. The personalized progressive lens is obtained with the aid of VisualMap DEVELOPER, an easy-to-use instrument patented by INDO. The VisualMap DEVELOPER captures an individual’s visual strategy (i.e. a unique visual map according to the user’s movements of eye and head) in a three dimensional manner. This ensures optimum vision across the full range of distances. EyeMADE is the result of a project collaboration between INDO and the Institute of Biomechanics of Valencia. Clinical studies on user satisfaction conducted by INDO indicate that the personalized EyeMADE lens provides greater comfort to users.

Figure 2: EyeMADE lens.

The Company: INDO INDO was founded in 1937 in Seville, Spain. Two years later, it was transferred to Barcelona where the Company first started to manufacture optical lenses and frames. During the 1940s, the Company opened more branches in Seville, Madrid and Valencia. The 1950s saw the construction of the company’s first industrial premises in Hospitalet (Barcelona) to manufacture frames, sun glasses and fused


981

bifocal lenses, accompanied by the first steps in melting optical glass. In the 1960s, INDO experienced a considerable growth in all ophthalmic and optics fields. This was also accompanied by large technological developments. In the 1970s, INDO inaugurated a new office in Bilbao and branches in other important Spanish cities: A Coruña, Zaragoza and the Canary Islands. The decentralization process of its productive structure became the first step of transformation into a international company. A few years later, the Company started to trade on the Stock Exchange. Two centers were opened in Morocco in the 1990s. At the same time, the traditional organizational structure was transformed into one based on business units. In 1997, INDO obtained the first quality certificate from AENOR, Spanish association for standardization and certification for the design, manufacture and distribution of lenses, and initiated trading on the Stock Market. The end of the 20th and the beginning of the 21st century may be defined as the beginning of a new Corporate Strategy at INDO. This change was due to several changes in the optical sector in Spain such as the entrance of major international optical operators and the massive importation of products manufactured in countries with lower labor costs. Thus, INDO started a process of Global Internalization aiming to:

Achieve a quantified target of international turnover. The 30% of the overall turnover figure.

Establish new factories in countries where labor costs are low.

Acquire shareholdings in companies that distribute INDO’s products. Prepare management a team, which has the required resources to achieve the above goals. INDO has three principal market strategies: efficiency through cost improvements and structural organization, differentiation by means of product innovation as well as higher added value services and the internalization of the Group. Innovation in new products and services is one of the major keystones in the Company’s development. The R&D group is not only formed by doctors and university graduates employed by the Company but supported by several professionals working in leading external universities and technological institutes cooperate in the development of INDO’s projects as well. Currently, 3% of the turnover is invested in R&D projects. In addition, it is worth highlighting that 50% of the Company’s turnover is achieved by products launched by the Company in the last four years. The INDO Group is currently made up of 15 companies: five production centers in Spain, China and Morocco, and commercial subsidiaries in the USA, France, Portugal, Morocco, Chile, Italy and Germany. The INDO Group also exports products to over 90 countries.

982


Progressive Addition Lenses Optimal Design Corrective lenses are personalized products. This is not only because they are made to the optical prescription of each wearer but also due to the fitting of the lens to the frame. In general, progressive addition lenses (PAL) compensate for the lack of accommodation that occurs physiologically to individuals over 40 years of age. PALs have several advantages over other forms of optical correction. For example, PALs are far more aesthetic than bifocals since there is no visible line on the lens. However, most PAL designs are made to fit an "average wearer", which may leave a non negligible part of wearers below the optimum performance. It is common that PAL users complain of adaptation problems. Research in this field has indicated that despite that users are able to distinguish among different PAL designs (Wittenberg et al. 1989; Pope 2000; Preston 1998; Brookman et al. 1988; Krefman 1991; Fowler et al. 1994), the PAL preference vary among users. The effect of different PAL designs on user head-eye mobility and the resulting subjective response has previously been investigated (Afanador and Aitsebaomo 1982; Von Buol et al. 1991). However, such studies are not conclusive and the subjective user preference has not been directly link to the different existing PAL designs. INDO in collaboration with a research center specialized in biomechanics (Institute of Biomechanics of Valencia, IBV) created a joint R&D project to investigate the following questions:

Users capability to discriminate, in terms of subjective response, among different PAL designs.

Quantification of differences in visual strategy provoked by different PAL designs.

Relationships between the users own visual strategy without a PAL and the modified visual strategy while wearing a PAL. It was also evaluated whether the subjective assessment of PAL preference corresponded to the natural visual strategy or not. To carry out the research studies at the R&D department of INDO, users performed visual tasks with targets placed at different distances, from infinity up to reading tasks. Movements of head and trunk were measured by means of a magnetic coil system, the 3SPACE FASTRAK (Polhemus, Colchester, VT 05446, USA) while an ASL Model 504 infrared camera (Applied Science Laboratories) followed the line of sight of their eyes. This is graphically shown in Figure 3.

The ultimate goal of this investigation was to find clues that could lead to the personalization of PALs to individual users rather than to an "average user", as done in the past. In contrast to conventional progressive lenses, personalized


983

progressive lenses are individually designed to ensure maximum comfort to the final user. Currently, there are four levels of personalization in the progressive lens market as shown in Figure 4.

Figure 3: Experimental session at the R&D department of INDO. The user is wearing the sensors at the head and trunk. The monitors show the location of the eye as recorded by the infrared camera and the scene that the user is viewing during the experimental session.

Figure 4: Four levels of personalization of progressive addition lenses.

Level 1: Pre-established designs The choice of the PAL, selected from mass produced options, is based on the use of the lens according to the wearer’s profile. Therefore, the personalization is limited to the range of available PAL designs restricting the level of personalization to a selection of pre-established designs rather than to the manufacture of a

984


unique, personalized lens for a specific user’s visual needs. Moreover, there’s no scientific foundation in the available literature related to the way of determining the user profile from a series of questions regarding their work or visual habits. Level 2: Frame/face measurements Including measurements related to the situation of the frame regarding the face. Typically this includes information such as pantoscopic angle, vertex distance, wrap angle of the frame and monocular pupillary distance. This level of personalization presents the following main disadvantages: The wearer’s visual strategy is not taken into account. Measurements such as vertex distance can vary during the day as a consequence of the frame weight and natural movements of the body. Level 3: 1 D visual strategy This personalization considers the user’s visual strategy to manufacture the lenses. Thus, an instrument has been produced to determine eye/head movement for a specific individual, the results of which can be incorporated into the progressive surface. The instrument consists of three light emitting diodes (LEDs) providing visual stimulus, the central one is viewed at 40cm from the center of the patient’s forehead, and the remaining two are located 40cm on either side of the central one. The patient is asked to look at the LED, which is illuminated, and their head movement is recorded by an ultrasonic signal, which is emitted by the system and reflected by a transponder attached to the special trial frame worn by the patient. The LEDs are randomly illuminated. The final output of this system is the average angle of rotation of the head plus the standard deviation of the measurements around it. The main pitfall of this level of personalization is that horizontal movements are measured in one dimension only. Thus, the output of the measurement is very limited to offer true personalization. Level 4: 3 D visual strategy 3 D visual strategy represents the highest level of personalization in the ophthalmic lens market, today. EyeMADE lenses by INDO are manufactured with the unique visual strategy of each user. From the raw data obtained with the VisualMap DEVELOPER, it is possible to obtain a graphical representation of the frequency of use of the visual space known as VisualMap. A VisualMap represents the areas most frequently used by the wearer when performing a threedimensional task. Research conducted by INDO and IBV has indicated that the visual strategy is a unique, repeatable feature for each individual subject and is easily understandable by users and clinicians. The fact that the measured


985

VisualMap was shown to be very repeatable for each subject but showed a large variety of shapes among different subjects, allowed INDO to create a customized PAL design taking the VisualMap information as the input for personalization.

Figure 5: VisualMap representation of two individuals performing the same task.

Figure 5 shows two VisualMap representations of two individuals performing the same task. The horizontal and vertical axis (centered at 0), represent the horizontal and vertical angle of movement of the eyes, whereas the grey scale indicates frequency of use of that particular combination of angles. The map on the left corresponds to a person that moves the head in a greater proportion than the eyes. The map on the right reveals a person that moves the eyes more than the head. Note also that the frequency distribution in both cases is not regular and some areas are privileged over others. In order to obtain a VisualMap representation in the clinicians practice, it was necessary to create a compact version of the research lab instrumentation that could provide quick, comfortable and non-invasive measurement of the visual strategy. The instrument developed contains two stereoscopic cameras that record the movements of the head and eyes with the aid of a wireless diadem.

Figure 6: Visual Map Developer instrument.

986


The process is controlled by means of a touch screen. The compact instrument (Figure 6) is made of two screens, which stimulate the visual system around two basic ergonomic postures:

The straight (also known as natural) posture, where the head stands up looking around the horizontal line of sight.

The flexion posture used when looking at the inferior field when doing activities such as reading. Please note that the shape of the frame as well as the design of the lenses worn by the user could influence the natural visual strategy measured. Thus, to record measurements using a VisualMap DEVELOPER it is required to remove any refractive correction worn by the user. The measure of the visual strategy using the VisualMap DEVELOPER instrument consists of four stages (Figure 7):

Step 1: Placing the user within the optimum field of vision of the cameras. The system detects the position of the diadem and suggests the required movements to place the user correctly within the two-camera optimal field of view.

Step 2: Checking location of the pupils and the diadem. A photograph is obtained from each of the cameras. The location of the pupils in the photograph should be indicated.

Step 3: Carrying out the test. The red luminous stimulus moves in all positions of the visual field with a constant speed. The user is asked to follow the stimuli in a natural and comfortable way. Both, eyes and head movements are freely allowed.

Step 4: Screen shows the resulting VisualMap. Once the test is finished a 12character alphanumerical code as well as the representation of the VisualMap appears on the screen. The alphanumerical code has the encrypted information required to manufacture the personalized PAL. An indicator located at the bottom right corner indicates the stage of the process by showing one to four green squares. It is easy to understand that a person whose VisualMap shows only slight vertical eye movement will not need such a long corridor length as someone whose eyes sweep across large vertical areas of the lens. Likewise, horizontal movements affect the width of the areas and the distribution of side aberrations. Also note that the size of the frame and therefore the required fitting height are important because the features of the design will have to be adapted to the size and shape of the frame, as well as the visual strategy. INDO has taken this into consideration and the EyeMADE lens does not restrict the users to freely choose the frame that they wish to wear. The algorithm


987

used in the calculation of the required lenses bears in mind the fitting height of the glasses to assign the corridor length. EyeMADE is targeted to users that demand the latest ophthalmic lens technology to achieve not only excellent vision but an easy adaptation to the lenses as well. A typical EyeMADE user is at least 40 years old and belongs to the middle and upper social classes. The lenses currently cost more than traditionally produced progressive addition lenses but the visual benefits are immediately apparent.

Figure 7: The four simple steps required to obtain a VisualMap.

Mass Customization Capabilities and Processes Up until now, ophthalmic lenses have been mass-produced. In contrast, the EyeMADE design is a unique concept that cannot be mass-produced in a traditional way. To be able to produce these lenses, two technologies have been developed in parallel with the developments in ergonomics that have led to the Visual Map Developer. One has been the automatically design system that converts the prescription plus the visual map code into a particular design suitable to offer the best performance. The other dealt with the way of producing personalized lenses, using the so-called free-form technology. With this flexible system, it is possible to produce customized unique lenses. Thus, each manufactured lens is

988


different in design and performance from all others according to the natural visual strategy of each wearer, the prescription and the fitting height. Also taken into account is the correlation between the size and frequencies of the VisualMap for different angles of vision and the size of the distribution of optimum powers and astigmatisms found through tests performed on wearers. Basically, free-form technology incorporates the use of numerically controlled machines for grinding the surface of the lens that contains the free-form design, typically combined with the toricity of the prescription on the backside of the lens and codified as a form of B-Spline shape. Then, a polishing machine with a flexible polishing system specially developed for the purpose of respecting and maintaining the ground surface, is used. Product development capabilities Despite the fact that INDO Group has several production centers for lenses in Spain, Morocco, China and Thailand, all EyeMADE lenses are produced at the center located in Barcelona. This laboratory has the latest technology capable of manufacturing optical lenses using free-form technology. With this technology, the personalized calculations can be made after the order is received and digitally transferred to a cutting machine with computerized numerical control that cuts the lens surface with a high level of precision. The computer numerically controlled (CNC) generator creates the lens surface according to the user-customized lens parameters. The CNC cutter is a single point diamond tool. The lens is then smoothed and polished using a flexible-polishing machine that ensures that the pads maintain the correct surface geometry while providing optimal lens clarity. With the latest technological advances, a lens can be transferred directly from the CNC generator to the polishing cycle without de-blocking. This guarantees that the path of the polishing tool follows precisely that of the generator. A summary of the laboratory processes the EyeMADE lens undergoe is shown in Figure 8. The result is EyeMADE, a unique lens for every user, made using today’s most advanced visual-measuring techniques and the latest manufacturing technologies protected by three patents. EyeMADE combines prescription and progression on the concave side of the lens. The lens includes the user’s initials marked by laser. The advantages of free form processing over traditional lens processing are as follows:

Free-form technology allows lens manufacturers to optimize the optics of each lens to the wearer’s prescription. It is also possible to compensate for the effects that variables such as vertex distance have on optical performance.


989

Figure 8: Flow-chart of the laboratory processes of the EyeMADE lens.

The lens inventory is limited to stocking uncut blanks that may be ground to any design prescription.

Free-form technology promises to become an excellent way for bringing small batches of low-demand lens designs to the market in a cost effectively way. Similarly, it allows manufacturing small batches of popular designs in lowdemand lens materials.

By tailoring the optical system closely to the wearer, personalized lenses will deliver a more natural visual experience. This is the philosophy behind the development of EyeMADE lenses. Finally, it is worth noting that the term free-form is not directly equivalent to customized lenses. There are free-form lenses available in the ophthalmic market that do not have any customization and are therefore comparable in performance to their semi-finished counterparts.

EyeMADE Product description EyeMADE is a completely personalized progressive lens given that:

Personalization is achieved through the VisualMap, an exclusive concept and technology patented by INDO. VisualMap technology records the wearer’s visual strategy, as well as the movements of the eyes and head while performing a visual task in all directions of gaze.

990


When designing the personalized lens, the ergonomic information obtained from the VisualMap is automatically applied to the prescription and the progression, which are finally combined on the concave side of the lens using a new algorithm patented by INDO.

INDO’s personalized lenses are manufactured using free-form technology. This is the result of patented computerized numerical control equipment and flexible-polishing machines. Such technology makes it possible to convert the unique design mapped for each specific person into a unique lens manufactured specifically for that user.

The lenses include the user’s initials marked by laser.

Description of sales channel INDO’s sales channels are independent retailers of ophthalmic lenses or opticians branches worldwide. EyeMADE lenses can be ordered by telephone, fax and/or through the website via Indonet Orders Service. When placing an order, the client should provide all standard details regarding the user prescription (i.e. sphere, cylinder, axis and addition), as well as the 12-character alphanumerical code from the VisualMap and the fitting height. Thus, only those practices that have the required VisualMap DEVELOPER apparatus are able to take orders. This enables optical retailers that distribute EyeMADE lenses to distinguish themselves from others by providing their customers with the latest ophthalmic lens designs in the market. Logistics and quality Once the lenses are manufactured, it is a key factor for INDO to minimize the lead time to customers given the extremely demanding and highly competitive optical market. INDO guarantees a delivery of the EyeMADE lenses within 5 to 10 working days. In addition, the company’s SAP system allows all the call centers in Spain and abroad to manage orders and distribution of the lenses effectively. Each pair of EyeMADE lenses is delivered in a box shaped as an eye with a display of the EyeMADE logo (Figure 9). The box contains the following:

Lenses are packed inside an exclusive EyeMADE envelope.

EyeMADE sueded microfibre cloth.

EyeMADE personalized warranty card.

DVD containing information regarding EyeMADE mass customization process.


991

Figure 9: Packaging EyeMADE lenses including: box, lenses envelope, microfibre cloth, warranty card and information DVD.

The completely personalized warranty card with the wearer’s full name also includes the VisualMap representation, the alphanumerical code and the product number, which facilitates immediate product traceability by providing the manufacturing date, time and batch. The Quality Management Systems of the Company have been adapted and certified by the Spanish Agency for Normalization and Certification (AENOR) under the new regulation, ISO 9001:2000, in the three Business Units. The control of the production registers and of the different quality tests certify that our products are valid to be commercialized in markets as competitive and demanding as, the US, Spain, France and Germany among others, fully complying with the international regulations required by the different countries. Likewise, INDO controls and examines all non-conformity from external and internal customers in order to review and adjust standards. Thus, adapting them to the demands of the different markets. Case Assessment A SWOT analysis of EyeMADE reveals a set of distinguished strengths, weaknesses, opportunities and threats (Table 1). Advantages are primarily based in the pioneer technology as well as the group experience in the optical industry. Like any other novel product, the main disadvantages are its potential competitors, which maybe expected to offer customized products in the near future. Currently, there is no other commercial firm offering a progressive addition lens with such high level of personalization in the optical market, which is one of its strengths.

992

HANDBOOK OF RESEARCH IN MASS CUSTOMIZATION & PERSONALIZATION Table 1: SWOT analysis.

Strengths Company with large experience in the optical market. Pioneer technology. Great acceptance and demand by clinicians. Lens with high added value. State-of-the-art lens at the forefront of technology. Opportunities No real competitor offering the same degree of personalization. High quality product. Potential for expansion in further markets. Capable of manufacture editions in small lot sizes or specific uses if needed. Progressive lens market in continuous growth.

Weaknesses Changing market particularly influenced by low-cost products.

Threats Potential competition from other optical lens manufacturers.

Table 2 shows that INDO management’s efforts concentrate on leadership and organization, product development, production technology and IT systems. These four competencies (or capabilities) of mass customization play the most important role for the success of INDO’s business. Table 2: Mass Customization success factors and competencies. Competence fields

Rank

Leadership and organization

1

Product development

2

Production technology

2

Logistics

7

IT systems

4

Product configuration systems

5

Complexity management

6

The implementation of mass customization concepts in the optical industry may be considered to be in its earliest stages. The EyeMADE lens has positioned INDO at the head of the progressive lens' field since, at present, no other


993

commercial firm offers such high level of personalization in the optical market. However, the appeal of other well-established optical and ophthalmic products despite the lack of personalization may represent the major competition for INDO. This is due to the fact that status and brand names constantly dominate our society. INDO is prepared to focus on product quality to maintain its position as the leader optical company in PAL personalization. Permanent sales monitoring, customer demands and technological development is among INDO’s priorities. It is also worth noting that INDO’s customers can choose among an extensive range of pre-designed PAL. This is the case for the recently launched LifeMADE lens collection, which is manufactured using the latest free-form technology but is not individually customized for each user.

References Afanador, A.J. and Aitsebaomo, P. (1982). The range of eye movements through progressive multifocals. Optometric Monthly. 73: 82–87. Brookman, K.E., Hall, E.A. and Jensen, M.J. (1988). A comparative study of the Seiko P-3 and Varilux 2 progressive addition lenses. J Am Opt Assoc. 39: 406–410. Fowler, C.W., Beavis, A.M., Bench, B.P. and Kempster, A.J. (1994). A wearer comparison of two progressive addition spectacle lenses. In: Vision Science and its Applications. Technical Digest Series (2), Optical Society of America, Washington DC. 1994: 6–9. Krefman, R.A. (1991). Comparison of three progressive addition lens desings: a clinical trial. Southern J. Optom. 9: 8–14. Pope, D.R. (2000). Progressive addition lenses: history, design, wearere satisfaction and trends. In: Vision Science and its applications, V. Lakshminarayanan (Ed) (OSA TOPS, 35): 1–16. Preston, J.L. (1998). Progressive addition spectacle lenses: design preferences and head movements while reading. PhD dissertation, The Ohio State University. Von Buol, A., Menozzi, M. and Krueger, H. (1991). Using multifocal and progressive lenses at VDU workplaces. In: Designing for Everyone: Proceedings of the 11th Congress of the International Ergonomics Association, London, Y. Queinnec & F. Daniellou (Ed.) (London: Taylor & Francis): 88–90. Wittenberg, S., Richmond, P.N., Cohen-Setton, J. and Winter, R.R. (1989). Clinical comparison of the True Vision omni and four progressive addition lenses. J Am Optom Assoc. 60: 114–121.

Author Biographies Mrs. Begoña Mateo holds a MSc. In Engineering at the Polytechnic University of Valencia. She is researcher at the Institute of Biomechanics of Valencia with 7 year of experience in functional and emotional evaluation and design of products. She is involved in European projects in the area of product customization and she is co-inventor of a patented method and device for determining human visual behavior and of a method for

994


customizing spectacle lenses (WO2005107576 A1). She has also published many papers about biomechanics, functionality, comfort and emotional assessment of products. Contact: www.ibv.org | [email protected] Dr. Rosa Porcar Seder leads de Competitive Intelligence Unit at Institute of Biomechanics of Valencia, Spain. As engineer specialised in statistics, she has focussed her activity in the last 11 years heading a group devoted to apply user orientation and ergonomics approaches to the improvement of industrial products. Contact: www.ibv.org | [email protected] Dr. Jose S. Solaz leads the Automobile and Mass Transport Dept. at the Institute of Biomechanics of Valencia. Before entering this position, he worked as a Researcher in the same Institute (2001-2006) after a professional experience in several companies of the automotive sector. His research focuses in human factors in automobile and mass transport, user focused design, perceived quality and emotional engineering. Contact: www.ibv.org | [email protected] Mr. David Garrido. His background is in Industrial Engineer. He studied at the Polytechnic University of Valencia. Since 2005 he is the responsible for the Functional Assessment Section of the Institute of Biomechanics of Valencia (IBV). He has been involved in the field of biomechanical research since 2002. He has participated in several projects related to development of new technologies at a European level, (e.g. DRIFTS and IPCA projects). As the responsible person of IBV’s Functional Assessment Section, he co-ordinates IBV’s R&D activities in that area at both, technical and scientific levels, He is also the author of several research studies related to functional assessment. Contact: www.ibv.org | [email protected] Dr. Juan Carlos Dürsteler leads the Research and Development dpt. of Industrias de Optica S.A., the leading ophthalmic optics company in Spain. He is also part time associated professor at Pompeu Fabra University in Barcelona. He has been engaged for the last 25 years in the research of progressive addition lenses and advanced ophthalmic lenses and surface coatings leading multidisciplinary teams composed of physicists, chemists, engineers and opticians. The fields covered include Numerical Computation, Optics, Computational Fluid Dynamics and Human Factors, all of them related to the development and manufacturing of ophthalmic lenses. Since 2000 his research focus has relied on Personalization of Ophthalmic Lenses, from the design to the enabling technologies that make it possible. Currently he leads an FP7 funded European Project which aim is the personalization of ophthalmic spectacles. Mrs. Antonia Gimenez Carol holds a postgraduate degree in optometry. She leads the Ergonomics (Human Factors) group in the Research and Development dpt. of Industrias de Optica S.A. (INDO). She has more than thirty years of experience in research of new ophthalmic products. She developed the first Clinical Trials in 1983 at INDO with the objective of integrating the user in the design of ophthalmic lenses. Since 2000 her research focuses on personalization of ophthalmic lenses including the ergonomic parameters of the user.


995

Mrs. Carmen Prieto is graduated in Optics & Optometry and she is finishing a Master in Optometry and Vision Science at UPC University. She works in the Research and Development dpt. of Industrias de Optica S.A. where she has acquired more than ten years of experience working at the Ergonomics group. Her research focuses on the Clinical Trials of user satisfaction, specifically on the methodology of evaluation of human factors to obtain objective information for the improvement of the products in development.

4.4


Richard H.Y. So Department of Industrial Engineering and Logistics Management, Hong Kong University of Science and Technology, China John Au Department of Industrial Engineering and Logistics Management, Hong Kong University of Science and Technology, China K.L. Leung Rehabilitation Engineering Center, Hong Kong Polytechnic University, China

Music played on a theater stage provides listeners with a sense of auditory spatial presence as well as a clear perceivable incident angle of where the music is coming from (Ando, 1985). In other words, listeners inside a theater can close their eyes and feel that they are inside a spacious building as well as being able to hear where the stage is. On the contrary, listening to music presented through a pair of earphones lacks both the auditory spacious feeling and, very often, listeners would perceive that the music is coming from the center of their heads (Blauert, 1997). Using personalized head-related transfer function (HRTF) filtering technology, it is possible to simulate the acoustics effects of a concert-theater for music presented on a pair of earphones. However, such a personalized solution can cost over US$2000.0 and may not be feasible for consumer products. Non-personalized solutions, on the other hands, do not work well. This paper discusses the problems associated with the non-personalized solutions and challenges and opportunities of masscustomized solutions. Progress in tackling various challenges is also reported. Potential industrial applications include high-end surround sound solutions for iPods and MP3 players.

Introduction Humans can perceive the incident direction of a sound cue. When a listener is inside a concert-theatre, he or she can perceive the correct incident angle of the source of music played on the stage. Incident sound waves interact acoustically with listeners' outer-ears (i.e., pinna) before entering the ear-canals. These 996


997

interactions modulate the incident sound waves according to their incident angles and listeners' auditory cortex are able to de-code the modulated signals to discriminate whether an incident sound is coming from the front or back and coming from above or below the ears. Because individuals have unique shapes of pinna, these modulations have large individual differences (Blauert 1997). For a particular ear and an incident sound from a particular angle, the frequency characteristics of such modulations can be measured as a transfer function called head-related transfer function (HRTF: Blauert 1997). In addition to perceiving the correct incident angle of a sound cue, directional echoes with controlled incident angles inside a concert-theatre can increase the perceived spaciousness without preventing listeners from perceiving the correct location of a sound source: theatres have been carefully designed so that echoes are coming from the appropriate incident angles. The result is that listeners can "hear" that the theatre is spacious as well as can "hear" where the stage is (Ando 1985). In other words, listeners of music played inside a concert-theatre can perceive the spaciousness of the theatre because of directional echoes. Such spatial information is missing when listeners are listening to music played through a pair of earphones. With earphones, sound is delivered directly to the openings of ear canals without the appropriate acoustic interactions with the pinna. The lack of pinna interactions greatly reduces the frontal / backward and up / down acoustic cue information normally embedded in a binaural sound. Consequently, listeners often perceive that sound transmitted via earphones is originated inside their heads (Blauert 1997). This can significantly reduce listening pleasure especially when the content is expected to contain surround sound information (e.g., live recordings from a concert). Although echoes are sometimes added to the music, these echoes lack the appropriate directional information. In summary, listeners of music played on a stage inside a concert theatre can perceive both the correct incident angles of the source of music as well as the spaciousness of the theatre. However, when listening to music played via a pair of earphones, the directional information (especially the frontal / backward and up / down cues) are missing from the music and the perceived acoustical spaciousness is reduced to inside the listeners' heads. Simulating the spatial aspects of sound: An expensive personalized solution and opportunities for mass customization Head-related transfer function (HRTF) filtering technology can simulate the modulating effects of acoustics interactions between an incident sound and a particular pinna. In other words, using the HRTF technology, it is possible to simulate acoustic cues with correct front / back and up / down directional information. The HRTF filtering technology, originally studied by Blauert

998


(1969/70), was later developed by United States Air Force to simulate directional audio warning cues for pilots of fighter jets. Studies have shown that while personalized HRTF solutions can produce directional sound cues within four degrees of accuracy, non-personalized (i.e., generic) solutions produce directional sound cues that carry inaccurate front and back as well as inaccurate up and down directional information (Wightman and Kistler 1989). In other words, HRTF technology can support the simulation of directional sound cues (as well as directional echoes) but the solution must be personalized. A personalized HRTF solution requires a listener to go through a series of tailor-made measurements conducted in an anechoic chamber equipped with specialized equipment. Typically, a series of personalized measurements will cost about US$2000 and take half-a-day (Blauert 1999). While personalized solutions are popular among military applications, it is not feasible for consumer product applications. In other words, it is not feasible for most users of earphones to invest US$2000 and half-aday of their time to acquire a personalized solution. In summary, personalized HRTF filtering technology can support the simulation of full-surround concert-theatre effects for music played through earphones but are too costly. Low-cost non-personalized HRTF filtering technology, on the other hand, fails to support the production of accurate directional sound and directional echoes. This suggests an opportunity for customizing low-cost non-personalized filtering technology to produce accurate directional sound and echoes for simulating concert-theatre effects. In the next section, the relevant problems associated with non-personalized HRTF technology will be reviewed and challenges to overcome these problems will be discussed. Problems and Challanges A review of literature indicates that directional sound cues produced by nonpersonalized HRTF filters are associated with problems of (i) front / back confusion errors and (ii) up / down confusion errors (Blauert 1997). This poses the first two challenges for customizing the non-personalized HRTF filters. First, the customized HRTF filters need to produce directional sound cues with low front / back confusion errors. Secondly, the customized HRTF filters need to produce directional sound cues with low up / down confusion errors. If these two challenges are solved, then the customized HRTF filters can be used to relocate the perceived origin of the music from the center of the listener’s heads to an appropriate location (e.g., that of a stage) related to the listeners. Also, the correct simulation of directional echoes can increase the perceived spaciousness of the music. Besides these two challenges, a third challenge is to minimize the


999

complexity of the customized HRTF filters while maintaining the necessary functionality. This is important if the customized HRTF filters were to find its way into MP3 players, iPods, and mobile-phones. Summary of Results Reducing front / back confusion through customizing HRTF filter responses A survey of the dimensions of 423 listeners' outer-ears has been conducted and analyzed into 4 groups (Figure 1). Two representative listeners from each group were tested in a series of sound localization experiments. Results indicated that their perceived directions of the same directional sound cues were significantly different. Individualized HRTFs were measured for these eight listeners and results of spectral analyses indicated that the differences in perceived sound localization performance were related to the spectral characteristics of their respective HRTFs (Figure 2). Further details of this study can be found in Ngan et al. (2005). Using the spectral characteristics identified in Ngan et al. (2005) and other literatures, six spectral features in a HRTF filter were determined to be responsible for producing a frontal or backward directional sound cue (So et al. 2007a). Using these six spectral features, 196 non-individualized published HRTFs (Algaz et al. 2001; Ircam and AKG acoustics 2004; Gardner and Martin 1995) were clustered into different groups. Six representative HRTF sets were determined from these groups and were used to produce choices of directional sound cues. In particular, frontal and backward binaural cues were produced using the six represented HRTF sets. When these 12 cues were presented to 15 listeners through a within-subject designed experiment, results of sound localization performance indicated that if an individual listener was given the chance to choose an appropriate directional sound cue produced from one of the six representative HRTF sets, significant reduction in front / back confusion errors can be achieved. In summary, a set of six representative choices of non-personalized HRTF filter sets have been successfully determined so that listeners can choose their "near" personalized HRTF filters without measuring their personalized HRTF sets. Further details about the experiment can be found in So et al. (2007a). At the Hong Kong University of Science and Technology, a set of Web-based auditory localization games has been developed so that gamers can choose their "near" personalized HRTF filters in a fun way.

1000


Figure 1: Samples of ears with typical differences in ear shapes (upper-left: an average ear; upperright: an ear with large normalized fossa height; bottom-left: an ear with large normalized cavum concha area; and bottom-right: an ear with wide normalized helix width) (adapted from Ngan et al., 2005).

Figure 2: HRTF spectra of four listeners with different types of ear dimensions. The HRTFs were measured for sound cues at 00 azimuth 00 elevation (i.e., directly in the front). As the sound cues were directly in the front, HRTFs for the left and right ears were similar (adapted from Ngan et al., 2005).

Reducing up / down confusion through customizing HRTF filter responses Hebrank and Wright (1974) reported that reducing the spectral energies of incident sound between 6kHz and 11kHz significantly affected its perceived elevation angles (i.e., up / down angles). They further proposed a "delay-and-add" theory to explain the formation of notches in the sound’s spectra due to the


1001

reflections of elevated incident sound off various reflecting surfaces located at one’s concha (Figure 3). Raykar and Duraiswami (2005) followed Hebrank’s work and reported that the formation of notches can be traced back to reflection points in the concha area of an ear. Based upon the "delay-and-add" theory, a series of studies have been conducted to investigate the relationships among reflection surfaces in the concha, spectral notches in a HRTF, and the up / down accuracy of the associated HRTF-produced directional sound cues. A survey of concha dimensions had been conducted and the "delay-and-add" theory was used to predict the spectral profiles between 6kHz and 11KHz. Results of sound localization experiments indicate that for an individual listener, the predicted spectral profiles based upon the measured dimensions of his or her concha can be used to select appropriate customized HRTF filters to generate accurate up / down directional sound cues. In particular, Au (2007) reported a novel matching index to predict the level of appropriateness between a particular non-personalized HRTF filter and an individual with a particular pinna shapes. Au successfully demonstrated that the perceived accuracies of HRTF filter sound cues for a particular listener significantly correlate with his proposed matching index. This is an important finding as it suggests that this matching index can be a useful tool for an individual listener to customize / select the most suitable non-personalized HRTF filtered sound cues.

Figure 3: A photo of an outer-ear (pinna). The opening of the ear-canal and the reflecting surfaces in the concha region are highlighted. Au (2007) used these reflecting surfaces dimensions to derive a matching method for listeners to choose the most suitable non-personalized HRTF filtered sound cues.

Optimizing the complexity of HRTF filter responses A series of studies have been conducted to optimize the complexity of HRTF filters while keeping the accuracy of the corresponding directional sound cues.

1002


Results indicated that the number of filtering coefficients can be reduced from 128 to 64 or even 32 without significantly reducing the front-back confusion errors of the corresponding directional sound cues. This is an important finding because it opens up the possibility of designing low-complexity (hence, low-cost) surround sound solutions. A prototype has been developed to simulate concert theater effects on earphone signals. A double blinded experiment has been conducted to compare the surround sound effects of this prototype and the DolbyTM stereo sound. Results indicate that the simulated concert theater binaural sound received significantly better surround sound quality ratings than the corresponded DolbyTM stereo sound (So et al. 2006). The sound quality was so impressive that a local DVD microprocessor manufacturer purchased the intellectual property (IP) rights of an associated algorithm. Final Remarks Head-related transfer function (HRTF) filtering technology enables the development of three-dimensional surround sound applications. The personalized nature of HRTF filtering technology opens up ample opportunities for mass-customized solutions. This paper summarizes the findings of a series of studies conducted towards the development of mass-customized virtual surround sound solutions for earphone listeners. Academically, knowledge obtained will further the understanding of how the individual variations in HRTFs are related to their ability to produce accurate directional sound cues. Also, original data concerning the interand intra-listener variations in the dimensions of outer-ear shapes are discovered. Potential applications of the findings include digital surround sound solutions for listeners of iPods and MP3 players. Acknowledgments The work was jointly supported by the University Grants Council as well as the Research Grants Council of the HKSAR Government (HKUST 6219/02E) and its counter-part in Germany, DAAD, (G-HK99/00.EG03, G_HK002/00E). The authors would like to thank Mr. Leung Ngan Ming and Mr. Brain Ngan for their MPhil work contribution and Prof. Jens Blauert for his advices. References Au, J. (2007). Optimizing the accuracies of up / down HRTF-filtered binaural cues. Unpublished MPhil thesis, Hong Kong University of Science and Technology. Ando, Y. (1985). Concert Hall Acoustics. Springer.


1003

Algazi, V.R., Duda, R.O., et al. (2001). The CIPIC HRTF database. Proceedings of IEEE ASSP workshop, October, New Paltz, NY: 111-123. Blauert, J. (1969/70). Sound localization in the median plane, Acustica. 22: 205–13. Blauert, J. (1997). Spatial hearing: the psychophysics of human sound localization. MIT Press. Blauert, J. (1999). Personal communications. Hebrank, J. and Wright, D. (1974). Spectral cues used in the localization of sound sources on the median plane. J. Acoust. Soc. Am. 56: 1829–34. Ircam (2004). recherche.ircam.fr/. Langendijk, E.H.A. and Bronkhorst, A.W. (2002). Contribution of spectral cues to human sound localization. J. Acoust. Soc. Am. 112: 1583–96. Ngan, B., So, R.H.Y., Leung, A., Altinsoy, E. and Avalos, A.G. (2005). Towards mass-customizing nonindividualized head-related transfer function (HRTFs). Annual Journal of IIE(HK). 25: 14–25. Raykar, V.C. and Duraiswami, R. (2005). Extracting the frequencies of the pinna spectral notches measured head related impulse responses. J. A.Soc.Am. 118: 364–74. So, R.H.Y., Leung, N.M., Braasch, J. and Leung, K.L. (2006). Towards a low-cost, non-individualized, head-related transfer functions-based surround sound system: An ergonomics study and prototype development. Applied Ergonomics. 37: 695–707. So, R.H.Y., Leung, N.M., Braasch, J. and Leung, K.L. (2006). A low cost, Non-individualized surround sound system based upon head related transfer functions. An Ergonomics study and prototype development. Applied Ergonomics. 37: 695–707. So, R.H.Y., Ngan, B., Leung, K.L., Braasch, J. and Blauert, J. (2008a). Optimizing the accuracies of generic binaural directional cues simulating frontward and backward directions: cluster analyses and an experimental study. Working paper. So, R.H.Y., Ngan, B., Leung, K.L., Braasch, J. and Blauert, J. (2008b). Effects of manipulating the spectra of non-individualized head-related transfer functions (HRTFs). Working paper. Wightman, F.L. and Kistler, D.J. (1989b). Headphone simulation of free-field listening II: Psychophysical validation. J. Acoust. Soc. Am. 85: 868–878.

Author Biographies Richard H.Y. So is Associate Professor of Ergonomics Engineering and Head of Computational Ergonomics Research Group in the Department of Industrial Engineering and Logistics Management at the Hong Kong University of Science & Technology. His research interests include computational model of human spatial audio and spatial vision processes. Prof. So is a Council member of the Hong Kong Ergonomics Society, and is a registered member of the Ergonomics Society and the Human Factors and Ergonomics Society. Contact: www-ieem.ust.hk/dfaculty/so | [email protected] Mr. John Au is a MPhil student studying at the Department of Industrial Engineering and Logistics Management, Hong Kong University of Science & Technology. His research concerns 3D spatial audio technology.

1004


Dr. K. L. Leung is an Associate Professor of the Department of Health Technology and Informatics of The Hong Kong Polytechnic University, Hong Kong Special Administration Region, PRC. He is a certified prosthetist-orthotist as well as a rehabilitation engineer. His research focuses on biomedical engineering for mobility rehabilitation including prosthetics, orthotics, foot biomechanics, body support interface and technology for enhancing human performance. Dr. Leung is currently the Award Coordinator of the MSc Degree in Health Technology (Biomedical Engineering) of The Hong Kong Polytechnic University. Contact: [email protected]

4.5

Simulation Models to Demonstrate Mass Customization Strategies Fazleena Bardurdeen Department of Mechanical Engineering, University of Kentucky, USA Haritha Metta Department of Mechanical Engineering, University of Kentucky, USA Brandon Stump Department of Mechanical Engineering, University of Kentucky, USA

Mass customization of products and services has been on the rise over the past several years across different industries. The increased interest in mass customization and personalization is evidenced by the growth in research in the area, in scholarly publications as well as the emergence of new courses on the subject. Introducing courses on mass customization and advancing knowledge on the subject are essential to continue research to make mass customization a sustainable strategy. The effectiveness of such courses can be enhanced to a great extent by incorporating practical demonstrations to provide students an opportunity for active/experiential learning. With this form of learning students take a participatory role rather than merely sitting to listen to lectures and often work together in teams to reflect upon the material taught in the classroom. It has been found that people often learn better through this approach. Therefore, innovative teaching aids that involve experiential learning, such as physical models and hands-on simulations can promote learning of and interest in mass customization. This paper presents a simple but versatile simulation that can be used in classroom environments, or even in technical fairs/exhibitions to help students (and visitors) understand the concept of mass customization and challenges to implementing the strategy.

Introduction Mass customization of products and services has been on the rise over the past several years across different industries. Many companies are now pursuing the strategy as a source of competitive advantage against others who still offer mere product differentiation. Successful mass customization manufacturing requires designing and deploying systems that have the flexibility to accommodate the diversity in product requirements as well as the dynamic demand while keeping costs low and delivery times short (Badurdeen and Masel 2007). Not only are 1005

1006


manufacturing processes new but mass customization also demands a change from the mass production paradigm in organizational structures and processes (Pine 1993) to transition from the conventional design-make-sell philosophy to one with customer involvement in value creation to a design-sell-make strategy. The increased interest in mass customization and personalization (MCP) is evidenced by the growth in research in the area, (for instance the attendance for MCP conferences has been on the rise), growth in scholarly publications in MCP as well as the emergence of new courses on the subject and related topics in business and engineering curricula. Sustaining continued interest to conduct more research and develop better approaches to successful mass customization requires better understanding of the concept and more effective methods to disseminate existing knowledge. Research shows that people’s learning styles differ significantly e.g. four types based on the Myers-Briggs Type Indicator (McCaulley 1990) as well as Kolb’s learning styles (Harb et al. 1993) and five types based on the Felder-Silverman Learning Style Model (Felder 1996). Most often the didactic teaching styles using lectures only are favored by only a small percentage of the student population. Therefore, the teaching of scientific phenomena as well as engineering applications can be enhanced to a great extent by incorporating practical demonstrations to provide students an opportunity for active/experiential learning (Bransford et al. 2000; McNertney and Garnett 2006). With active/experiential learning students take a participatory role rather than merely sitting to listen to lectures and often work together in teams to reflect upon the material taught in the classroom (Candido et al. 2007). Thus, this form of learning translates experiences into knowledge. Their benefit is attributed to addressing "cognitive and affective learning issues and in facilitating interactivity, collaboration, peer learning and active learning" (Lean et al. 2006). The advantages offered by teaching materials that require active participation to make learning more enjoyable and enhance knowledge retention have been emphasized by many (Feinstein 2001; Lean et al. 2006; Verma 2006). Therefore, innovative teaching aids that involve experiential learning, such as physical models and hands-on simulations can promote learning of and interest in mass customization particularly in classroom environments. As reported by Verma (2006), the Encyclopedia of Educational Technology states that "simulation-based learning involves the placement of a student in to a realistic scenario or situation. The student is then responsible for any changes that occur as a result of their decisions". These educational simulations can facilitate increased awareness and understanding or know-how of the environment being simulated (Riis 1995).


1007

This paper presents a simple but versatile simulation that can be used in classroom environments, or even in technical fairs/exhibitions to help students (and visitors) understand the concept of mass customization. The simulation aims to,

Contrast mass customization manufacturing and conventional make-toforecast manufacturing

Demonstrate customer involvement and co-design in mass customization

Demonstrate the use of product configurators for the co-design process

Illustrate different types of mass customization strategies

Exemplify the challenges to implementing mass customization, particularly with a focus on managing manufacturing-related issues such as inventory control, production planning and control etc. The paper is organized such that the next section provides a description of the simulations including the mass customization strategies demonstrated, the product used and the manufacturing system design. An outline of the product configurator developed for the simulations follows. A description of the simulation runs and the learning outcomes are presented together with conclusions and future work.

Description of the Simulations The various types of mass customization strategies used to design the simulations, the simple product used, the manufacturing systems involved and the product configurator developed are all presented in this section. Mass customization strategies Various frameworks have been presented to contrast mass customization with conventional manufacturing and classify different mass customization strategies. One of the clearest is the classification based on the extent of customer involvement in a product’s value chain as put forward by Lampel and Mintzberg (1996) presented in Figure 1. The strategies vary from pure standardization where the customer has no opportunity to customize his/her product to pure customization where a product/service offering is developed from scratch, with customer involvement beginning at the design phase to create a unique offering for each customer. As a result, the entire value chain of design, fabrication, assembly and distribution are affected and a unique product is produced. Other strategies lie along the continuum between pure standardization and pure customization with varying degrees of customization.

1008


Figure 1: Mass customization strategies (Lampel and Mintzberg, 1996).

Pure standardization refers to conventional make-to-stock production with no customer involvement. Nevertheless, product variety to meet different kinds of customer demand is possible, although it typically depends on a forecast rather than real-time customer input. Lean manufacturing, through the application of flow and pull production techniques, is the best known strategy to improve quality, reduce cost and delivery lead time for pure standardization. With pure standardization, when demand is accurately forecast, load leveling can be done using the principle of heijunka. The leveled production schedule can then be released to the shop floor. With this system, the assembly process acts as the pacemaker, while prefabricated components are delivered to assembly in a Just-inTime (JIT) manner based on the production schedule. A pull system with capped WIP can be implemented once appropriate buffer sizes are determined. Kanban can be used to manage and control the flow of material between processes and increase visibility across the system. Customized standardization (or some times known as standardized customization) refers to a mass customization strategy in which customer involvement takes place in the assembly stage of the value chain, and is often characterized by heavy usage of product modularity. A good example of this strategy is Dell Computers [though only about 30% of Dell’s computers are custom made (Gray 2007)]. Dell’s mass customization strategy allows the consumer to select a standard base product, and


1009

then customize it by choosing various options to add/remove to/from the system such as memory, DVD/CD recordable drives, etc. through an interactive process. With customized standardization, relatively high variety can be achieved through the different combinations in which a consumer can select the options to configure the product. A product configuration interface is needed to enable the customers select from the options available. Here, the number of part options can be low in comparison to the number of possible variants, and thus the parts inventory required is manageable. However, in contrast to pure standardization based on forecasts, such products are not assembled until the consumer configures the product through the co-design process. When such assemble-to-order customization is practiced by a company, particularly in conjunction with a larger proportion of the output made-to-forecast, such as with Dell or many automobile manufacturers, lean manufacturing principles and practices can be applied very effectively. If the percentage of mass customized products (through customized standardization) is low, variability in operations is relatively low which permits effective production mix and volume leveling. Thus, combining a set of make-to-forecast and make-to-order products for a certain period of time allows for production leveling and order release to final assembly just as in pure standardization. However, as the percentage of assemble-to-order products increase, the variability increases and managing process becomes more complex. Tailored standardization presents a greater challenge to manufacturing because customer involvement comes in the fabrication stage. Stand-alone or user-assisted product configurators, more sophisticated than those used in customized standardization, play a vital role in the co-design of products to meet individual requirements. Here, the variety of raw materials and components that are used to create the finished goods can be much higher than with customized standardization, thus creating a potentially infinite variety of end products for the consumer. Because forecasting customer demand is impossible due to high variety, components and part options cannot be pre-fabricated because of the risk of high WIP levels. Therefore orders must be launched at the fabrication stage, which becomes the pacemaker in the manufacturing system. The use of lean manufacturing principles and practices become much more difficult with this strategy because of the high variability that is present so early in the value chain. Cycle times depend on product specifications and are difficult to predict, and thus concepts of flow, pull and JIT are difficult to apply. However, other lean manufacturing tools such as 5S and visual controls, organizational learning, team work and team member empowerment are essential to achieve low

1010


cost and short delivery with mass customization particularly as customer involvement moves up the value chain. Based on these approaches to MC, three runs of the simulation are created to demonstrate pure standardization, customized standardization and tailored customization strategies. Pure standardization through the application of lean manufacturing is included to better demonstrate the challenges to mass customization manufacturing. The two approaches for mass customization are chosen because they are most widely used by companies in many industries and lend themselves to being simulated quite easily. Further, these strategies will also help demonstrate the increase in the challenges to mass customization manufacturing as the extent of customization increases and customer involvement begins earlier in the value chain. Product description A four-module Origami paper product which we call the Star Badge is used for the simulations. The supplies required to make the Star Badges are different types of paper, stickers for labels, pins to attach the badges and a paper cutter or scissors. The bill of materials (BOM) for the Star Badge is shown in Figure 2. The production steps involved in making the Star Badge are illustrated in Figure 3. The amount of choices available for the consumer is used as a means to create the pure standardization and the two different mass customization scenarios described above.

Star Badge

Subassembly

Module 1

Pin

Label

Module 2

Module 3

Customized Components

Figure 2: Bill of materials for star badge.

Module 4

1011


(a) Raw Materials (modules, label, pin)

(d) Assembled Star Badge

(b) Four modules fabricated

(c) Two sets of modules assembled

(e) Finished product with label

Figure 3: Steps involved in the production of star badges.

For the purpose of the simulations, cutting the selected paper, i.e. raw material, to the required size and folding the modules of the Star Badge following Origami techniques (particular design chosen from Boursin 2005) is considered fabrication. Customization is provided by varying the size of the badge, paper color/patterns and the label that adorns the middle. Thus, Star Badges are made in three different sizes with the module size varying from 3"x3", 4"x4" to 5"x5" squares. For each of the four modules of the Star Badge, a selection of 8 different paper colors/patterns is available. The customer also has the option of selecting from four different labels to adorn the badge; UK (stands for University of Kentucky), UK wildcat, Catspaw, and four different colored papers. This gives rise to a total of 27,216 Star Badge variants. The use of Star Badge variants (from those discussed above) and other features for the three simulation runs presented in this paper are summarized in Table 1.

1012

HANDBOOK OF RESEARCH IN MASS CUSTOMIZATION & PERSONALIZATION Table 1: Product details for simulation runs.

Simulation Run Product Variant Description Other Details 1. Pure Only 4"x4" Star Badges, 4 types of paper, 2 types of labels No customer interaction Standardization Variants are predefined as follows: Star Badge variants have No product configurator all modules of same color or opposite modules of same color Total product variants = 20 2. Pure + Only 4"x4" Star Badges, 4 types of paper, 2 types of labels Standardized ~ 70% of the demand is make-to-stock with the same Customization specifications as pure standardization ~ 30% of Star Badges are assemble-to-order. No restriction how the papers can be combined to make custom Star Badges

No customer interaction when make-to-stock. Product configurator is used in the co-design of assemble-to-order Star Badges

3. Tailored 3 sizes of Star Badges, 8 types of paper, 7 types of labels Customization Variants available are as follows: No restriction how the papers can be combined to make custom Star Badges. 100% custom orders. Materials for all Star Badges is processed (cut and folded) and product assembled after receiving customer order.

Product configurator used for co-design process.

Manufacturing system design The production process is organized into three work stations. The first work station (Station 1: Paper cutting) cuts the paper to the required size. A paper cutter with an in-built scale and color-coded templates to indicate paper sizes (3'x3", 4'x4" and 5'x5") is used as equipment at this station. The paper modules are folded to the required shape in the second work station (shown in Figure 3(b)). All the modules are assembled and the label & pin are attached at the third work station (steps shown in Figure 3 (c), (d) and (e)). The manufacturing system so designed can be run with three operators (students in this case). Cycle times of approximately 60 seconds could be achieved for the folding operation (station 2: Folding) and assembly of the product (station 3: Assembly) after a few minutes of training. The design and operation of the manufacturing systems for pure standardization and standardized and tailored customization are described below. The first run of the simulation is aimed at demonstrating pure standardization through the application of lean manufacturing principles. The layout of the system is shown in Figure 4. All orders are generated based on a short term forecast, load leveled and sequenced in production control based on the principles of heijunka before being released to assembly. Products are made to a takt time, with the assembly process as the pacemaker. Operations at every station in the manufacturing system could be completed with a one minute cycle time with some practice. Therefore, a takt time of 69 seconds (85% of cycle time) is used for the simulation.

1013


Forecast Orders (100%) Production Control

Leveled Production Schedule

Paper cutting

Folding

Cut Module storefront

Assembly

Folded Module storefront

Lead Time

Figure 4: System layout for pure standardization.

The folding and cutting stations build to replenish what is pulled from the storefront between itself and the downstream process. A buffer of four pieces of paper (or modules) of each type needs to be maintained to maintain flow and prevent starvation. The second run of the simulation involves a combination of strategies where 70% of the orders are made-to-forecast (pure standardization) and 30% are assemble-to-order (i.e. customized standardization). The system operation for this case outlined in Figure 5 is very similar to that with pure standardization. However, in addition to the forecast demand for the period which is communicated to production in advance, this scenario includes consumers who use the product configurator and place custom orders on an impromptu-basis. The amount of product variants available for pure and customized standardization is as described in Table 1. The assemble-to-order products are to be inserted in the leveled production schedule to practice change-to-order as far as possible thereby minimizing process disruptions. For the third run of the simulation, all orders (100%) are received after the codesign process through the product configurator. As indicated in Table 1, there are literally no restrictions on how consumers can configure their Star Badges. Therefore, orders are fabricated-to-order (tailored customization) by releasing them to paper cutting as shown in Figure 6. The lead time for processing is the time taken from paper cutting to assembly. Due to the large number of product variants feasible, no work-in-process inventory is maintained between the processes. Orders proceed through the system as they are processed.

1014


Figure 5: System layout for pure and customized standardization.

Figure 6: System layout for tailored customization.


1015

Product configurator for mass customization Customer involvement in value creation, through co-design, is essential for mass customization. In practice this is achieved by the use of a web-based or standalone interface that customers can navigate on their own or with the assistance of a dealer/salesman, respectively to configure the product. Therefore, to demonstrate the customer involvement in mass customization through the simulations, a simple product configurator is designed for use with the simulations. This is briefly described below. The product configurator for the Star Badge was developed using Visual C++.Net software. The different options for the customizable features of the Star Badge as well as compatibility considerations are incorporated in the product configurator. With this model the customers are able to select from the various options, including the size of the badge in the case of tailored customization. Finally a BOM is presented to the consumer for verification and acceptance. Once approved, the BOM is sent as the production order and a copy is given to the consumer. Several interfaces of the Star Badge configurator are shown in Figure 7.

Figure 7: Some interfaces of the star badge configurator.

1016


Running the simulations A description of how the simulations are performed is briefly described below. For each round of the simulations students play a specific role such as operators, production control or customers. Simulation run 1: Pure standardization The simulation is run for eleven minutes and 30 seconds which allows for an even number of 10 products to be produced based on the takt times established. Demand is created for a mix of 10 products (from the possible variants as shown in Table 1). Three students are needed to operate the manufacturing system and one person acts as production controller for this run. Before running the simulation, the production controller (with the assistance of the group of students) sequences the products when releasing to the pacemaker process (Assembly). Once the simulation begins, data is collected on the actual delivery/completion times (on time vs. late), their conformance to quality specifications (correct color modules, label and if product well assembled and for symmetry), amount of scrap and WIP in the system. Simulation run 2: Pure + customized standardization For this round of the simulation students are given the same set of orders as before at the beginning of the process. These account for the approximately 70% of make-to-stock Star Badges. Five more custom orders (~ 30%) are generated by asking five students in the class to place custom orders through the product configurator, at random, within the first 15 minutes. A person to assist customers (students) with the co-design process using the product configurator is necessary for this simulation run. When running the simulation, production controller sequences the pure standardization orders, about which information is available, and releases to production. As custom orders are placed for assemble-to-order (customized standardization), the production controller reviews sequenced orders to verify if any of them can be changed to accommodate the custom order. If so, the first such opportunity is used to insert the custom order in the sequence. If not a new order is inserted in the sequence but there is a time delay and the new order can be inserted only after the third product in the sequence. The students document the completion time for orders (including start—at product configurator—and end times for custom orders) as well as the other details as in the first run. The simulation is run for approximately 18 minutes (~ 69 sec x 15) during which time 15 products must be processed.


1017

Simulation run 3: Tailored customization This run involves the custom fabrication of all orders after the customer has decided what he/she wants. All orders are placed through the product configurator after the co-design process and the BOMs are released to the paper cutting station (pacemaker for this run). To reflect the random pattern of order receiving, arrival times are generated in advance and provided to the class. Since only a limited time is available to run the simulation, these times are generated such that there will be multiple orders (or batches of customers) arriving together (or very close to each other) as well as to reflect that there could be spans of time during which there are very few orders. If there are multiple orders arriving very close to each other, the production controller attempts to sequence them to reduce changeover. Else they are directly released to production (i.e. paper cutting) as the BOM is released after the product configuration process. The random arrivals of orders, with a large influx of requests during some intervals and sparse orders during other times, is created to help students appreciate the challenges involved in balancing the work load in manufacturing and managing inventory to meet demand fluctuations in mass customization environments. The simulation is run for 20 minutes (more time must be allowed to illustrate the dynamic customer arrivals and the impact of that on production) and data is collected as for the other simulation runs. Learning outcomes from the simulations The total time for all three runs of the simulations is about 1.5 hours. At the end of the simulations the data collected is reviewed and observations about system operations are discussed within the group. The increase in the lead times for mass customization, compared to pure standardization, helps understand how the variability in demand affects production. Comparison of performance with simulation run 1 against 2 and 3 helps demonstrate the increase in challenges faced as customer involvement moves to earlier in the value chain. The difficulties encountered with maintaining consistent quality with mass customization manufacturing, as diversity of products increase making the use of standard specifications less meaningful, is also appreciated better after the experience with the simulations. The use of a product configurator in the simulations helps understand that mass customization can eliminate the intermediate parties in the supply chain, such as distributors and retailers, exposing manufacturers directly to customer requirements and their variations. Conclusions With the design of this simulation a versatile teaching tool has been developed for demonstrating the practice of mass customization. Participants in the simulation

1018


are able to experience first-hand how an organization must constantly deal with turbulent market conditions and large amounts of product variation when implementing mass customization. The simulation demonstrates some key concepts associated with mass customization manufacturing such as launching jobs to the floor only after the customer places an order, the capability needed to handle products with completely different materials, characteristics, and processing needs and the need for and function of a product configurator. The Star Badge simulation requires minimal resources to develop and the use of a paper product (any Origami product that requires a few modules would work) makes for easy replication in any environment with little preparation. The simulation therefore is a time- and cost-effective method for demonstrating the principles and practices of mass customization to promote active learning. Further, it can be replicated to involve as many student groups in a classroom. To quantify the effectiveness of the simulation, a survey can be designed to measure the pre-simulation and post-simulation knowledge of students on mass customization. Having customers (students) engage in role-playing to characterize demanding and difficult customers can be used to demonstrate difficulties encountered with the product configuration and the co-design process. The Star Badge product configurator as shown does not reflect the product configured, and ordered, but shows a generic image of a Star Badge. Further improvement is possible to display the Star Badge actually configured. However, given this is a teaching tool, excessive effort on such finishing touches may not be necessary. The Star Badge mass customization simulation involves students as active participants and provides a more hands-on approach to learn about the concept of mass customization and the challenges to producing such products. References Boursin. D. (2005). Easy Origami. New York: Firefly Books. Bransford, J., Brown, A. and Cocking, R. (2000). How People Learn: Brain, Mind, Experience and School. National Academy Press: Washington. Badurdeen, F. F. and Masel, D. T. (2007). A Modular Minicell Configuration for Mass Customization Manufacturing. International Journal of Mass Customization. 2(1/2): 39–56. Candido, J. P., Murman, E. M., McManus, H. (2007). Active Learning Strategies for Teaching Lean Thinking. Proceedings of the 3rd International CDIO Conference, Cambridge, MA, June 2007. Feinstein, A. H. (2001). An Assessment of the Effectiveness of Simulation as an Instructional System. Journal of Hospitality and Tourism Research. 25(4): 421–443. Felder, R. M. (1996). Reaching the Second Tier: Learning and Teaching Styles in College Science Education. Journal of College Science Teaching. 23(5): 286–290.


1019

Gray, M. (2007). Dell’s Unique Approach to Supply Chain Management. Keynote Presentation, MCPC 2007 World Conference on Mass Customization & Personalization, October 11 – 12, Montreal, Canada. Harb, J. H., Durrant, O. S. and Terry, R. E. (1993). Use of the Kolb Learning Cycle and the 4MAT System in Engineering Education. Journal of Engineering Education. 82(2): 70–77. Lampel, J. and Mintzberg, H. (1996). Customizing Customization. Sloan Management Review. 38(1): 21– 30. Lean, J., Moizer, J., Towler, M. and Abbey, C. (2006). Simulations and Games: Use and Barriers in Higher Education. Active Learning in Higher Education. 7(3): 227–242. McCaulley, M. H. (1990). The MBTI and Individual Pathways in Engineering Design. Engineering Education. 80: 537–542. McNertney, E. M. and Garnett, R. F. (2006). Using a Simple Simulation Model to Help Students „Think like Economists“ in Intermediate Macroeconomics. Computers in Higher Education Economics Review. 18: 34–39. Pine II, J. B. (1993). Mass Customization: The New Frontier in Business Competition. Boston: Harvard Business School Press. Riis, J. O., Johansen, J. and Millelsen, H. (1995). Simulation Games and Learning in Production Management. England: Chapman and Hall, 3–25. Verma, A. K. (2006). Teaching lean manufacturing concepts using physical simulations thin engineering technology program. Paper presented at the American Society of Engineering Education Conference, Chicago, IL, June 18–21, 2006.

Author Biographies Fazleena Badurdeen is an Assistant Professor in Mechanical Engineering and also affiliated with Center for Manufacturing at the University of Kentucky. She has a Ph.D. in Integrated Engineering and M.S. in Industrial Engineering from Ohio University, USA. She also holds an MBA from the Postgraduate Institute of Management, Sri Lanka. Her interests are in manufacturing systems design and optimization with emphasis on mass customization, sustainable manufacturing and sustainable supply chains. Contact: www.engr.uky.edu | [email protected] Haritha Metta is a PhD candidate in Mechanical Engineering at the University of Kentucky. Her research focuses on coordinated sustainable product and supply chain design and optimization. She is the student chair of the Society of Manufacturing Engineers, University of Kentucky student chapter (2008-2009) and a member of the Delta Epsilon Iota and Golden key International Honor Societies. Contact: [email protected] Brandon Stump is obtained his MS in Mechanical Engineering with a focus in manufacturing systems at the University of Kentucky. His research was focused on investigating the possibilities of applying lean manufacturing as well as other manufacturing strategies in various types of mass customization environments in order to improve system performance. He is currently attached to Toyota Boshoku America, Erlanger, KY, USA.


5

From Mass Customization to Open Innovation

1021


5.1

User Innovation and European Manufacturing Industries: Scenarios, Roadmaps and Policy Recommendations Philine Warnke Fraunhofer Institute Systems and Innovation Research ISI, Germany Karl-Heinz Leitner Department of Technology Policy, Austrian Research Centers, Austria François Jégou Strategic Design Scenarios, Belgium Wolfram Rhomberg Department of Technology Policy, Austrian Research Centers, Austria

Recently the emergence of more open innovation models which draw on a greater diversity of distributed knowledge sources often including users and customers has received growing attention not only in academia and industry but also in policy circles. Many governments have launched initiatives to explore how to benefit from these developments and how to support companies in their adoption (e.g. the EU lead market initiative, Danish User Innovation Lab duci.dk/). The rationales for these activities are manifold. First of all, policy makers are recognising the growing relevance of open innovation models which is being driven by changing socio-economic framework conditions on the one hand and availability of enabling technologies such as innovation interfaces, connecting platforms and rapid manufacturing technologies on the other (v. Hippel 2005). Accordingly, policy makers strive to enable companies to unlock the potential benefits by adopting concepts of user involvement such as the lead user strategy (Lüthje and Herstatt 2004). At the same time it has been shown that the adoption of democratized innovation models is likely to yield substantial benefits for welfare (Henkel and v. Hippel 2005). Additionally, the empowerment of innovating users is responding to a recognised societal demand with a high potential to increase quality of life in many domains where the uptake of user centered innovation models will better match the high diversity of user needs and the growing demand for creative experience. Finally, for industrial policy there is a very concrete motivation behind the interest in such innovation models. In the face of increasing relocation of manufacturing activities to low wage production sites, concepts of production and consumption patterns that place large part of the value chain close to the customer such as distributed production in mini factories are becoming increasingly attractive to keep jobs and access to high quality products within the country. In many high wage locations where whole sectors have been disappearing,

1023

1024


policy initiatives towards personalized production and customer integration are motivated by this goal (e.g. for US and Canadian furniture industry cf. Lihra et al. MCPC 2007). To sum up, there are many good reasons for policy makers to support the transition towards democratized innovation models within economy and society. However, to achieve this goal tailored and efficient policy actions aligning research and innovation policy with measures from other realms such as IPR and regulation are needed (Chesbrough 2006, v. Hippel 2006).

Introduction The emergence of open innovation models which draw on a greater diversity of distributed knowledge sources has received growing attention in industry, academia and policy circles. In particular the integration of users into various elements of the innovation process has been explored as a promising way of underpinning competitiveness within increasingly diverse global markets. A number of governments have launched studies and initiatives to explore how to benefit from these developments and how to support companies in their adoption. Examples are the EU lead market initiative which explicitly draws on insights from user innovation studies emphasised in the recommendations of a EU high level expert group on (EU 2005), the OECD project on "Globalization and Open innovation" the Dutch advisory report on open innovation (Awt 2006) and the Danish Innovation Lab (e.g. the Danish User Innovation Lab duci.dk/). The rationales for these activities are manifold: First of all, policy makers are recognizing the growing relevance of user centric innovation models which is being driven by societal demand and availability of enabling technologies such as innovation interfaces, connecting platforms and rapid manufacturing technologies (von Hippel 2005). Accordingly, policy makers strive to enable companies to unlock potential benefits of user involvement in order to enhance competitiveness. At the same time it has been shown that the adoption of democratized innovation models is likely to yield substantial welfare benefits (Henkel and von Hippel 2003). In addition, the empowerment of users is responding to a recognized societal demand with a high potential to increase quality of life in many domains where user centered innovation models may better match the high diversity of user needs and the growing demand for creative experience. These more immediate motivations are backed up by recent insights of innovation studies emphasising the quality of interaction between users and producers as a key factor for innovation capability and thereby quality of innovation systems (Christensen and Lundvall 2004). Also the high relevance of context related knowledge of users complementing scientific and codified knowledge for the success of innovation


1025

within learning economies has increasingly been recognised (Foray 2004). More recently it was stressed that the emergence of more collective forms of user innovation is most likely to lower transaction cost and may be a means of speeding up the generalisation of local interactive learning in order to unlock its benefits for the whole innovation system (Lundvall 2006). Finally, for industrial policy there is another strong motivation for investigating into new innovation models. In the face of increasing relocation of manufacturing activities to low wage production sites, production and consumption paradigms that place large part of the value chain close to the customer such as distributed production in mini-factories (Piller et al. 2003) are becoming increasingly attractive to keep highly qualified jobs and access to high quality products within the country (e.g. Lihra et al. 2007 for the case of US Furniture Industry). Thus, there are many reasons for policy makers to explore transition towards democratized and open innovation models within economy and society. At the same time the emergence of user driven innovation in manufacturing challenges industry- and technology policy in various ways. Due to the scope and complexity of the transformation sophisticated governance with careful coordination between policy goals such as competitiveness and sustainability is required. In order to realize a customer driven innovation approach, manufacturing needs to undergo a major transition involving a number of diverse aspects. New manufacturing approaches such as "personal fabrication", "user manufacturing" or "fabbing" and "desktop manufacturing" are emerging in connection with user innovation, often involving new technologies such as advanced simulation techniques and new generations of rapid manufacturing technologies. Nevertheless, technological innovation is only one aspect of the transformation manufacturing will have to face when adopting these approaches. New patterns of using, distributing and producing goods will have to go along with the uptake of these technologies. Concepts such as "virtual co-creation platforms", "Product Configuration Toolkits" or "Product Platforms" that allow users to actively engage into product creation, call for different management approaches and organizational concepts from the manufacturer and related value adding networks. Finally, all these elements from different realms such as technology, organizational concepts and skill base need to be aligned into workable configurations. Within this transformation process there are some "classical" areas for policy activity such as legislation and regulation related to accessibility of knowledge and products for user innovation (Chesbrough 2006, v. Hippel 2006, awt 2006). However, in order to create momentum for change towards user centric innovation

1026


patterns, concerted policy measures are needed that combine activities in several policy realms. We address this issue by providing results from an international research project funded by the European Commission within the 6th Framework of Research Funding: UCIM (User centered and manufacturing) (All deliverables and outcomes are available on the project webpage: www.sustainable-everyday. net/UCIM/UCIM_site/UCIM_HOME.html). Departing from the considerations outlined above, the UCIM project aimed to explore pathways towards user centered innovation models likely to underpin competitive and sustainable manufacturing in Europe. On the base of these insights UCIM was expected to propose policy activities in support of promising options and in particular to suggest priorities for future EU research funding in the area of industrial technologies. Currently, literature dealing with user innovation hardly addresses policy issues and if so is focussing on specific questions related to either technical or organizational issues. UCIM was adopting an integrated perspective focussing on the interplay of technical, business, social and economic factors. The project comprised four main consecutive stages: First, a typology of approaches of user involvement in manufacturing industry was developed. Based on this typology, possible future situations of user involvement were drafted and discussed with experts and stakeholders from different fields. Building on these assessments more comprehensive scenarios were developed for two industries (furniture and machine tool sector). Later these scenarios were generalised into four more generic visions applying to a broader range of sectors. In a next step a roadmapping approach was used to identify crucial enabling elements for the realisation of these scenarios. Finally, UCIM developed policy recommendations suitable to foster pathways towards desirable scenarios and suggested complementary research priorities. The chapter is structured as follows. After a short introduction of the methodological framework adopted by the UCIM project, we outline the main results of the scenario building and roadmapping process and then discuss the UCIM policy recommendations. In the conclusions we propose some wider interpretations and issues for further research. UCIM Methodology In order to explore options and pathways for user centered innovation the UCIM project deployed a Foresight approach. Foresight is the set-up of a systematic futures“ dialogue among experts and stakeholders with the aim of generating forward looking intelligence on the one hand and creating common ground and


1027

momentum for change on the other (ForLearn 2006). From the policy perspective Foresight functions as a systemic innovation policy instrument directed at improving innovation capability by setting up joint learning spaces within innovation systems (Smits and Kuhlmann 2004). A number of formal methods are used in Foresight to structure the process of collective reasoning. In the case of UCIM a combination of roadmapping and scenario building was used to capture visions as well as pathways for user innovation. A roadmap provides a clear graphical representation of steps towards a particular future state. Usually a roadmap aligns information from different realms such as technological developments and market evolvement. There are a number of different types of roadmaps each serving a different purpose and structuring information in a specific way way. Some roadmaps are taking an explorative approach outlining expected developments along a timescale into the future. Others start from a desired state of the future such as a certain product to be ready developed in order to define the steps needed to get there in time (backasting). Within a Foresight process roadmapping usually serves to structure strategy oriented dialogues. While the method was originally mainly applied by companies to support strategy building, more recently a number of research institutes and think tanks have made significant efforts to adapt this methodology to the provision of intelligence to the policy making process (Da Costa et al. 2003, Könnölä T. 2007). These roadmaps aim to provide strategic intelligence needed by policy makers to optimize public R&D investments and to ensure their relevance to society. In UCIM the roadmapping was applied in a backcasting manner exploring steps needed to achieve desirable future states of user innovation – the UCIM scenarios. Scenarios are plausible, internally consistent images of the future (Schwartz 1991). Similar to roadmaps, scenarios come in many shapes and sizes. Some describe different operational environments such as socio-economic framework conditions in order to understand tehir impact on a certain organization (outsidein) others explore the consequences of different courses of action within these framework conditions (inside-out). In order for a set of scenarios to cover a wide space of relevant possible futures it needs to be developed in a systematic analysis of relevant driving factors. The UCIM scenarios did not attempt such a coverage. Rather they were picking up spots within the possibility space by illustrating distinctively different desirable and possible UCIM realisations. To sum up, in the UCIM scenarios describe where things may be heading, whereas the roadmaps outline what is needed to realise the concepts envisaged within the scenarios. Both, roadmaps and scenarios were generated in close

1028


interaction with stakeholders from industry, research and policy through interviews and interactive workshops. The futures dialogue was fed by the UCIM team with input from literature and outcomes of previous steps. UCIM did not aim to predict specific developments or points in time for the occurrence of certain states but rather to initiate creative thinking beyond established pathways. Therefore, strong emphasise was placed on providing input and outcomes of the process in a highly vivid format. In the following paragraphs we will describe the UCIM process and its results along the three basic stages: analysis, scenario building, roadmapping and development of policy recommendations. Stage 1: Analysis: The UCIM Typology Integration of customers or users into the innovation process comes in many forms and shapes. Depending on the nature of the product, company strategy and many other factors, models feature different degrees of user activity ranging from simple choice of predefined options to complete "do it yourself" with only marginal support from the manufacturer. As a base for the UCIM Foresight process the project team set out to develop a typology of approaches. Based on a literature review we analyzed various concepts such as co-creation approaches (Prahalad and Ramaswamy 2004), the lead user method (Lüthje and Herstatt 2004) customization and personalization strategies (Tseng and Piller 2003) and selfproduction concepts (Gershenfeld 2005, Neef and Burmeister 2005). A first classification was made regarding the nature of interaction between manufacturer and user/customer resulting in five different clusters (Van Zandvoort-Roelofsen and Warnke 2006). Finally, the polarity diagram displayed in Figure 1 emerged as a useful structure to organize the further UCIM futures dialogue. The diagram systematizes the user integration approaches with respect to the type of support the manufacturer provides on the one hand and the role of users on the other. As shown, the manufacturer may support the production of user driven innovations or support the creation of innovation from users or customers. The user may be involved either in the creation phase of innovation or in the production. For further investigation, the machine tool sector and the household furniture sector were chosen as pilot cases in order to capture the different situations in B2B and B2C sectors. For these two sectors the research team developed fictive future situations of user involvement across the quadrants of the polarity diagram involving experts and stakeholders from industry, policy, and academia (cf. Figure 2). These situations were generated in a creative way by blending and radicalising existing approaches suggested in the literature as well as in discussions with experts. Figure 2 shows 14 situations related to the involvement of users within


1029

the furniture sector. In the user-driven innovation quadrant we defined amongst others the "Kitchen Lab" situation, where groups of users are observed in their day-to-day practices at their home or through living laboratories in order to detect emerging demand and integrate it in the conception of new products. In the "Fab on demand" quadrant, where users are involved in creation and have manufacturer realizing it, we defined amongst others the situation "Local designer shop", where in short sessions, local designers (equivalent to hairdressers or tailors…) are helping creative people to fine tune their furniture ideas/models and have them ready to be produced. Users are involved in production process through the combination of elements preset by the manufacturer in the quadrant composition. "Furniture components mall" is here a situation where multiple modular elements from different manufacturers are combined together allows user to buy components and compose on the spot a quasi-original product. Finally, "Almost Done" is an example of a situation in the quadrant self-care which is characterised by manufacturer which supports the involvement of users in production.

Figure 1: Different types of interaction between user and manufacturer for innovation (Source: Warnke et al. 2007. All images in the UCIM project were created by Solutioning, solutioning-design.net/ )

1030


My personal chair

User involved in

Creation

Your furniture studio Local design shop

Domus worklab

Kitchen Lab

www.design-factory.com

USER -DRIVEN INNOVATION

FAB. ON DEMAND

First of series Only unique™

Manufacturer supports

Open options Lead architects group

_365 Ideas

Decorator & Furniture creation

Manufacturer supports

Production

Creation Furniture component mall

COMPOS ITION

Long Lasting Furniture

Almost done ™

SELF -CARE Infinite options Furniture Academy Corner fab workshop

User involved in

Production Figure 2: Situations of user innovation in furniture industry (Source: Warnke et al. 2007).

For the machine tool industry we developed different fictive situations (not displayed here) in a similar way. Amongst others we defined the situation "Virtual Models". In this situation the machine tool producer is offering a web platform where its clients are able to virtually build or assemble the machine they need using standard components or creating new ones. They can simulate the specific task they want to achieve or integrating it in their virtual production line. The machine tool producer can draw trends from all his clients' virtual trials. Stage 2: Exploration: The UCIM Scenarios Building on the results of the analysis, UCIM set out to develop the user involvement scenarios. As a first step, selected fictive situations were synthesised into tentative scenarios describing in more detail possible settings of user centred innovation within the selected two sectors. In this process, situations from all quadrants of the polarity diagram were taken into account. The draft scenarios were assessed, modified and further developed in an interactive scenario building workshop. Each scenario comprises a concrete realisation of user involvement with a description from the users' perspective as well as an outline of the realization from the manufacturers' side and a tentative sustainability assessment.


1031

Fictive advertising campaigns (cf. Figures 3 and 4) were used to illustrate the scenario plots. The scenarios were generated not only with a view of facilitating a maximum of user integration, they were also meant to describe solutions where production is largely based in Europe and negative impacts on the environment are avoided. In the following paragraphs we describe the basic features of the two scenarios for the furniture sector and give a short summary of the two machine tool scenarios. Furniture scenario 1: Own sweet home Own-sweet-home is based on a regional cluster of existing furniture producers using a very flexible manufacturing system and collaborating in a close network to offer a complete range of models that the customers can adapt and customize to meet their needs and tastes. Own-sweet-home from the user point of view: A family goes to local manufacturer for a consultation in which they will create and order a new bedroom for their kid. They are supported by a coach in an hour of "creative session" looking for basic models to start with, touching samples of materials, modifying the models on a computer system and finally visiting a 3D visualization of the bedroom new interior design. If they order it, Own sweet home staff will deliver and assemble the pieces at home some weeks later.

Figure 3: Simulation of an advertising campaign for scenario Own-sweet-home (Source: Warnke et al. 2007).

1032


Own-sweet-home from the business point of view: A cluster of regional furniture producers organizes itself in a partnership in order to provide a complete offer on the local market emphasizing high quality and environmental standards. The cluster of companies is constituted in order to cover most material and furniture products types involved in the household. According to the orders received, the companies connect to each other on a project base to realize it. A typical cluster relies on 3-4 regional points of sale that look like a chain of small branded showrooms and advisory desks. These clusters have a brand which is franchised at the national or European level but is based on regional clusters of existing furniture companies. The regional system organizer can be an independent new company or one existing furniture producer that takes the leadership to federate the others. Own-sweet-home from the manufacturing point of view: Solutions provided are based on existing companies that have a very flexible production able to make pieces one-by-one at medium-low cost and in full interaction one with the other. For this purpose, network of furniture manufacturing small and medium-sized firms (SMEs) together with machine tool manufacturers have developed highly agile multi-purpose manufacturing equipment that is affordable even for the smaller network members and for all classical materials of the furniture sector. A common furniture data standard is used by all actors in the regional network such as the design software, the machine tools, the ERP system, etc. Moreover, a specific software is able to support the "furniture advisor" in showing and shaping customized furniture integrating possibilities and limitation in terms of manufacturing and logistic. The same integrated software is also able to co-ordinate and streamline the production and flow of material between the different clustered companies. The basic models that are implemented in the software have been created on the base of an open design concept that enables easy change and offers many possibilities for variation around a limited number of basic models. Own-sweet-home from the sustainability point of view: Local production and coordinated logistic within the cluster as well as the home delivery allow to reduce transport and to eliminate overproduction, hence, furniture is only produced on demand. In addition, due to both personal involvement of the users and long lasting quality products a longer product lifespan has positive effect with respect to sustainability. Strong environmental standards are part of the marketing strategy. Finally, there is also less need for furniture packaging (i.e. reusable protections like movers practices).


1033

Furniture scenario 2: Rent-a-piece Rent-a-piece is a chain of leasing points renting furniture for short-medium periods of time. They ensure regular maintenance and light renewal of products along leasing as well as complete re-fabrication in the regional production plants of the brand. Rent-a-piece from the user point of view: Furniture is leased for short or medium period of time addressing the need of a more mobile population with more diverse living patterns or desire for a more frequent renewal of the household interior design. For instance, a family would lease a bed for its new born child and change model when he grows; people often moving will travel lighter not having to care for their furniture; garden furniture will be delivered in the summer period, taken back and stored in the winter; a new trend in families will be to change periodically the furniture in a continuous metamorphosis of the interior design. The user can chose and try the furniture from the park of furniture available at the showroom or leasing-points as well as order them on an on-line catalogue. The customization to the user needs exists at two different levels: Along all the leasing period, the pieces of furniture may be exchanged to fit to the evolution of the client requirements. After a period of leasing, each piece of furniture is revised and renewed. According to the state of the product, the maintenance or refabrication process needed is the occasion to customize it according to the next client expectations. Additional options of this scenario may be that home delivery is provided.

Figure 4: Simulation of an advertising campaign for scenario Rent-a-piece (Source: Warnke et al. 2007).

1034


Rent-a-piece from the business point of view: The leasing business can be developed aside a current manufacturer activity or be developed from scratch from a third party. A typical organization competing in this scenario is based on a reduced number of production plants which are able to perform heavy refabrication process and a larger number of distributed leasing points which only manage a park of pieces of furniture and perform simple maintenance operations. A range of models are available in several exemplars at each leasing point constituting a park of furniture for leasing. The production is most likely to be local in order to facilitate exchange, maintenance and re-fabrication along product lifespan. Within rent-a-piece the payment is made according to the leasing period or by subscriptions to evolving solutions over a certain period of life (e.g. a kids bedroom per year, garden furniture per season, full house furnishing per renting periods of the apartment). Extra charges are requested for anticipation of the maintenance program and specific customization requests (i.e. changing color and finishing fabrics, adding components, etc.). In addition, the service can be used as a possibility to renew and maintain pieces of furniture over a long period without exchanging it but it may also cover specific markets such as renting office furniture for companies or providing furniture for furnished apartments. Rent-a-piece from the manufacturing point of view: Furniture is specifically designed for easy maintenance and renewal (i.e. easy disassembling, renewal per parts, quick substitution of finishing elements, re-fabrication and restyling possibilities, reuse of certain components at the end of life…) and the pieces of furniture are simple and robust. Hence, they are not adaptable but can be easily exchanged according to the evolution of users needs. In order to adapt to changing leasing conditions (i.e. robust assembling and highly resistant technical parts) the furniture is made of high quality. Further, the renewal and re-fabrication of the products allow a certain level of customization (i.e. by combination of the various products components, during the renewal of used parts, etc.) within this scenario. Rent-a-piece from the sustainability point of view: Rent-a-piece is associated with a longer product lifespan through robust and long lasting furniture and different levels of renewal and re-fabrication and thus a more sustainable model. The service provided in this scenario responds to the desire of the users to renew their furnishing through leasing and periodical renewal instead of the manufacturing of new pieces. The machine tool sector scenarios In a similar way as for the furniture sector we elaborated two scenarios in more detail for the machine tool sector. They are briefly summarized here:


1035

Taking no chances: MTsystem Inc. is a machine tool producer working very closely to its clients companies organizing frequent collaborative workshop with them, sharing their virtual modeling system to customize flexible machine tools solutions and also to capture permanently clients needs and adapt their machine tools development strategy. Leasing my needs: MTservices Ltd. is an intermediate agency set up by a group of different machine tools producers to organize leasing services on the basis of the different machine tools they produced. Responding to the needs of a large pool of clients they play also the role of an observatory following evolution of the demand of user companies to orient new product development of each machine tool producer. To generate more general insights, four generic UCIM scenario types were developed. For this purpose, in a further workshop involving experts from different industrial sectors, the sector specific scenarios were assessed as to their potential for generalization and application in other sectors. Based on the outcomes of this workshop the following four generic scenarios were generated:

Generic scenario 1: MyProductValley: Individualized production in local production clusters with a joint space for interaction with the customer or networks of shops where individual data is captured. The above described scenario Own-sweet-home is an example of this type in the furniture sector.

Generic scenario 2: Create and carry: Shops where some product components are produced according to users design or manufacturing centers where products are customized through modification of existing components and addition of self produced parts and produced on the spot. Such a scenario may be realized for end consumer markets such as glasses but also for investment goods e.g. for spare part production for machinery.

Generic scenario 3: Leasing my long-term needs: Provision of individual product service systems that are adapted to customer needs over the whole life time either through exchange of product or through continuous adaptation of one long lasting product is the idea of this scenario. The scenarios Leasing By Needs in the machine tools sector or Rent-a-piece in the furniture sector are illustrations of this scenario type.

Generic scenario 4: Co-innovation: Close long-term collaboration between manufacturer and (lead) users characterised by joint workshops for generation of new products as well as proactive observation of customer needs. This is a strategy which is already realized partly in some industries but certainty has a

1036


potential for many manufacturing industries. The above sketched Taking no chances describes such a vision for the machine tool sector. Together these scenarios cover the space of the polarity diagram and the whole range of fictive situations that were assessed positively by the stakeholders and experts. Stage 3 Backcasting: UCIM Enablers and Roadmap Starting from the four generic scenarios, UCIM explored what would be needed to realize promising visions of user involvement within a wide range of sectors of European manufacturing industry (backcasting). Within two expert workshops, a set of twenty crucial enabling elements from different realms were identified and assigned to the UCIM generic scenarios (For an in depth description cf. Warnke et al. 2006.). The resulting roadmap was visualized in the form of a metro-plan as shown in Figure 5. The plan indicates that there are some basic requirements for all types of user involvement such as better understanding of users' attitudes and adequate staff skills as well as some specific enablers required by each usercentered innovation model. Enabling elements of user-centred Innovation in the manufacturing include (adapted from Warnke et al. 2006):

High Mix- Low volume manufacturing systems: Manufacturing systems which are flexible enough to produce very low batch sizes in a short time. One-off manufacturing technology: Flexible machines and processes able to efficiently process and quickly create a range of variations, even if it is just for one piece. The processes need to meet high environmental standards. Flexible finishing processes: Finishing processes that can be applied at the last moment in a high number of variations. Rapid Manufacturing technologies: Additive manufacturing processes. Special UCIM requirements: Improved speed, quality and materials variety as well as applicability in other than factory environment (shop, home) and by laypersons, environmentally benign. Management of data flows: Standards and procedures to facilitate the complex data flows within smart company networks including the customer/user. User creation interface: An easy to use design platform to enable users to define their own design freely but taking into account ergonomics, functionality as well as material and production constraints, and finally to visualize and assess the result. The use of the platform is supported by a human mediator as needed.

1037

VOLUME 2: APPLICATIONS AND CASES MyProductValley Furniture: Own-sweet-home®

Create and Carry Spare Parts: HandymanDepot Stores Glasses: OptiOptions

Smart networks high mix - low volume, manufacturing system

Network quality management

Integrated design

Shoes: Ferramo Rapid Manufacturing Technologies One-off manufacturing

Leasing my Long-term needs

Flexible finishing processes User creation Interface

Furniture: Rent-a-piece Machine Tool: MT services Ltd.

Data flow Management New materials

Adaptive product design

Smart logistics management

New business models

Usage monitoring systems

Virtual simulation Mutual Guidance Instruments

Staff skills User attitudes

Current situation

Co-Innovation Machine Tool: MTsystem Inc. Healthcare: Alzheimer patients

Figure 5: Roadmaps of UCIM enablers visualized as Metro Plan for the Furniture sector (Source: Warnke et al. 2007).

Smart collaborative networks: Company networks combining the necessary competencies to manufacture a wide range of product variants according to customers needs and offering a unified interface to the user. Underlying requirements: Organizational concepts, standards, collaborative software, collaboration skills. Adaptive design: The adaptive design concept aims to design products in a way that leaves a maximum degree of freedom for adaptations and variations for users and customers all along the product life cycle including adaptations after some time of use, re-manufacturing, second use (c.f. Lindemann and Maurer 2006). In the case of product service systems this also embraces the service component. Adaptive design needs to be facilitated by "open product standards" ensuring e.g. flexibility and compatibility of components (e.g. standardized joints). Integrated design: Integrated product and process innovation going beyond current approaches towards simultaneous engineering or concurrent engineering to allow for a quick transfer of individualized product specifications into

1038


production. Many processes will have to be designed rather for flexible multipurpose use than optimized for a specific product model. Although research into such concepts is ongoing there are still many challenges to be addressed and there is a need to gain experience in various sectors and types of products. The integrated design approach needs to be complemented by organizational measures to ensure a close co-operation and co-ordination between design and production departments. Smart logistics management: Smart logistics are a key component of the supply chain management function in order to manage successfully supply chains through complex networks and single users. Network Quality Management: Quality standards to enable distributed production of customized products as well as suitable measurement and control techniques and procedures for all network partners to meet the standards. Business models: Companies are shifting their strategy for generating value towards the involvement of users in the creation and/or the production of the products they use. To do so, they need to adopt business models that are more collaborative and allow for different value configurations such as the "outsourcing" of value adding elements to the user. User attitudes: Users need to shift from passive to active attitude towards involvement in creation and production of the products they use. Staff skills: Individual skills need to be developed within the manufacturing companies to facilitate cooperation between producers and customers, to assist customers in the customization processes, to cooperate within a network of companies, to operate agile and flexible manufacturing systems. New materials: Materials for small-batch and one-off production and rapid manufacturing processes as well as long lasting materials sustaining product adaptations over the whole lifecycle. Materials need to meet the high environmental standards. Usage monitoring systems: Concepts and technologies allowing manufacturers to observe product usage. Key issues are privacy and security to ensure an exchange of relevant information based on mutual trust. Organizational measures are needed to ensure the adequate uptake of the observation results. Virtual Simulation: To enable joint design of complex products virtual models are needed that can be used for collaborative design as well as for testing in the users' application context. Mutual Guidance instruments Concepts assisting articulation of customers needs with a long term horizon.


1039

This list of enablers comprises a range of technologies and concepts needed to realize various forms of user involvement into innovation processes in manufacturing industry. However, the UCIM project stressed that much more than the availability of individual enabling elements, intelligent integration of different enablers will be crucial to facilitate change towards user-centred innovation models. For instance, the "user-creation interface" needs to fit tightly with the personalized manufacturing system on the one hand and the user-centred business model on the other. Furthermore, as the scenarios show, a crucial element of usercentred innovation is the formation of new constellation of actors along the value chain. So for instance in the furniture sector manufacturers and retailers as well as software providers need to establish new modes of co-operation to facilitate the UCIM scenarios. The integration of users into innovation processes requires new technologies, business models and organizational structures along the entire value chain, and hence concerted process, product and organizational innovation. Stage 4 Prescription: The UCIM Recommendations The UCIM project aimed to advise policy makers in meeting the challenges of user driven innovation and in particular to propose priorities for EU research funding underpinning user-centred innovation concepts in favor of sustainable and competitive manufacturing in Europe. Therefore, following the roadmapping process, the project was exploring policy measures in support of the UCIM scenarios. Taking into account the whole range of established instruments of research and innovation policy (Andersson 1999), concrete measures such as direct R&D funding awareness raising, standardisation and public procurement were proposed for each of the enabling elements described above (Leitner et al. 2006). However, it was recommended that rather than pushing enablers individually, research and innovation policy should address groups of enabling elements in concerted measures supporting their aligned development. Building on the results of the expert workshops the following four priority areas for such integrated policy activities were proposed:

Manufacturing system for UCIM

User interface for UCIM

Adaptive product and service design

Smart and open production networks

For each of these priority areas topics for research funding and complementing policy measures were proposed and crosscutting issues to be taken into account outlined. Finally, UCIM recommended a set of overarching policy activities in

1040


order to support the uptake of user-centred innovation by a wide range of companies. These recommendations for policy measures in support of user centred innovation models include (Leitner et al. 2007):

Real life learning for UCIM: Establishment of attractive UCIM pilots to raise awareness among users and create learning spaces for companies and other actors. Two concrete experiments were suggested: the "UCIM co-creation pilot" and the "co-operative furniture showroom" Orienting research towards UCIM: Integration of users and user research into publicly funded R&D projects Make the case for UCIM: Collection and documentation of experience with UCIM applications. Establishment and promotion of success stories best practice and transfer of concepts between sectors. Open up knowledge for UCIM: Fostering of debate on IPR revisions in favour of user innovation. Align actors for UCIM: Targeted set up of user-producer-dialogues in dedicated innovation areas through local clustering and in particular Foresight initiatives. Purchasing for UCIM: Public procurement with a view to promoting user involvement using experience from "green procurement" and "procurement for innovation". Tailoring UCIM: Develop tailored UCIM strategies for companies with the aid of innovation researchers taking the Danish experience as a model. Find out more about UCIM: Launching of additional socio-economic research to explore in more detail the nature of change towards user centred innovation for various sectors, products cultural contexts as well as its social and economic impacts and the emerging requirements for innovation policy.

Conclusions and Outlook As discussed in the first section of the chapter the emergence of user centred innovation models holds many potential benefits both from a macro socioeconomic perspective and from the point of view of individual companies. The UCIM scenario exploration provided imaginative pictures of possible futures usercentred innovation landscapes which many of the actors involved assessed as desirable. The roadmapping process has highlighted how the involvement of users by manufacturing industries requires a number of diverse elements not only on the company level but within wider socio-economic framework conditions. Accordingly, for a wider transition to occur, not only companies need to embrace new approaches towards innovation. Also other actors such as regulators, associations,


1041

education bodies may facilitate changes of socio-economic framework conditions in phase with company activities. The UCIM analysis clearly indicated that change towards user-centred innovation models is inherently systemic. Many more than individual novel elements, aligned changes on various levels are needed. A change towards user centred innovation models can be interpreted as a complex multi-level and multi-actor learning process as they have been described and analyzed by scholars from an evolutionary economics perspective (Geels 2004). The UCIM policy recommendations reflect this insight: they point to the need of concerted proceeding between policy actors concerned with issues such as innovation, industry, regional development, consumer protection, education, and information society. Furthermore, UCIM emphasised more systemic modes of research funding rather than single new topics. As the UCIM project was targeting recommendations for research and innovation policy, it spelled out the consequences from a policy perspective. However, it could well be concluded that similar implications hold for strategies on a managerial level. Instruments for visioning and strategy building across actor groups such as foresight dialogue and transition management will most likely have a central role to play in both realms. For companies bridging between visions of actors with different positions in the value creation process such as product designers, marketing and maintenance will be crucial for successful integration of users into innovation processes. Approaches like the UCIM foresight dialogue using imaginative futures' images to facilitate joint visioning beyond established lines of thinking may be useful. References Andersson, T. (1999). Managing a Systems Approach to Technology and Innovation Policy, OECD STI Review 22. Awt (2006). Dutch Advisory Council for Science and Technology Policy awt: Opening up: Policy for Open innovation. Advisory report 68. Chesbrough, H. (2006). Open Innovation and Open Business Models. Presentation to Joint OECD/Dutch Ministry of economic affairs conference on globalization and open innovation December 2006. Christensen, J.L., Lundvall, B.-Å. eds. (2004). Product Innovation, Interactive Learning and Innovation Performance. Amsterdam: Elsevier. Da Costa, O., Boden, M., Punie, Y., Zappacosta, M., (2003). Science and Technology Roadmapping: from Industry to Public Policy. IPTS Report 73, April 2003. EU (2005). Creating an Innovative Europe. Report of the Independent Expert Group on R&D and Innovation appointed following the Hampton Court Summit and chaired by Mr. Esko Aho. Foray, D. (2004). The Economics of Knowledge. Cambridge MA: MIT Press. FORLEARN (2006). The For-Learn Online Foresight Guide. forlearn.jrc.es/guide/1_whyforesight/characteristics.htm.

1042


Geels, F.W. (2004). Processes and patterns in transitions and system innovations: Refining the coevolutionary multi-level perspective. Technological Forecasting & Social Change. 72(6): 681–696. Gershenfeld, N. (2005). FAB: The Coming Revolution on your Desktop: From Personal Computers to Personal Fabrication. New York: Basic Books. Henkel, J., von Hippel, E. (2005). Welfare implications of user innovation. Journal of Technology Transfer. 30(1/2): 73–87. Könnölä, T. (2007). Innovation Roadmap: Exploring Alternative Futures of Industrial Renewal. Role and Dynamics of Corporate R&D. First European Conference, IPTS Joint Research Centre of European Commission, Seville, Spain, October 8th-9th 2007. Kemp, R., Schot, J., Hoogma, R. (1998). Regime Shifts to Sustainability through Processes of Niche Formation. The Approach of Strategic Niche Management. Technology Analysis and Strategic Management. 10(2): 175–195. Lihra, T., Buehlmann, U. and Beauregard, R. (2008) Mass customization of wood furniture as a competitive strategy. International Journal of Mass Customization. 2(3/4): 200–215. Leitner, K.-H., Rhomberg, W. and Warnke, P. (2007). UCIM: Work Package 4: Deliverable 7 Policy Analysis Report. www.sustainable-everyday.net/UCIM/UCIM_site/UCIM_HOME.html. Lindemann, U. and Maurer, M. (2006). Early evaluation of product properties for individualized products. Int. J. Mass Customization. 1(2/3): 299–314. Lüthje, C. and Herstatt, C (2004). The Lead User method: an outline of empirical findings and issues for future research. R&D Management. 34(5): 553–568. Lundvall, B.-Å. (2006). Interactive learning, Social Capital, and Economic Performance. In: Advancing Knowledge and the Knowledge Economy. Brian Kahin and Dominique Foray (eds.), Cambridge MA: MIT Press: 63–74. Neef, A., Burmeister, K. and Krempl, S. (2005). Vom Personal Computer zum Personal Fabricator Hamburg: Murmann. Reichwald, R., Piller, F., Jaeger, S. and Zanner, S. (2003). Economic Evaluation of Mini-Plants for Mass Customization. A decentralized setting of customer-centric production units. In: Tseng, Mitchell M.; Piller, Frank T. (Eds.) The Customer Centric Enterprise. Heidelberg: Springer: 51–70. Phaal, R., Farrukh, C. and Probert, D. (2004). Technology roadmapping – A planning framework for evolution and revolution. Technological Forecasting and Social Change. 71(1-2): 5–26. Tseng, M. and Piller, F. (2003). The Customer Centric Enterprise: Advances in Mass Customization and Personalization. Berlin: Springer. Prahalad, C.K. and Ramaswamy, V. (2004) The Future of Competition: Co-creating Unique Value with Customers. Boston, MA: Harvard Business School Press. Schwartz, P. (1991). The Art of the Long View. Chichester, New York, Brisbane: Doubleday. Smits, R. and Kuhlmann S., (2004). The rise of systemic instruments in innovation policy. Int. J. of Foresight and Innovation Policy. 1(1/2): 4–32. Van Zandvoort-Roelofsen, C. and Warnke P. (2006). UCIM Work Package 1: Deliverable 1. Description of clusters and concepts. von Hippel, E. (2005). Democratizing Innovation. Cambridge, MA: MIT Press. von Hippel, E. (2006). Innovation is Democratizing: What can governments do? Presentation to Joint OECD/Dutch Ministry of economic affairs conference on globalization and open innovation December. Warnke, P., Leitner, K.-H., Jegou, F. and Cahill, E. (2007). UCIM Final Roadmapping Report. Work Package 3: Deliverable 6.


1043

Author Biographies Dr. Philine Warnke is a senior researcher in the foresight team of the Fraunhofer Institute of Systems and Innovation Research (ISI) since 2008. She has an engineering background and holds a PhD in sociology of technology. From 2005-2007 she was a senior researcher at the European Commission’s Institute for Prospective Technological Studies (JRCIPTS). At IPTS she was substantially contributing to the activities of the Foresight Action such as the FORLEARN online Foresight guide and the FORLEARN mutual learning process. Before joining IPTS Philine was a researcher at the Fraunhofer ISI competence center for innovation in production. Philine’s research focus is on foresight, innovation and co-evolution of technology and society. She has carried out numerous prospective research projects as well as foresight exercises and technology assessment studies both on a national and European level. Dr. Karl-Heinz Leitner is Senior Researcher in the Department of Technology Policy at the Austrian Research Centers. He received his PhD in Economics and Social Sciences from the Vienna University, teaches Innovation Management at the Technical University of Vienna was Visiting Research Scientist at the Copenhagen Business School (Fall semester 2007). His main research interests cover innovation processes in companies, strategic management, research policy, and the valuation of intellectual capital. Amongst others he has studied the best 50 Austrian industrial innovations between 1975-2000. KarlHeinz Leitner carried out research and consultancy projects funded by companies, ministries and the European Commission and has published in R&D Management, Management Accounting Research and Creativity and Innovation Management. Contact: [email protected] François Jégou, Strategic Design Scenarios Brussels. François Jégou is a Strategic Design Consultant with a degree in industrial design and teaches as visiting professor at the Faculty of Design of the Politecnico in Milan and La Cambre School of Visual Art in Brussels. Since 1990 he has run the consultancy DALT-SDS based in Paris and Brussels, specialising in co-designing scenarios and new product-service system definition. François Jégou is active in various fields including: sustainable design, interaction design, cognitive ergonomics, senior friendly design, compliance and security of pharmaceutical products, innovation in food products. He is involved in several EU research projects. Contact: www.sustainable-everyday.net & www.solutioning-design.net | [email protected] Wolfram Rhomberg is Senior Researcher at the Department of Technology Policy at the Austrian Research Centers. He studied Social Sciences and Economics at the Vienna University and at the University of Edinburgh. He received his masters degree in Sociology in 2000. His main research interests cover innovation processes in companies with focus on empirical studies in the manufacturing sector, research policy consultancy, and the evaluation of RTD programs. Lately he has put a special focus on process innovations and user centred innovations. Wolfram Rhomberg carried out consultancy and research projects funded by Companies, Ministries and the European Commission.

5.2

Bridging the Innovation Gap: From LeadingEdge Users to Mass Market Salah Hassan George Washington University, USA Philippe Duverger George Washington University, USA

This paper presents an integrative model for innovation diffusion using a market orientation approach to radical innovation creation and market adoption intention. It is argued that lead-users possessing high opinion leadership qualities (i.e. social influence, community active, innovation/ modification sharing) will adopt radical innovations at a faster rate than lead-users with low opinion leadership propensities. This integrative model is expected to help managers in utilizing the opinion leadership qualities of select leadusers while developing and introducing new innovations. This integrated research approach is not only appropriate for the "mainstream" product development, but calls for the marketers to identify a new adopter category personified by the lead-users with high opinion leadership qualities and to consider them as a particular group of interest for future marketing research and practice. Finally, research propositions are offered based on modelling these relationships in order to drive future empirical research.

Introduction For hundreds of years, firms utilized a manufacturer-centric method of developing innovations that represented a source of high-risk due to uncertainties related to the product development and marketing processes. This led to high failure rates in the marketplace and multi-billion dollar waste of scarce economic resources. These failure rates were attributed to the difficulty in evaluating the factors associated with accelerating the rate of diffusion and the inappropriate application of innovation diffusion models. Hence, a better understanding of factors and models influencing innovation diffusion is becoming a high priority for researchers and managers, particularly those in high-tech firms. Consequently, the open innovation method emerged as an alternative offering greater advantages represented in helping firms to minimize the risk by developing new products that translate users' needs and hence there is a high probability to be accepted by the mass market.

1044


1045

Some diffusion researchers have long maintained that a particular set of factors such as evaluation of innovation attributes and opinion leadership variables were the best predictors of diffusion rates. Other researchers have championed lead user characteristics as the most effective method for the innovation development process. The introduction of innovations is influenced by a complexity of factors both controllable and uncontrollable by the firm. This chapter presents a diffusion model that synthesizes concepts from key research traditions and concludes with a discussion of managerial implications to accelerate the innovation diffusion rates and provides a future research agenda. There are a number of researchers that theoretically and empirically investigated the influence of lead users on the innovation process as they modify the existing products to be later developed by firms to become commercial products (Morrison et al. 2000). It is argued that this lead-user innovation approach helps the firm to minimize the risk associated with introducing new products to the market (Henkel and von Hippel 2005; Lüthje and Herstatt 2004). This lead-user based innovation process "offer greater advantages over the manufacturer-centric innovation development systems that have been the mainstay of commerce for hundreds of years" von Hippel (2005). As a result, the possibility of an accelerated rate of diffusion is far greater than in comparison with the traditional innovation method due to lead users' opinion leadership qualities. However, there are only limited researchers that comprehensively evaluated the influence of lead users' innovations on the rate of diffusion (Morrison et al. 2000). The purposes of this chapter are: (1) to introduce an integrative model of innovation diffusion that evaluates the lead users influence on accelerating the diffusion rate through highlighting the link with opinion leadership and (2) to discuss possible directions for future research and managerial implications for accelerating diffusion rates to bridge the innovation gap in the marketplace. Nature of Innovation Adoption Innovations are defined here as new products, services, application methods, tools, offerings, processes, organizational improvements, or value delivery systems that facilitate problem solving for potential adopters. Different adopters perceive and assess innovation in a variety of ways. Rogers (1983) suggests that analysis of innovations should be made in the context of the potential adopter’s own perspective and situation; in other words, to emphasize the subjective nature of innovations. Considerable research efforts by diffusion researchers have found that adoption decisions followed a hierarchy of effects model that led to the cognitive assess-

1046


ment of cost/benefits associated with innovations (Rogers 1962; Fliegel and Kivlin 1966; Rogers and Shoemaker 1971; Zaltman and Stiff 1973). Investigations of adoption decisions have gained broader recognition when marketing researchers became concerned about acceptance of innovations. Consequently, the new product adoption process is most often viewed as a hierarchal sequence from knowledge/ awareness and evaluation to full adoption (Robertson 1971). It is argued that communication of information about new products is essential in order to create positive perception of the benefit and favourable attitude toward the innovation being described. Traditional diffusion models (Rogers 1983) are based on the assumption that making consumers aware of innovations will produce positive attitudes which will facilitate acceptance. It is assumed that consumers will act on their perceptions, once they become aware of the desirability of adopting a particular innovation. Once the consumer becomes aware of a felt need and possesses the means to satisfy the need, he or she begins a process of innovation evaluation. In contrast, the lead-user based model of innovation diffusion presented in Figure 1 depicts that innovation evaluation occurs at two levels. The first level of evaluation is a comparison of the innovation with the set of products currently being used. This comparison is suggested by research conducted by Taylor (1977), Rosch (1978), Hirschman (1980), Cohen (1982), and von Hippel (1984). In this context, the innovation is compared with a category of products being used by the potential adopter in order to determine if it is compatible. All other factors being equal, Taylor (1977) suggests that heavy users of a product category are likely to be early adopters of compatible innovations. The second level of evaluations concerns a comparison of the innovation with analogues products. Existing analogues products are found within categories of complementary products which the consumer is presently using or has considered using. Each user is thought to "customize" this evaluation process due to personal experience, perceived pressures from cultural or reference groups, and/or differences in levels of knowledge/ awareness. These factors are depicted in Figure 1 under comparative evaluation. In this context, the integrative approach presented here focuses on "comparative evaluation," where the user evaluates the analogues products according to their perceived costs and benefits and then evaluates his/her expectations for future product innovation which will supersede even the presently introduced innovation. This innovation evaluation process was identified as discriminating attributes between adopter classes/ segments. It is observed that perceptions of innovations and evaluation of their attributes are positively related to high rates of innovation diffusion. Take for example how Apple was able to realize that in meeting consumers' expectations for download-


1047

ing music they had think beyond their own computer industry category. Apple realized that "leading edge-users" think beyond this computer category, they think in terms of music and entertainment as they download music by the track, create digital playlists, and swap playlists and tracks of music as they engage in socialization within a network of music hobbyists. In this case, Apple was able to harness the understanding of user expectations and to leverage this "lead user" innovation by creating the iPod/ iTunes as a new subcategory that represents a combination of entertainment, music sharing, and computer-based inventory of content.

Figure 1: Comparative evaluation of innovations.

Users as Source of Product Innovation Traditionally, firms utilized a classical product development method of developing new products for the marketplace (Gupta and Rogers 1991). This traditional product development method is characterized by high-risk due to the uncertainty associated with the development of new products. In addition, most of the organizations are working on line extensions and modifying the existing products rather than creating new products (von Hippel et al. 1999). It was argued that the participation of the customers in the innovation process might minimize risk (Lüthje and Herstatt 2004). That is to say that this approach helps firms to minimize the risk by developing new products that translate the customers' needs and hence there is a high probability to be accepted by the market (Henkel and von Hippel 2005).

1048


Despite these advantages, there are important limitations to customers' participation in the innovation process through the traditional marketing research methods that are based on interviews and focus groups. First, customers will provide feedback related to their personal experience without taking into consideration future trends in the market. Second, the problem solving steps are relatively complicated especially when the consumers try to describe their needs. Third, usually the marketing research methods that are used during the innovation process are based on the evaluation of users' preference and perception of different product attributes. Hence they depend on the users' previous experience and not enhance the innovation concept (von Hippel 1986). Therefore, the transfer of information from the consumers to the organization is relatively costly due to sticky information syndrome (von Hippel and Katz 2002; Franke and Piller 2003; Piller et al. 2004). In this context, stickiness of information is defined as "the incremental expenditure required for transferring a certain unit of information to specified locus in a form that is usable to the information seeker" (Piller and Walcher 2006, p.307). A number of researchers argue that most of the information that describe the consumers' needs such as taste are subjective and hence difficult to describe and translate to customize the product (von Hippel and Katz 2002; Piller et al. 2004). As a result, firms depend on their database as a source of information to develop products that will satisfy the most common needs among consumes. Based on this strategy, firms tried to satisfy the heterogeneous needs; hence a number of dissatisfied users will appear in the market (von Hippel 2005; Franke and Reisinger 2003). There are remarkable efforts done by firms in order to overcome these limitations. In addition, there are new trends in the market such as technology development, global competition, sophisticated customers, etc., that encourage firms to involve external source of information in the innovation process (Gupta and Rogers 1991). One of the methods is to increase the involvement of selected type of consumers/users in the innovation process depending on their creativity and problem solving skills. Clarified by von Hippel (2005), this method as he mentioned, "the user manufacturer categorization of relationship between innovator and innovation can be extended to specific functions, attributes, or features of products and services" (P3). Empirical researches proved that there are highly significant correlations between lead users and innovations by users in many major industries (von Hippel 2005). As a result, this open-source innovation method was adopted by several firms like 3M, Adidas, Lego, and BMW (von Hippel et al. 1999; Morrison et al. 2000). This method looks at dramatic developments in users' ability to innovate as a result of the steadily improving quality of computer software and hardware which improved access to easy-to-use tools and compo-


1049

nents for innovation access to a steadily richer innovation commons. As a consequence, innovation by users will continue to grow. Who are the Lead Users? As it was mentioned before, innovation researchers found that many important commercial products are developed by users, who are ahead of the market trends in terms of their unsatisfied needs, rather than by manufacturers (von Hippel 1999). The lead users as innovators method was empirically investigated to show how dissatisfied users were able to develop their own products that were adopted later on by firms (Franke and Piller 2003). Lilien et al. (2002) summarized the three main phases of the lead users idea generation to start with goal generation followed by pyramid networking which identifies the lead users in the target market and finally, develop several workshops for the lead users to develop the product idea. This form of consumer involvement was conceptualized by the literature as the toolkits approach that is based on transferring the product and service development tasks from the research and development department to pre-qualified lead users (von Hippel and Katz 2002; Franke and Piller 2003; Piller and Walcher 2006). One of the main advantages of the toolkits approach is that it minimizes the sticky information transfer costs since the consumers will be participating directly in most of the stages of the product development process (von Hippel and Katz 2002). In addition, this approach helps the organizations to develop new products that are accepted by the market (Henkel and von Hippel 2005). As a result, the possibility of an accelerated rate of diffusion is far greater than in comparison with the traditional innovation method (von Hippel 2005). To summarize, lead users' innovations will result in the development of innovations and new products with an expected market potential. Hence, it is more effective to involve the lead users in the early stages of the innovation process (Lüthje and Herstatt 2004). However, the challenge that is facing firms is how to identify the lead users and differentiate them from the traditional consumers. The next section will address this challenge. Why Identify Lead Users? Introducing new products to the market is associated with the high risk of failure (Lüthje and Herstatt 2004). Hence, to minimize this risk, it will be beneficial to connect the consumers with the company during the innovation process and consider them as the main source of information on new product ideas (von Hippel 1986; Lüthje and Herstatt 2004). Lead users creativity is extremely

1050


important in generating new ideas and participating in the product development process. A number of researchers studied the frequency of users' innovations have found that relatively a large number of users develop their own products (Franke and Piller 2003; von Hippel 1986; Piller and Walcher 2006) and others play a major role in developing new products based on their previous experience (Lüthje and Herstatt 2004; von Hippel 1986). Lead user creativity helps in increasing the efficiency of the innovation process as it enables manufacturers to produce products with high market potential (Henkel and von Hippel 2005). In addition, it reduces the sticky information cost since transferring the information from the consumer to the manufacture is costly (von Hippel and Katz 2002; Franke and Piller 2003). Hence their involvement should be in the early stages of the product development process (Lüthje and Herstatt 2004). These types of creative consumers are defined as being in advance of the market in terms of their needs, motivations and qualified to develop products that satisfy their needs (Urban and von Hippel 1988; Lilien et al. 2002; Lüthje and Herstatt 2004; Hippel and Katz 2002). In addition, these types of users are motivated to modify the existing products or innovate new products as they enjoy the problem solving techniques (von Hippel 2005). However, the challenge facing the companies is to identify these lead users, which represents the initial developers of new products ideas (Lüthje and Herstatt 2004; Morrison et al. 2000). The lead users might not be from the existing customers as they could be competitors' customers or even not served by any producer in this market (von Hippel 1986). As a result, it is extremely important to identify the characteristics of lead users in order to be able to evaluate them. Lead Users Attributes As mentioned earlier, lead users are ahead of the mass market in term of their ability to identify needs and create product prototypes or modify existing products to satisfy their own needs (Lüthje and Herstatt 2004; von Hippel 1986; Morrison et al. 2000). As a result, they have unique characteristics in comparison with the mainstream consumers that are summarized in Table 1. Identifying lead users is an extremely important part in the innovation process. As mentioned earlier, one of the main advantages of lead users' innovations is that there is a high rate of probability that the market will accept their newly developed product. Another advantage is that the lead users could act as opinion leaders and hence spread positive word of mouth about the product, which will positively affect the rate of diffusion (Morrison et al. 2000). Goldsmith and Witt (2003) identified opinion leadership as one of the dimensions of lead users. In addition, a number of

1051


researchers mentioned that innovation is one of the main attributes of opinion leadership (Myers and Robertson 1972; Lyons and Henderson 2005). In order to investigate the link between lead users and opinion leaders and their impact on the rate of diffusion, the next section will define opining leadership and highlight its main attributes. Table 1: Lead users attributes. Lead users attributes

Literature support

Capability New needs and benefits recognized early and they are perceived as being ahead of the trend Knowledge and brand awareness Experience Technical Skills

Lüthje and Herstatt 2004 von Hippel 1986 Morrison et al. 2000 Franke and Piller 2003 Goldsmith and Witt 2003

Motivation Intensity of dissatisfied need High levels of benefits expected

Lüthje and Herstatt 2004 Hippel 1986 Urban and von Hippel 1988 Morrison et al. 2000

Modification Sharing Networks Connections Perceived as Lead User

Morrison et al. 2000

Lead User Interpersonal Factors Search Behavior High level of involvement Value Conscious: more interest on new products rather than the price. Fashion Conscious

von Hippel 1986 Goldsmith and Witt 2003

Opinion Leaders Attributes Opinion leaders are usually among the first adopters of new products and use their word of mouth communication skills to influence the behavior of other people in terms of search, purchasing and usage of new products (Goldsmith and Witt 2003; Gupta and Rogers 1991). Generally, the influence of the opinion leaders is informal; however, they play a major role to influence the consumer decisionmaking process as they represent a reliable source of information. As a result, marketers worked to create communication channels to reach opinion leaders in

1052


order to encourage them to spread positive word of mouth (Lyons and Henderson 2005). The literature highlighted several characteristics that are used to identify opinion leadership (Lyons and Henderson 2005). These main characteristics are summarized in the Table 2. It should be noted that individuals with attributes such as higher level of exploratory behavior are more likely to be involved in and aware of new trends in the marketplace. In addition, they modify new products in order to adopt them (Lyons and Henderson 2005). This results on a higher rate of diffusion of new innovations. Table 2: Opinion leaders attributes. Opinion leaders attributes

Literature support

Knowledge Acquired more information about the product Technical Competence Experience

Goldsmith and Witt 2003 Myers and Robertson 1972 Childers 1986 Lyons and Henderson 2005 Eastman et al. 2002

Social influence Central of interpersonal communication networks Social accessibility

Myers and Robertson 1972 Goldsmith and Witt 2003

Innovativeness Earlier adopters of Innovation

Myers and Robertson 1972 Lyons and Henderson 2005 Goldsmith and Witt 2003

Opinion Leaders Interpersonal Factors Conformity to system norms Higher level of curiosity and exploratory behavior Higher level of involvement with the product category

Myers and Robertson 1972 Goldsmith and Witt 2003 Lyons and Henderson 2005

It is argued in the literature by a number of researchers such as Myers and Robertson (1972) that opinion leaders are not innovators. However, it is mentioned that there is a moderate relationship between opinion leadership and innovative behavior. On the other hand, Goldsmith and Witt (2003), proved a positive correlation between innovativeness and opinion leadership across several product categories. In addition, a number of researchers mentioned that innovation is one of the main attributes of opinion leadership (Myers and Robertson 1972; Lyons and Henderson 2005). Consumer innovators influence the consumer decision-making process through positive word of mouth. In addition, they act as


1053

a role model to be imitated (Goldsmith and Witt 2003). Hence, opinion leaders could convince other consumers to adopt new innovations. This relationship was supported by diffusion researchers, which highlight that opinion leaders influence the evaluation of new innovation and hence they affect the rate of diffusion (Gupta and Rogers 1991; Lyons and Henderson 2005). The next section emphasis this point in more details. Rate of Diffusion Diffusion can be defined as the process by which a new innovation is communicated through mass media as well as word of mouth in a specific market (Forlani and Parthasarathy 2003). Diffusion researchers studied the diffusion theory and empirically tested it across several industries and countries. There were attempts to study the time taken for an idea or a product to be adopted in the market (Bulte 2000; Gupta and Rogers 1991; Lüthje and Herstatt 2004). A number of researchers believed that products that are recently launched to the market diffuse faster in comparison with the situation several years ago (Bulte 2000). This fact is a result of emerging different players in the market. One of the main players that accelerate the diffusion process is the opinion leaders' word of mouth effect (Forlani and Parthasarathy 2003). Another main player that influences the rate of the diffusion is the lead users. It is believed that lead users tend to wide spread their innovation in different ways. The first method is through informal communication through assisting other people to solve their problems and satisfy their needs. The other method is through organized networks and communities that enhance the interactions and diffusion of new innovations. These collaborative communities "can increase the speed and effectiveness with which users and manufacturers can develop, test and diffuse their innovations" (von Hippel 2005, p.11). Moreover, from an economic welfare point of view, all the parties will benefit from the willingness of the lead users to freely share information relating to a new innovation with manufactures as well as other users (Morrison et al. 2000). This is to say, lead users benefit from other users as well who suggest modifications to the innovation and hence this results in mutual benefits (Raymond 1999). In addition, lead users will benefit from the positive word of mouth that will result in increasing the rate of diffusion of their innovations (von Hippel 2005). It was empirically proven that lead users used to freely reveal information about the products that they have developed in order to enhance their reputation. The first study was by Allen (1983) who mentioned that the lead users' innovations started in the iron industry in the eighteenth century (von Hippel 2005). In addition, other

1054


empirical researches documented the free revealing of information by lead users and its effects on new innovations. For instance, von Hippel and Finkelstein (1979) investigated the roles of lead users in developing medical equipment, and Franke and Shah (2003) also made a similar investigation for sporting equipment. Future Research Agenda Users' ability to innovate is improving radically and rapidly as a result of the steadily improving quality of computer software and hardware which improved access to easy-to-use tools and components for innovation and access to a steadily richer innovation commons. As a consequence, innovation by users will continue to grow even if the degree of heterogeneity of need and willingness to invest in obtaining a precisely right product remains constant. As we have seen, the information needed to innovate in important way is widely distributed; the traditional pattern of concentrating innovation-support resources on a few individuals is hugely inefficient (von Hippel 2005, p.13). One of the key implications of this research is to identify lead users that possess high opinion leadership qualities (i.e. strong social influence, high modification sharing capabilities) that could be used in the marketing efforts to reach non-lead users in order to influence the word-of-mouth effect. The research model presented in this paper is expected to pave the way for future empirical studies to test whether lead users with high opinion leadership qualities can positively affect non-lead users' perceptions of future innovation attribution. Finally, further studies could expand on the proposed model and link it to accelerating the rates of innovation diffusion. Future research should attempt to test this research model at the consumer as well as the aggregate level. Empirical studies should make use of the proposed conceptualizations and relationships in order to test whether characteristics of lead users positively affects innovation attribution and accelerates the diffusion rate. Managerial Implications Several management consulting firms developed open innovation applications to utilize the lead users in generating innovative product/ service ideas for leading firms. For example, the Open Innovations Dashboard (mktgSPECTRUM.com) provides a detailed and systematic process for profiling "Leading-edge Users" who have the ability to innovate and are considered as savvy users on the cuttingedge of knowledge in the marketplace. This trend represents a great opportunity for the firm to bridge the innovation gap in the marketplace and raises key


1055

strategic questions (Figure 2). For example, is this innovation gap going to be bridged by bringing mainstream and continues improvements in products and services as represented in annual modifications in features and product line extensions? Or, is this innovation gap going to be bridged by bringing breakthroughs represented in leading-edge innovations to the market? These are two different strategic questions that will face the firm as it attempts to bridge the innovation gap (Figure 2).

Figure 2: Bridging the innovation gap.

In addressing the first strategic option, the innovation management team that aims to bring incremental product modification will rely on involvement of mainstream consumers in product research and development. For example, Dell developed a website that engages mainstream consumers to share their ideas www.ideastorm.com. This new website helps Dell in encouraging mainstream consumers to share product improvement ideas and suggestions that are then screened for the development of future pproduct modifications. On the other hand, developing a strategic direction for the firm that brings leading-edge breakthroughs to the market requites a different brand management approach. An approach that focuses on identifying leading-edge consumers with high opinion leadership qualities (i.e. strong social influence, high modification sharing capabilities) that could be used in the marketing efforts to bring to the market innovations with great level of influence and high level of word-of-mouth effect.

1056


This strategic approach was used by Apple in bringing iPod and iPhone to the marketplace. In this case, Apple was able to harness "lead user" innovation in the commercialization of the iPod/iTunes combination. Apple was in the computer industry, not the music or the consumer electronics business at the time it introduced the iPod. The practices of downloading music by the track, creating digital playlists, swapping playlists and tracks of music were all pioneered and socialized by music hobbyists as leading-edge users in the entertainment business. In this case Apple leveraged this "lead user" innovation by creating the iPod/ iTunes as a new subcategory that represents a combination of entertainment, music sharing, and computer-based inventory of content. Consequently, a better understanding of the factors influencing lead-user innovation diffusion is becoming a high priority for managers, particularly those in high-tech firms. Many other examples of involving mainstream as well as leading-edge consumers are now under way. One of these examples is "Participatory Marketing", which encourages customers to help not only in co-creating innovative brands but also innovative marketing campaigns. The open innovation process of harnessing leading-edge users is at the heart of developing a more customer-centric strategy for the firm. It is more than a tactic to attract user attention. Approached in the right way, open innovation represents an opportunity for firms to start co-creating product & service brands as well as marketing campaigns with their leading-edge customers. Companies have a lot to win from leading-edge users. In this case, firms can capture cutting-edge innovative ideas and be linked with social networks of leading-edge users (online or offline) in order to further commercialize these innovations ahead of competitors in a way that will leverage enhanced brand differentiation and gain strategic market leadership position.

References Childers, T. L. (1986). Assessment of the Psychometric Properties of an Opinion Leadership Scale. Journal of Marketing Research. 23(5): 184–188. Eastman, J. K., Eastman, A. D. and Eastman, K. L. (2002). Insurance Sales Agents and the Internet: The Relationship Between Opinion Leadership, Subjective Knowledge, and Internet Attitudes. Journal of Marketing Management. 18: 259–285. Forlani, D. and Parthasarathy, M. (2003). Dynamic Market Definition: An International Marketing Perspective. International Marketing Review. 20(2): 142–160. Franke, N. and Piller, F. T. (2003). Key Research Issues in User Interaction with Configuration Toolkits in a Mass Customization System. The International Journal of Technology Management. 26(5/6): 578–599.


1057

Franke, N. and Shah, S. (2003). How Communities Support Innovative Activities: An Exploration of Assistance and Sharing Among End-Users. Research Policy. 32(1): 157–178. Goldsmith, R. E. and Witt, T. S. (2005). The Predictive Validity of an Opinion Leadership Scale. Journal of Marketing Theory and Practice (winter). 28–35. Gupta, A. K. and Rogers, E. M. (1991). Internal Marketing: Integrating R&D and Marketing Within the Organization. The Journal of Service Marketing. 5(2): 55–68. Henkel, J. and Hippel, E. Von (2005). Welfare Implications of User Innovation. Journal of Technology Transfer. 30(1/2): 73–87. Lilien, G. L., Morrison, P. D., Searls, K., Sonnack, M. and Hippel, E. Von (2002). Performance Assessment of the Lead User Idea-Generation Process for New Product Development. Management Science. 48(8): 1042–1059. Lüthje, C. and Herstatt, C. (2004). The Lead User Method: An Outline of Empirical Findings and Issues for Future Research. R&D Management. 35(5): 553–568. Lyons, B. and Henderson, K. (2005). Opinion leadership in a computer-mediated environment. Journal of Consumer Behavior. 4(5): 319–329. March, A. (1994). Usability: The New Dimension of Product Design. Harvard Business Review. 72(5): 144–149. Morrison, P. D., Roberts, J. H., Hippel, E. Von (2000). Determinants of User Innovation and Innovation Sharing in a Local Market. Management Science. 46(12): 1513–1527. Myers, J. H. and Robertson, T. S. (1972). Dimensions of Opinion Leadership. Journal of Marketing Research. 4(2): 41–46. Piller, F. T. and Walcher, D. (2006). Toolkits for Idea Competitions: A Novel Method to Integrate Users in New Product Development. R&D Management. 36(3): 307–318. Piller, F. T., Moeslein, K. and Stotko, C. (2004). Does Mass Customization Pay? An Economic Approach to Evaluate Customer Integration. Production Planning & Control. 15(4): 453–444. Takada, H. and Jain, D. Cross-National Analysis of Diffusion of Consumer Durable Goods in Pacific Rim Countries. Journal of Marketing. 55 (April): 48–54. Urban, G. L. and von Hippel, E. (1998). Lead User Analyses for the Development of New Industrial Products. Management Science. 34(5): 569–582. Van den Bulte, C. (2000). New Product Diffusion Acceleration: Measurement and Analysis. Marketing Science. 19(4): 366–380. von Hippel, E. and Katz, R. (2002). Shifting Innovation to Users via Toolkits. Management Science. 48(7): 821–833. von Hippel, E. (1986), Lead Users: A Source of Novel Product Concepts. Management Science. 32(7): 791–805 von Hippel, E. (2005), Democratizing Innovation. Cambridge, MA: MIT Press. von Hippel, E. and Finkelstein, S. N. (1979). Analysis of Innovation in Automated Clinical Chemistry Analyzers Science. Public Policy. 6(1): 24–37. von Hippel, E., Thomke, S. and Sonnack, M. (1999). Creating Breakthroughs at 3M. Harvard Business Review. 73(5): 47–56. Yeniyurt, S. and Townsend, J. D. (2003). Does Culture Explain Acceptance of New Products in a country? An Empirical Investigation. International Marketing Review. 20(4): 377–396.

1058


Author Biographies Salah S. Hassan is chairman of marketing and professor of Global Brand Management at the School of Business, The George Washington University. He helped audiences around the world understand the influence of brands and innovations on mobilizing performance and delivering extraordinary results. He received his Ph.D. in 1984 from The Ohio State University and was recognized in 2005 as "Outstanding Marketing Teacher" by the Academy of Marketing Science. His research focuses on strategic brand management, diffusion, and open/user innovation. He completed numerous consulting assignments and executive/ coaching programs for leading organizations like The World Bank, Smithsonian Institute, Porter/Novelli, US-AID, International Food Institute in Australia, Saudi Airlines, the Dubai School of Government, and the American Chamber of Commerce in Egypt. He is the founding partner of SPECTRUM Brand Strategy Group. SPECTRUM has deep experience in offering start-to-end marketing, communications, and branding solutions for clients' toughest problems. In working with clients, SPECTRUM is known for blending the creative and scientific aspects of branding to inspire actionable solutions. It addresses client needs anywhere in the world based on innovative solutions with the right tools for the right business environment. Contact: www.mktgSPECTRUM.com | [email protected] Philippe Duverger is a Research Assistant at the International Institute of Tourism Studies, School of Business, The George Washington University. Contact: [email protected]

5.3

Ordinary Users and Creativity: Fostering Radical or Incremental Innovation? Peter R. Magnusson Service Research Center, Karlstad University, Sweden Per Kristensson Service Research Center, Karlstad University, Sweden Christiane Hipp Chair of Organization, HR, and General Management, Brandenburg Technical University Cottbus, Germany

Managers aiming to utilize the potential of involving ordinary users in ideation for innovation currently receive very little guidance from the existing literature as regards how to do this in a satisfactory way. This paper aims to fill this knowledge gap by contributing towards better understanding of how users contribute towards the ideation process of technology-based services, as well as how they may satisfactorily be managed within it. This is accomplished by identifying and investigating different ideation patterns, as well as their effects on the created ideas' characteristics, in the context of mobile telephony services. The paper is based on a quasi-experimental study conducted over a period of twelve days and involving 56 ordinary users and 12 professionals as idea creators. Three different groups of users were used, as well as one reference group of professionals. The paper inductively identifies four different ideation patterns leading to different types of ideas as regards innovativeness (incremental/radical). These are further related to the existing literature. The paper concludes with managerial implications concerning how to manage this type of user involvement in order to obtain ideas that are either more incremental or more radical.

Introduction The service sector and its peculiarities concerning innovation are increasingly being pushed towards the centre of economic policy and innovation management research (Djellal and Gallouj 2001; Drejer 2004; Gershuny 1978; Miles 1994; Sundbo 1997; Tidd and Hull 2003). The roles of innovation, technology, and know-how, in the context of economic development and technological change, are of growing interest here and widely discussed on different economic levels (Hipp and Grupp 2005). In more recent decades, we have seen increasing growth in 1059

1060


technology-based services (Bitner, Brown and Meuter 2000; Dabholkar 1996; Meuter, Ostrom, Roundtree and Bitner 2000), i.e. services based on technology and normally combining hardware (technology) and software. These services are, for example, embedded in our mobile phones, radio receivers, etc. They are consumed by means of man-machine interaction. The underlying hardware is often rather standardized and fixed, while the services themselves can be implemented via software in an almost infinite range of permutations (Magnusson, In Press; Reid and de Brentani 2004). Innovation in this context is often linked to increased user-value by providing services based on an existing, or improved, technology platform. Knowledge of what brings value to the end-users – "use knowledge" – has its main locus among the end-users themselves, while knowledge of how to implement the services technically – "technological knowledge" – is possessed by the professional developers (experts) at the developing firm. "Use knowledge" and "technological knowledge" need to be combined in order to achieve innovation. Traditional marketing methods are oriented towards trying to understand "use knowledge" through different modes of interrogation or surveys. However, there are alternatives, i.e. allowing users to create the innovations themselves as well as trying to stimulate their creativity. Another method consists of searching for innovations already developed by so-called "lead users" (von Hippel 1986). Lead users are regarded to be in possession of both "use knowledge" and "technological knowledge", and – combined with a strong motivation to innovate for their own benefit – they are also regarded to be excellent innovators. For a long time, lead users have been recognized as valuable contributors of new ideas for new products (e.g. Franke and Shah 2003; von Hippel 1977, 1986). There are, however, problems which mainly involve identifying and engaging lead users. For business-to-consumer (B2C) markets, there is also the problem of knowing whether or not lead users are representative of the bulk of the future market (Mahajan, Muller and Srivastava 1990; Martinez, Polo and Flavián 1998; Moore 1991; Rogers 1962). This brings us to exploring the potential of the "ordinary users", i.e. the users who do not have any in-depth knowledge of the underlying technical systems. These are easier to identify and recruit, and can be expected to be more representative of the majority of ordinary users. Recently, it has been advocated that ordinary users can make valuable contributions to the ideation process (Kristensson and Magnusson 2005; Magnusson, Matthing and Kristensson 2003). There are, however, only a few studies investigating how different ways of involving users in the ideation process affect the quality of the created ideas. Most studies conclude that user involvement is beneficial; however, understanding the ideation process – the creation of novel ideas – on a micro level


1061

remains a rather under-investigated area (Dahl and Moreau 2002; Moreau, Dahl, Iacobucci and Anderson 2005). However, Kristensson and Magnusson (2005) found that the users“ technical knowledge of the underlying systems was negatively related to their ability to produce original, technology-based service ideas. Furthermore, based on more psychological research, it has been found that "priming" the users has a decisive effect on the creativity of the outcome. However, managers aiming to utilize the potential of user involvement in ideation currently receive very little guidance regarding how to do this in a satisfactory way. This paper aims to fill this knowledge gap by enabling better understanding of how users contribute towards the ideation process of technology-based services. This is accomplished by investigating different ideation processes as well as their effects on the created ideas' characteristics, in the context of mobile telephony services. We further investigate whether or not the chosen ideation process or pattern is affected by the way the user is involved in the ideation process. More explicitly, this enables the research questions to be formulated thus: Which different "ideation patterns" can be identified among the ordinary users involved in an ideation process? How does the ideation pattern affect the characteristics (innovativeness) of the created ideas? Is there a dependency between the strategy for involving the ordinary users in the ideation process and the ideation pattern they adopt? The remainder of this paper is arranged as follows. In the next section, the theoretical background is described. This section includes a discussion of the role played by innovation and technology in service industries and discusses the balance between incremental and radical innovation. Furthermore, there is a review of creativity and ideation research as regards the differences between experts and novices, and the various ideation patterns. The third section of the article describes the quasi-experimental methodology of the empirical study. The fourth section presents the results and analyses, including the findings relating to examining the research questions. The fifth section discusses the findings of the study, including a discussion structured around the research questions. The article ends with some managerial implications.

1062


Theoretical Background The role of innovation and technology in service industries In much of the traditional literature on innovation, services are neglected due to the low technology-intensity and the inability to develop or use technological innovations in products and processes (Pavitt 1984). More recent studies, however, underline the interdependencies between the techno-economic paradigm in manufacturing and services, in particular. Gallouj (1997) identify five different relationships between services and technologies: 1) the substitution relationship (replacing human capital with technical capital); 2) the identity relationship (consubstantiality between tools and services, e.g. electronic mail); 3) the determination relationship (technological innovation determines the emergence of new service functions); 4) the diffusion relationship (services help to diffuse technological innovations); and 5) the production relationship (services produce technological innovations). In particular, the identity relationship, the determination relationship, and the production relationship are of importance to technologybased service companies such as those offering telecommunications services. Companies producing and operating technology-based services are important agents in the development of fresh knowledge; they assist – among other things – the expansion of both use and technology knowledge since their interaction with customers and users leads to a better understanding of new services as well as the underlying technology. Due to their "boundary-spanning" role and their ability to both create new knowledge and combine existing knowledge, technology-based services play a significant role as creative knowledge-generators and brokers in the innovation systems of companies, industries, and nations (Hipp 1999). The nature of innovation in services The traditional innovation literature often classifies innovations according to their degree of innovativeness. Often, two categories are used, e.g. incremental/radical, continuous/discontinuous, etc (e.g. Crawford and Di Benedetto 2000; Tidd, Bessant and Pavitt 2005). Garcia and Calantone (2002) are of the opinion that these dichotomies are often ill defined and too binary. They have thus introduced the term "really new", which lies in between. Callahan and Lasry (2004) prefer the term "very new" instead. The rationale behind this division is that both types are necessary for companies. The incremental has a lower risk and a better short-term profitability while the radical is more risky but is aimed at the future (Tushman and O'Reilly III 1996).


1063

Sundbo (1997) discusses the difference between radical and incremental service innovation in detail, concluding that, because service innovations are easily copied, a continuous innovation process is necessary, which in turn affects the initiation of radical innovations. In order to cope with continuous innovation, it should not be the innovativeness that is prioritized, rather different innovation speeds for service innovation based on use knowledge and innovations based on the underlying technical systems. Consequently, for technology-based services, both incremental (short-term oriented) and more radical (long-term oriented) innovations should be of interest to companies. However, special attention will be necessary in order to match the interfaces between new service applications with the underlying (existing or improved) technical platforms. Creativity among experts and novices In many fields, experts have been shown to be able to recognize, store, and retrieve meaningful information faster and more qualitatively than novices (Chi, Glaser and Farr 1988). Thus, experts generally solve problems more effectively than novices due to their well-structured and easily-activated domain knowledge. However, in new product/service development, the requirement for more radically-new products may require a very broad search for solutions, i.e. a solution possibly residing outside the search space (domain) where the experts are at an advantage. If novices use a much broader and unexpected search strategy, they may come up with ideas that are considered more creative and better than those of the experts. In a study by Wiley (1998), this scenario was indeed the case and the experts were at a disadvantage when performing several creative problemsolving tasks. An important theoretical advancement regarding human reasoning capabilities concerns commonalities in how individuals function when engaged in a creative problem-solving. Commonly, prior knowledge plays an important role in structuring novel ideas when these are being generated. The greater prior knowledge one has, the less novel the created solutions will be (Marsh, Ward and Landau 1999). Of interest here is the notion that prior knowledge is activated automatically and thus requires no intention. When an individual identifies that a certain type of prior knowledge will be useable in a certain problem, then this prior knowledge will act as a stimulus which activates a set of predetermined response tendencies of which the individual will be unconscious. Unconscious responses like this are generally referred to as "priming" (Bargh and Chen 1996). More specifically, the theoretical consequences of priming have been discussed. Marsh, Landau, and Hicks (1996) discovered that ideas generated by individuals

1064


conform to examples which they had been shown before becoming engaged in creative problem-solving, i.e. participants reduced their creative performance by thinking in a convergent manner. From a theoretical point-of-view, people seem to take the path of least resistance by retrieving existing solutions or information which seems likely to immediately contribute towards a future solution (Ward 1994). One reason for mental shortcuts like this may be that creative problemsolving is considered to be a very demanding task mentally. Thus, in order to maximize creativity, it seems better not to have too much primed knowledge, or easily retrieved examples, to hand. This means that, once people have been shown examples, have acquired some type of information that may be related to the problem in question, or possess prior knowledge, they will not be able to avoid using this information or knowledge – i.e. they get stuck searching for solutions within a predefined box. Ideation patterns Goldenberg, Lehmann and Mazursky (2001) maintain that the marketing literature has paid little attention to the way ideas are generated, called "ideation patterns" in our paper; nevertheless, research has shown that ideation patterns affect the quality of the ideas. Drawing on previous research (Finke, Ward and Smith 1992; Smith, Ward and Finke 1995), Goldenberg et al. (ibid) suggest that ideas are composed of "functions" and "forms". Functions are related to consumer needs while forms are solutions to user needs. Function and form are thus related to two types of knowledge; "use knowledge" (function) and "technological knowledge" (form). Depending on the genesis of the idea, they classify according to different patterns: (1) need spotting – need identification precedes solution development (form); (2) solution spotting – the identification of a technology and the inventor’s search for meaningful user applications; (3) mental invention – represents the cases where needs and solutions are identified concurrently. In addition to these three patterns, they also name two more; market research for new products and following a trend; however, these can be regarded, in our opinion, as variants of (1) and (2) above. Methodology Research design In order to investigate the research questions, a comparative quasi-experimental design was employed which involved users during the ideation phase. Four different ideation trials (user involvement strategies) were used; three different user trials and one control trial using professional developers. The task of all the


1065

trials was to derive service ideas for an existing service platform for mobile telephony services. All the trials were conducted over a period of twelve days and the participants were equipped with mobile phones prepared with eleven sample services illustrating the potential of the available service platform. In total, 354 ideas were collected and assessed relating to three dimensions (originality, uservalue, and producibility). Participants The first trials, "professionals", consisted of 12 professional service developers all recruited from Telia Mobile (Sweden’s largest mobile telephony operator). All of them were from an R&D unit responsible for developing new non-voice mobile services, i.e. services based on SMS, WAP, GPRS, and suchlike. Their professional experience in the field varied between 1 and 10 years. The other three trials consisted of ordinary users who were students at a Swedish university and who had volunteered their services. To encourage the submission of ideas, it was also stated that an idea would remain the intellectual property of the creator of the idea. In order to make the experimental setting more stimulating, the participants were informed that an award of 80 Euros would be given in recognition of the best service idea. All the participants were students on non-technical study programs, e.g. social sciences, teacher training, business administration, etc. The main reason for choosing students was that they are one of the most frequent SMS user groups, thus representing users in general and constituting a target group of great interest to mobile service providers. The students were randomly assigned to one of the three trials. Trial two, "ordinary users" (19 people), managed idea creation by themselves, while the third trial (20 people) consulted a professional service designer, in groups of 4-5, during two controlled 1-2 hour meetings. The feedback given by the professionals was restricted to whether or not an idea was feasible; they were also allowed to tell the participants that the proposed idea already existed. This approach provided the participating users with the opportunity to learn the technical possibilities and limitations of the underlying technical systems in a more individualized way. This group will henceforth be referred to as "consulting users". The last group, "creative ordinary users" (17 people), had participated, before entering the trial, in a university course on which they practiced different creativity techniques, e.g. brainstorming, slipwriting, random input, and six thinking hats. There were no significant differences between the four trials with regard to any of the background variables – age, experience of mobile telephony, and SMS usage.

1066


Neither were there any significant differences with regard to the three personality tests: (i) the FS test, which correlates to a person’s creativity (Holmquist and Ekvall 1986); (ii) the LOT (Life Orientation Test) indicates whether a person has a positive or negative disposition, (Scheier and Carver 1985), and (iii) TR (Technology Readiness), which indicates a person’s willingness to adopt new technology (Parasuraman 2000). The groups' scores on the personality tests, including the descriptive data, are shown in Table 1. Table 1: Personal characteristics of the participants.

FS test

Pro-fessionals (N=12)

Ordinary Users (N=19)

Consulting Users (N=20)

Creative Ordinary Users (N=17)

M

5.92

6.37

4.55

5.76

SD

1.56

1.77

1.23

1.86

M

23.08

24.53

23.50

23.47

SD

3.70

4.34

3.75

5.04

M

8.00

5.79

4.65

1.88

SD

4.11

5.92

4.97

6.06

M

36.50

23.79

22.10

27.53

SD

8.13

2.18

2.02

9.07

Females

2 (17%)

4 (21%)

8 (40%)

9 (53%)

Males

10 (83%)

15 (79%)

12 (60%)

8 (47%)

M

10.42

3.60

3.88

4.32

SD

6.04

2.33

2.50

3.28

LOT

TR

Age

Gender

Mobile phone experience (years)

Dependent variables The literature mentions numerous criteria that can be used to evaluate the "quality" of an idea; however, there are no uniformly-accepted general criteria (Balachandra and Friar 1997; Cooper 1993), and it would seem that different criteria should be chosen on the basis of the context (Hart, Hultink, Tzokas and Commandeur 2003; Hauser and Zettelmeyer 1997; Tzokas, Hultink and Hart 2004). An idea can, for instance, be perceived as the "most innovative", the "easiest to use", the "cheapest to implement", or the "best fit with the current business model". Ultimately, what is deemed most important will depend on the


1067

business context; this in turn will decide whether the emphasis should be on more incremental ideas or on more radical ones. To enable comparison of the ideas' characteristics (i.e. dependent variables), without taking different business contexts into account, the present study has derived criteria that are productspecific, irrespective of context. This was achieved using a focus group of five experts from Telia Mobile who had experience of assessing mobile telephony services. Following a discussion about the intrinsic factors differentiating successful mobile telephony services from less successful ones, the participants individually and collectively suggested the following three dimensions: (i) originality, the newness of the service idea; (ii) user value, the estimated perceived value to the user; and (iii) producibility, the ease with which the service idea can be implemented, thus taking the producer’s perspective. It should be noted that the first two criteria correspond to the definition of creativity (see, for instance, Amabile 1996), supporting the validity of our three dependent variables. Procedure The experimental procedure consisted of four stages: (i) initiation; (ii) idea creation; (iii) delivery; and (iv) evaluation. Each trial was conducted over a period of twelve days. Initiation stage. During the initiation stage, the participants were tasked with creating one or more ideas for SMS-based services. The users were asked to come up with proposals for new services that would be of value to them, while the professionals were asked for proposals that would be of use to the participating users; all the groups thus had the same target group for their ideas (students at the a specific university). The participants were not organized into teams; however, they were free to collaborate if they wished. If this turned out to be the case, the names of their cocreators were noted. The ideas were expected to include at least one new service idea which utilized the existing application platform (AP), which was essentially a converter of SMS messages into http calls over the Internet; i.e. the AP enabled access to information on the Internet by sending and receiving SMSs. To give the participants a sense of how these services worked, and to provide inspiration, they were given access to a sample of about ten implemented services (sample services). They were also equipped with a mobile phone and a pre-paid card allowing them to send approximately 150 SMSs. All the participants received hands-on training in how to use the phone by means of testing the sample services.

1068


Idea-creation stage. The idea-creation stage of the study was conducted over a period of 12 days. The only trial that featured any interaction with the researchers was the "consulting ordinary", as previously described. In the other trials, the users managed the creation process without assistance. Each participant was given a diary (notebook) and instructed to document the ideas arising, as well as the activities triggering idea creation. The diary data was used in order to gain a deeper understanding of the individual idea creation process, thus making explicit the factors influencing the creation of the ideas and solutions. Delivery stage. After 12 days of idea creation, each group was assembled and the ideas were delivered, in a predefined format, together with the diaries. Additionally, all the participants were interviewed within the following two weeks, with the interviews being semi-structured in nature. The interviews were tape-recorded and transcribed. The purpose of these interviews was to track the relevant process data, e.g. the important events causing the participants come up with especially good ideas. Accordingly, much of the interview was spent on discussing how the submitted service ideas had been triggered. Evaluation stage. The Consensual Assessment Technique (CAT) (Amabile 1996) was used during the evaluation stage. Six experts, all of whom had experience of evaluating service ideas for mobile communications, independently assessed the service ideas. Three of the judges were engineers working in the R&D department of Telia Mobile. All three had each more than five years' experience of assessing mobile services; additionally, all three were engineers working in the R&D department. The other three judges' experience was a blend of technical and marketing outside Telia Mobile. To reduce the workload, each judge assessed two-thirds of the ideas. However, each idea was assessed by at least four judges. The assessments were performed during a two-day intensive workshop which was held at a resort. The ideas were ranked on a scale of one to ten for all three dimensions— originality, user value, and producibility (see previous paragraph "Dependent variables")—with a score of one representing the idea that was the least original, least valuable, and hardest to produce, and a score of ten indicating the idea that was the most original, most valuable, and easiest to produce. A test was conducted in order to calibrate the judges“ assessments. During this test, five ideas were chosen for individual assessment by the judges, followed by a discussion between the judges regarding the results. If an individual assessment was found to differ markedly from the others, this would be discussed and judgment anomalies addressed. After completion of this test, the service ideas of the participants were


1069

formally evaluated. Each assessment was made individually, and no discussions between the judges were allowed. Analysis and Results Identification of the different "ideation patterns" First, all 354 ideas were analyzed in order to identify possible ideation patterns, using an inductive approach. The purpose of the analysis was to investigate the extent to which the participants had utilized the available sample services during their ideation process, i.e. the priming effect of the sample. The input for the analysis consisted of: the written service descriptions, the notebooks (administered by the participants) wherein the origins of the ideas were documented, and, finally, the transcripts from the concluding interviews with the idea creators. These three sources jointly provided the basis for analysing the used ideation patterns. We then analyzed and categorized these and were able to identify four different typical ideation patterns: (i) "Improvement", entailing an improvement of one of the available sample services, the intention being to make a slight improvement in efficiency or to add some minor function. For instance, one of the sample services was an electronic bus timetable; several of the created ideas proposed minor improvements to this. (ii) "Context translation", ideas where it is traceable that one of the sample services acted as the trigger for proposing the same type of application in a new context, i.e. extending the application context. An example of this is the previously-mentioned bus timetable which someone proposed for conversion into a train timetable. Both the previously-described ideation patterns thus originated from one of the sample service’s functions, i.e. a type of priming. Another type of priming was also in evidence among the ideas. These were cases where the participants adopted an existing application outside the sample services (often web-based), proposing that this should be implemented on the application platform; we call this ideation pattern (iii) "application adoption". The fourth and final ideation pattern we call, (iv) "unprimed application", a novel idea which cannot be traced, either to any of the sample services or to any other pre-existing service. These ideas just seem to have "popped up" at some creative point during the trial, constituting either the solution to an encountered problem or a spotted opportunity. The distribution of ideation patterns is presented in Table 2.

1070

HANDBOOK OF RESEARCH IN MASS CUSTOMIZATION & PERSONALIZATION Table 2: Distribution of ideation patterns. Ideation pattern

Total

Improvement

Context translation

Application adoption

Unprimed application

Professionals

7 (13%)

4 (7%)

32 (58%)

12 (22%)

55

Users

46 (15%)

35 (12%)

125 (42%)

93 (31%)

299

Total

53

39

157

105

354

The ideation patterns' effect on the characteristics of the created ideas To investigate the characteristics of the idea, and whether these can be classified as more radical or rather incremental, two different indexes were computed on the basis of the three evaluated dimensions, as proposed by Magnusson (In Press). For operationalization, the following formula was used: Type of innovation (α∗Originality, β∗Producibility, γ∗User value) The coefficients α, β, and γ were determined in accordance with the type of innovation (incremental or radical). For incremental innovations, "producibility" and "user value" are most important (i.e. the service should be both easy to implement and valuable), while "originality" is (by definition) low. For more radical ideas, "originality" is the most important factor, at the cost of "producibility"; "user value" is not unimportant, but this dimension can be accorded a lower weighting (compared to an incremental innovation) because the actual user value can be rather difficult to establish during the idea stage of a new, original idea. On the basis of this rationale, the following values were assigned to the coefficients for the two types of innovation: Incremental_index 0.475*producibility + 0.475*user value + 0.05* Originality Radical_index 0.10*producibility + 0.35*user value + 0.55*originality The two indexes were used to analyze whether the different ideation patterns lead to more radical or more incremental ideas. A one-way ANOVA conducted for the

1071


four ideation patterns yields significant differences with regard to both the incremental_index (F3,350 = 13.805, p

Handbook of Research in Mass Customization and Personalization (2 Volume Set)

Handbook of Identity Theory and Research, 2-Volume Set

Encyclopedia of Polymer Research (2 Volume Set)

Accountants' Handbook, 2 Volume Set (Accountant's Handbook)

Handbook of Analytical Techniques, 2 Volume Set

Handbook of Semiconductor Technology (2 Volume Set)

Handbooks of Management Accounting Research, Two-Volume Set: Handbook of Management Accounting Research, Volume 2

Mass Customization Information Systems in Business

Advances in Semantic Media Adaptation and Personalization, Volume 2

Protein Engineering Handbook (2-Volume Set)

The Cancer Handbook, 2 Volume Set

Mass Customization: An Exploration of European Characteristics

Advances in Semantic Media Adaptation and Personalization, Volume 2

Advances in Semantic Media Adaptation and Personalization, Volume 2

Theocritus 2 Volume Set

Hydrocolloids : 2-Volume Set

Theocritus 2 Volume Set

International Handbook of Research in Arts Education 2-volume set (Springer International Handbooks of Education)

Encyclopedia of Survey Research Methods (2 Volume Set)

Handbook of Research on User Interface Design and Evaluation for Mobile Technology 2-Volume Set

Handbook of Research on User Interface Design and Evaluation for Mobile Technology (2-Volume Set)

Handbook of Neuroscience for the Behavioral Sciences, 2 Volume Set

Growing Modular: Mass Customization of Complex Products, Services and Software

Mass Customization. Ein wettbewerbsstrategisches Konzept im Informationszeitalter

4 Volume Set, Handbook of Liquid Crystals

Handbook of Mass Measurement

Immigration in U.S. History (2 volume set)

Handbook of Metalloproteins, 3 volume set

Handbook of Thin Films, Five-Volume Set

Handbook of Shock Waves, Three Volume Set

Handbook of Spectroscopy (2 Vol. Set)

Handbook of Research in Mass Customization and Personalization (2 Volume Set)

Handbook of Identity Theory and Research, 2-Volume Set

Encyclopedia of Polymer Research (2 Volume Set)

Accountants' Handbook, 2 Volume Set (Accountant's Handbook)

Handbook of Analytical Techniques, 2 Volume Set

Handbook of Semiconductor Technology (2 Volume Set)

Handbooks of Management Accounting Research, Two-Volume Set: Handbook of Management Accounting Research, Volume 2

Mass Customization Information Systems in Business

Advances in Semantic Media Adaptation and Personalization, Volume 2

Protein Engineering Handbook (2-Volume Set)

The Cancer Handbook, 2 Volume Set

Mass Customization: An Exploration of European Characteristics

Advances in Semantic Media Adaptation and Personalization, Volume 2

Advances in Semantic Media Adaptation and Personalization, Volume 2

Theocritus 2 Volume Set

Hydrocolloids : 2-Volume Set

Theocritus 2 Volume Set

International Handbook of Research in Arts Education 2-volume set (Springer International Handbooks of Education)

Encyclopedia of Survey Research Methods (2 Volume Set)

Handbook of Research on User Interface Design and Evaluation for Mobile Technology 2-Volume Set

Handbook of Research on User Interface Design and Evaluation for Mobile Technology (2-Volume Set)

Handbook of Neuroscience for the Behavioral Sciences, 2 Volume Set

Growing Modular: Mass Customization of Complex Products, Services and Software

Mass Customization. Ein wettbewerbsstrategisches Konzept im Informationszeitalter

4 Volume Set, Handbook of Liquid Crystals

Handbook of Mass Measurement

Immigration in U.S. History (2 volume set)

Handbook of Metalloproteins, 3 volume set

Handbook of Thin Films, Five-Volume Set

Handbook of Shock Waves, Three Volume Set

Handbook of Spectroscopy (2 Vol. Set)

Recommend Documents