J PROD INNOV MANAG 2005;22:303–321 r 2005 Product Development & Management Association Interfirm Modularity and Its Implications for Product Development Nancy Staudenmayer, Mary Tripsas, and Christopher L. Tucci Industries characterized by interfirm modularity, in which the component products of different firms work together to create a system, are becoming increasingly widespread. In such industries, the existence of a common architecture enables consumers to mix and match the products of different firms. Industries ranging from stereos, cameras, and bicycles to computers, printing, and wireless services are now characterized by interfirm modularity. While the increasing presence of this context has been documented, the implications for the product development process remain underdeveloped. For the present study, in-depth field-based case studies of seven firms experiencing an environment of interfirm modularity were conducted in order to deepen understanding of this important phenomenon. What unique challenges did this context pose and why? What solutions did firms experiment with, and which seemed to work? Based on an inductive process of data analysis from these case studies, three primary categories of challenges raised by this environment were identified. First, firms were frustrated at their lack of control over the definition of their own products. The set of features and functions in products were constrained to a great extent by an architecture that the firm did not control. Second, while an environment of interfirm modularity should in theory eliminate interdependencies among firms since interfaces between products are defined ex-ante, the present study found, ironically, that interdependencies were ubiquitous. Interdependencies continually emerged throughout the product development process, despite efforts to limit them. Third, firms found that the quantity and variegated nature of external relationships made their management exceedingly difficult. The sheer complexity was daunting, given both the size of the external network as well as the number of ties per external collaborator. Partners with whom control over the architecture was shared often had divergent interests—or at least not fully convergent interests. The solutions to these challenges were creative and in many cases counter to established wisdom. For instance, research has suggested many ways for a firm to influence architectural standards. While the firms in the present sample followed some of this advice, they also focused on a more neglected aspect of architecture— the compliance and testing standards that accompany modules and interfaces. By concentrating their efforts in a different area, even smaller firms in this sample were able to have some influence. Instead of focusing on the elimination of Address correspondence to: Christopher Tucci, Ecole Polytechnique Fédérale de Lausanne, EPFL-CdM-CSI, Odyssea 1.04, Station 5, CH-1015 Lausanne, Switzerland. Tel.: þ 41.21.693.0023. Email: christopher.tucci@epfl.ch. We wish to gratefully acknowledge the support and feedback we received on this article from the editor of JPIM, Abbie Griffin, two anonymous JPIM referees, Lynda Applegate, Michael Cusumano, Michael Hitt, Alan MacCormack, Spencer Palocz, and Lori Rosenkopf. We also wish to thank the Hartman Center at The Fuqua School of Business, Duke University, the Ecole Polytechnique Fédérale de Lausanne, and the Division of Research at the Harvard Business School for funding this research. 304 J PROD INNOV MANAG 2005;22:303–321 N. STAUDENMAYER, M. TRIPSAS, AND C. L. TUCCI interdependencies, it was found that firms benefited from concentrating on the management of interdependencies as they emerged. Finally, while layers of management and ‘‘bureaucracy’’ are often viewed as unproductive, these firms found that adding structure, through positions such as Relationship Manager, was highly beneficial in handling the coordination and control of a wide range of external relationships. Introduction T he importance of the product development process, its complexity, and the resulting managerial challenges have been widely acknowledged by academic scholars and practitioners (Terwiesch, Loch, and Niederkofler, 1998). The BIOGRAPHICAL SKETCHES Dr. Nancy A. Staudenmayer was assistant professor of management at Duke University’s Fuqua School of Business since 1997. She graduated magna cum laude from Wellesley College in 1986 with a B.A. degree in mathematics and economics, and she was inducted into Phi Beta Kappa. She earned her M.A. degree in applied statistics from the University of California at Berkeley and her Ph.D. from the Sloan School of Management at the Massachusetts Institute of Technology. A specialist in corporate strategy, she taught the Foundations of Strategy course to first-year students at Fuqua. Nancy passed away in November 2000. Dr. Mary Tripsas is assistant professor of business administration in the Entrepreneurial Management Unit at the Harvard Business School. She earned her Ph.D. from the Sloan School at the Massachusetts Institute of Technology, an M.B.A. from the Harvard Business School, and a B.S. from the University of Illinois at Urbana. Dr. Tripsas’ research examines how radical technological change affects organizations and overall industry evolution. She studies both innovation in established firms and entrepreneurial opportunities with her recent work focusing on the relationships among firm capabilities, managerial cognition, and inertia. She explores these issues through in-depth, longitudinal industry studies and is currently studying the emergence of digital imaging. Dr. Christopher L. Tucci is associate professor of management of technology and chair in corporate strategy & innovation at Ecole Polytechnique Fédérale de Lausanne. He received the degrees of B.S. in mathematical sciences, B.A. in music, and M.Sc. in computer science from Stanford University. He also received a master of science in technology and policy from Massachusetts Institute of Technology (MIT) and a Ph.D. in management from the Sloan School of Management at MIT. His prior work experience was as an industrial computer scientist at Ford Aerospace, where he was involved in developing Internet protocols and applying artificial intelligence tools to industrial problems. Dr. Tucci’s primary area of interest is in technological change and how waves of technological changes affect incumbent firms. He is also studying how the technological changes brought about by the popularization of the Internet affect firms in different industries. He was recently elected to the division leadership track of the Academy of Management’s Technology and Innovation Management Division. inherent nature of the process creates complexity in that it requires the simultaneous coordination of multiple functional areas within the firm and multiple constituents external to the firm. This coordination often takes place in an environment of high uncertainty, rapid change, and competitive pressure (Brown and Eisenhardt, 1998). While a number of tools for better managing the product development process have been identified, including quality function deployment (QFD) (Griffin, 1992), matrix organizations and cross-functional teams (Kahn, 1996; Larson and Gobeli, 1988; Swink, Sandvig, and Mabert, 1996), heavyweight product managers (Clark and Wheelwright, 1992), technology integration (Iansiti, 1998), standardized product platforms (Cusumano and Nobeoka, 1998; Meyers and Lehnerd, 1997), tightly integrated supplier relationships (Dyer, 1996), and close user relationships (Athaide, Meyers, and Wilemon, 1996; von Hippel, 1986), relatively little research has focused specifically on the increasingly common industry context of interfirm modularity. This article explores the implications of this context for product development, shedding light on yet another layer of complexity in the product development process. Drawing upon the work of Baldwin and Clark (2000), Langlois and Robertson (1992), and Schilling (2000), the present article defines an industry as being characterized by interfirm modularity when different firms are responsible for designing and developing the various subsystems of the industry’s products. In this context, a given firm’s product is but one component of an overall product system, and it must interconnect with components developed by a variety of other companies in order to have value. For instance, a Sanyo receiver, Sony compact disc (CD) player, and Bose speakers all work together to create a stereo system. The decentralized set of relationships among components is guided by what Baldwin and Clark (2000) called ‘‘design rules’’ that describe how different parts of the system will interact, ensuring compatibility among system modules produced by multiple firms. Interfirm modularity thus results in MODULARITY AND PRODUCT DEVELOPMENT the creation of a set of subindustries containing products that are only useful when combined with products from the other subindustries that are part of the overall product system. In this environment, users can customize their own systems, mixing and matching component modules from different firms. Some industries, such as stereo systems and bicycles, have long been characterized by interfirm modularity. However, this decentralized industry structure has become increasingly common, driven by higher levels of specialization and the movement to open interface standards that enable products from multiple firms to work together relatively seamlessly (Langlois and Robinson, 1992; Matutes and Regibeau, 1988). The shift in the computer industry from a set of vertically integrated firms to a set of horizontal subindustries including central processing units (CPUs), disk drives, monitors, printers, and software is one commonly cited example of interfirm modularity (Baldwin and Clark, 2000; Yoffie, 1997). More recently, industries ranging from televisions, digital cameras, and personal digital assistants (PDAs) to telecommunications services are all characterized by interfirm modularity requiring the coordination of multiple parties, sometimes numbering in the hundreds (Iansiti and Levien, 2004). While the increasing presence of interfirm modularity has been documented, the implications for the product development process remain underdeveloped. What unique challenges does this context pose? What solutions have firms experimented with in order to address those challenges? What solutions seem to work and under what circumstances? Using data from a set of field-based exploratory case studies at seven firms, the present study examines the implications of interfirm modularity for the product development process. This context first is characterized. Next, based on data from the present field research, the article summarizes the most challenging issues confronted by managers and useful practices for addressing those challenges across three domains: product definition, interdependency management, and relationship management. Characterizing the Environment The article begins by developing a taxonomy with which to distinguish differing types of interfirm modularity. The focus is on two dimensions that affect the product development environment: (1) the number of J PROD INNOV MANAG 2005;22:303–321 305 organizations involved in defining and/or controlling the architecture of the system; and (2) the number of firms involved in producing a ‘‘systemic’’ product, regardless of who controls the architecture. For this study’s purposes, the architecture of a system comprises three elements, based on Baldwin and Clark’s (2000, p. 78) three categories of design rules: Module partitioning: The definition of what the modules in the system will be and what function each will perform. As part of defining the modules, decisions are made about which modules are ‘‘hidden’’ and which are ‘‘visible.’’ The internal architecture of a hidden module is independent of the rest of the system, so changes within the module do not affect other modules. Visible modules, on the other hand, have interdependencies with other system modules. Interface specification: The delineation of how modules will work together. The interface specification will typically include multiple levels including, for instance, the physical connection or ‘‘pipe’’ between components, the communication protocols, and the format of the content that travels down the pipe. It is possible for the interface specification to change independently of the modules themselves changing (Henderson and Clark, 1990). System compliance and testing: The definition of standards for testing modules in the system for both performance and compatibility with other modules. Control over the architecture and its subsequent evolution can, at one extreme, be directed by a single firm, a centralized industry trade organization, or specific governmental body. At the other extreme, no single firm, institution, or coalition of firms defines the system architecture, but instead a broad range of players are involved in a highly decentralized process, with different players defining different interfaces. A single system may incorporate architectural standards developed by formal institutional standards bodies [e.g., International Telecommunications Union (ITU)], industry-sponsored standards bodies (e.g., the W3C or World Wide Web Consortium), industry consortia (e.g., the Compact Flash association for flash memory), individual firms (e.g., Adobe’s Postscript printer page description language), or communities and nonprofit foundations (e.g., Linux) (O’Mahoney, forthcoming). There also may be multiple competing and/or coexisting design rules for any given system, or even a specific interface. The direct connection between a personal computer (PC) and 306 J PROD INNOV MANAG 2005;22:303–321 N. STAUDENMAYER, M. TRIPSAS, AND C. L. TUCCI a printer can be accomplished with at least three different standard connections: a serial port, a parallel port, and a universal serial bus (USB) port (of which there are at least two different plugs). The second dimension of interest is the number of firms involved in producing a product. At the low end, a single firm controls the production of an entire system. This firm may use subcontractors or suppliers to produce specific modules, but those suppliers simply comply with a set of requirements defined by the focal firm; they do not experiment or innovate in the development of the component module. The suppliers also do not sell directly to the user—the overall product is still sold by a single firm. At the other extreme, different firms might produce and sell each of the modules of a system. Figure 1 combines these dimensions into a matrix and provides diagrams of product examples from each cell of the matrix. The boxes in each diagram represent modules in a systemic product, and the letters in each box denote the organization producing that module. Lines connecting the boxes represent interfaces between modules, and the letter above each line denotes the organization(s) that define these interfaces. Appliances Products in the ‘‘Appliance’’ quadrant of Figure 1 (Langlois and Robertson, 1992) are low on both Q QKP S S S S STU High KM QOP R JN OR N M U P J J KR K STV TUV T T SS J PQ O QOPR JKN Number of firms involved in defining the system architecture Low Partial Integration Development Web Eg. printer-camera bundle Eg. digital camera, cell phone D D AA A A A A B E C C C C G C C F Appliance Single-apex Cluster Eg. toasters, autos, photocopiers Eg: stereo system, plug-compatible mainframe system Low dimensions: the number of organizations defining the system architecture and the number of firms producing modules. They represent the base case of little or no interfirm modularity. Examples include not only appliances such as washing machines or toasters but also many other products such as photocopiers, portable cassette players (e.g., the Sony Walkman), and automobiles. In the example from Figure 1, a single firm (Firm A) decides upon a set of features and an overall architecture for the product (low on the y-axis) and also produces the entire product (low on the x-axis). Firm B produces a subcomponent but, as discussed already, simply meets the specifications supplied by Firm A. Within the industry other firms producing the same product may have entirely different architectures, but since each system is self-contained, these differences are transparent to users. Consumers may either accept or reject any of the preset packages offered, but they do not configure their own products. The overall design of ‘‘appliances’’ may still be modular even if the design is controlled by a single firm. In fact, modular design principles have been shown to create value in a number of ways. They can significantly improve the communication and coordination within single-firm product development projects (Ulrich and Eppinger, 2000), can increase the potential for reuse of components—thus improving the efficiency of product development (Garud and Kumaraswamy, 1995)—and can create option value by enabling module-level experimentation and innovation (Baldwin and Clark, 2000). High Number of firms involved in producing the “systemic” product Figure 1. The Evolution of the Boundaries of the Product Development Process (Adapted from Baldwin and Clark, 2000; Langlois and Robertson, 1992) Single-Apex Cluster In the ‘‘Single-apex cluster’’ quadrant (cf. Baldwin and Clark, 2000), an individual or small set of organizations defines the system architecture (low on the y-axis), but different firms produce the components (high on the x-axis). For instance, the interfaces between components of a stereo system are centrally agreed upon by all firms in the industry, but different firms produce the components. Sometimes a single firm dominates an industry and defines the architecture. IBM’s System 360 mainframe computer architecture was centrally controlled but allowed for other firms to produce ‘‘plug compatible’’ components such as disk drives, creating a number of subindustries (Baldwin and Clark, 2000). In the ‘‘Single-apex cluster’’ diagram of Figure 1, firm C would be the equivalent of IBM, both defining the interfaces and producing some modules. Other firms (e.g., D, E, and MODULARITY AND PRODUCT DEVELOPMENT F) also produce modules in the system and sell those products directly to users. Partial Integration In the ‘‘Partial integration’’ quadrant of Figure 1 firms have chosen to ‘‘reintegrate’’ and to produce multiple components of a system even though definition of the overall system architecture is decentralized, and multiple players could produce those different components. Firm S in the diagram is producing three modules of the system and has taken back ownership of the interface between those modules. But ‘‘S’’ has followed the decentralized design rules for other interfaces, and the other modules of the product system are made by firms T and U. An example of this type of interfirm modularity is found in the realm of digital imaging. Digital cameras have standard interfaces with PCs, printers, memory cards, and PDAs. Some firms, such as Canon, however, have created proprietary elements in the interface between their digital cameras and printers to optimize the functionality and, they hope, to gain market share. Development Web In the top right quadrant is found the most complex form of interfirm modularity. This context is called here a ‘‘Development Web’’ insofar as it evokes the image of a spider’s web or complex network of relationships. The complexity of this quadrant is best illustrated with an example. In February 2003, Motorola announced a cell phone that incorporated ‘‘the ideal features of a mobile phone with the capabilities of a PDA, digital camera, video player, MP3 player, speakerphone, advanced messaging, instant Internet access, and Bluetooth wireless technology (Motorola Incorporated, 2003). In the development of this product, Motorola clearly did not control many aspects of the cell phone’s architecture. Compatibility with multiple existing standards constrained the firm’s design process. The most obvious constraint was in the kind of cellular communications network available in various countries around the world. Cellular network standards could be established by national telephone carriers, government-sponsored standards bodies, private consortia, or dominant firms. However, the constraints on the architecture exceeded the simple choice of network. In storing a digital photo, Motorola could J PROD INNOV MANAG 2005;22:303–321 307 theoretically have developed its own proprietary file format, but then only people with the correct decoding software would have been able to look at the photo. Instead industry standard file formats were used. Likewise, Bluetooth compatibility was used for short-range identification of and communications with digital devices/appliances. Bluetooth was originally developed by Motorola’s competitor Ericsson, but since 2000 Motorola has been involved in the further evolution of Bluetooth. That involvement, however, involves coordination with at least eight other firms, including Ericsson, IBM, Intel, Nokia, Toshiba, 3COM, Agere (Lucent Technologies), and Microsoft. In fact, each ‘‘feature’’ of this new cell phone was actually a different technology that was part of a larger system, each interface controlled by parties external to Motorola or at best with Motorola’s interdependent involvement. Methods and Data To better understand the implications of each of these quadrants for the product development process, an exploratory field-based study of seven firms was conducted. This study’s description of the challenges and potential solutions associated with product development in the context of interfirm modularity is thus inductively derived, initially based on informal observation of changes in product development and subsequently supported and modified based on a set of structured interviews at these firms (Glaser and Strauss, 1967; Yin, 1984). The present authors’ collective experience in new product development (NPD), assorted media articles, cases, and anecdotal comments from industry insiders suggested significant implications for the management of product development. The goal of this research was to explore the challenges posed by this context and some potential solutions to those challenges. It is therefore important to note that the aim of this study is not to establish the validity or statistically test the generalizability of any particular approach. Nor can conclusions be drawn about product or project performance, although the data are suggestive of some outcomes. Seven firms were selected in which to conduct interviews: Red Hat, Lucent Technologies, The Gale Group, InterWorld Corporation, Adobe Systems, Motorola, and Lexar Media (see Table 1). Every firm that was approached agreed to participate in the study. Since this study sought to improve 308 J PROD INNOV MANAG 2005;22:303–321 understanding of the phenomenon, the decision was made to focus on the technology and telecommunications sectors where interfirm modularity was more pronounced. As shown in Table 1, some variability was introduced in the sample in order to refine understanding of the phenomenon (Eisenhardt, 1989; Yin, 1984). For example, The Gale Group is a major business unit of an established publishing firm and develops subscription-based information services for the World Wide Web. Red Hat coordinates development of ‘‘open source’’ software and related services and support from a single North Carolina location. Motorola is a large, established communications and electronics technology firm with a global presence. Thus, the firms varied in terms of size (number of employees), geographic location (North Carolina, East Coast, West Coast, global), type of business (telecommunications, PC operating systems, publishing), relative age (startup, spinoffs and older firms), and how much control they had over the system’s architecture. Nineteen interviews were conducted from the summer of 1999 through spring of 2000. To ensure that this study’s observations were not simply an artifact of interviews coinciding with the Internet bubble, a second set of 11 interviews was conducted during the winter and spring of 2003. In choosing the interviewees within each firm, individuals were selected from a mix of functional areas as well as multiple levels of the hierarchy. Among the titles of the study’s subjects were director of new product development, vice president of new product development, founder and chief technology officer, and director of strategic alliances. Each interview lasted more than an hour, and confidentiality was guaranteed at the subject and project level. Thus, the name of the firm is disclosed in the study’s illustrations and quotations except when such disclosure might compromise the identity of the speaker. The interviews were taped, with permission of the subjects, and were transcribed by a professional transcription service. Before conducting the interviews, profiles of each of the firms were developed, based on public data such as Annual Reports, press releases, marketing literature, trade journals, and the business press in order to have a basic understanding of each firm’s business. An interview protocol also was developed that defined a common introduction and description of the research goals and purpose and a set of specific questions. The interviews had three parts. First, interviewees were each asked to describe the process for two or three completed product development projects in an open- N. STAUDENMAYER, M. TRIPSAS, AND C. L. TUCCI ended fashion. Second, detail was gathered on how each product fit into an overall system—how it worked with products of other firms and how the interfaces with other firms’ products were defined. Third, the interviewees were probed for information about external relationships with questions such as, ‘‘What interactions did you have with external firms throughout the development process? What was the purpose, form, depth, and duration of each interaction?’’ In most cases it was possible to corroborate information about specific projects with more than one individual at each firm. Data analysis began with a simple inductive process of category creation (Glaser and Strauss, 1967; Yin, 1984). Each researcher began by searching the interview transcripts from the set of firms for which they were responsible, identifying both common words and common topics in order to highlight important issues and to classify them according to a common scheme. Some of the classifications were identified in advance based on related research (e.g., relationship management), while others emerged as a result of their ubiquity across the interviews (e.g., product definition). To improve the reliability of the study’s conceptual scheme, the interviewers subsequently traded transcripts and repeated the sorting process. The primary researchers then met and compared the results, reconciled any differences, and repeated the categorization again. Overall, the analysis process went through four iterative rounds of culling the interviews, discussion, adjustment, and refinement (Aldenderfer and Blashfield, 1984). Challenges and Potential Solutions Based on analysis of the interviews, the challenges posed by interfirm modularity and potential solutions to these challenges were classified into three domains: (1) product definition; (2) managing interdependencies; and (3) managing relationships. Key points are summarized in Table 2. Product Definition: Challenges Many studies in product development research carry an implicit assumption that individual firms maintain control over their product architecture—the basic set of components and interfaces present in the product. Constraints on product development therefore are considered primarily internal, imposed by technological barriers, market analysis, or cost/resource Suppliers Customers Durham, NC ‘‘Open source’’ software products and associated service and support 1993 Expand market share of Linux OS and increase revenues from services and support Suppliers Customers Geographic Locations Primary Products or Services Founded Strategic Challenge Examples of Key Types of External Partners Integrate recent acquisitions; manage very large software and hardware development projects Result of corporate breakup of 100year-old firm in 1996 Communications equipment (e.g., digital switches) for wired, wireless and cellular phone services New Jersey, Illinois, Denver, and several international locations Large (n 5 25,000– 35,000 employees) Small (n 5 160 employees) Size (Approximate Number of Employees) Spinoff of AT&T Bell Labs; develops real time global telecom equipment and services A software firm that introduced open source software strategy Lucent General Description Red Hat Table 1. Summary of Firm Sample Customers (librarians)individuals Content suppliers Transition from print to electronic information products 1995 in three-way merger Electronic subscription-based information services Farmington, MI Medium (n 5 1800) A major independent business unit of an established publishing firm The Gale Group Semiconductor manufacturers Camera manufacturers Managing through the uncertain evolution of new industries such as digital imaging 1996 Removable digital storage media for digital photography and consumer electronics Fremont, CA Small (n 5 200) Cirrus Logic spinoff that makes removable memory for digital photography and other markets Lexar Media Technology development partners Channel partners Managing channel partnerships to allow EDI and other kinds of electronic commerce in a variety of firms; manage growth 1995 Custom software enabling companies to engage in electronic commerce New York Small (n 5 200) Startup in electronic commerce Interworld Media/content companies Customers Suppliers Establishing their technologies as world-wide industry standards, establish strong position in wireless Internet access Adapting imaging strategy for the Internet Graphic designers 1928 Communications and computer components, cellular telephones, pagers Chicago, IL, Austin, TX, Phoenix, AZ, and several international locations Large (approximately 100,000) Established electronics company that is transitioning into consumer telecom products Motorola 1982 Electronic document formats, graphic design, and imaging software San Jose, CA Medium (n 5 2700) Established software company Adobe Systems MODULARITY AND PRODUCT DEVELOPMENT J PROD INNOV MANAG 2005;22:303–321 309 Other wireless device producers Other SW companies Standards institutions Standards institutions Digital camera and scanner producers Printer and image-setter manufacturers N. STAUDENMAYER, M. TRIPSAS, AND C. L. TUCCI Software developers Distributors Hardware producers Other firms developing software in system architecture Complementary Standards technology partners institutions PDA, cell phone manufacturers Specialized knowledge and technology suppliers International governments Open source developers Examples of Key Types of External Partners (Cont’d.). Table 1. (Cont’d.) Red Hat Lucent The Gale Group Lexar Media Interworld Adobe Systems End-user hardware companies J PROD INNOV MANAG 2005;22:303–321 Motorola 310 constraints. Much work in the realm of operations management related to product design and mechanical engineering has focused on optimizing development in this context (e.g., Nelson, Parkinson, and Papalambros, 1999; Papalambros, 1995). Given a requirements specification that outlines the desired performance and cost of the product, developers follow a process that provides the best match between technology and the requirements. For instance, Nelson, Parkinson, and Papalambros (1999) described the conditions under which it might be optimal to share a component across two designs, given that the performance of one or both designs might suffer due to the design changes. In this case, the implicit assumption is that the firm controls the interface between the component and the product for both possible products. In the presence of interfirm modularity, however, an increasing proportion of the crucial architectural decisions are imposed on companies by external firms and institutions. The system-level product architecture— the modules and interfaces—is often not within the control of the firm. Instead, the definition of the system architecture, and the associated ownership and control issues, become areas of intense negotiation. Dominant market players, formal standards bodies, and informal alliances among companies therefore have a strong influence on any given firm’s product architecture. The firms in the present sample recognized that they were operating in an environment where key architectural decisions were often out of their control. As an executive at Interworld noted, ‘‘[Our products] fit into a broader system or existing system. Whether they were payment processing systems, or order-fulfillment systems, or inventory systems, or accounting systems, they were plugged into . . . an existing infrastructure and had to connect to those existing infrastructures . . . [we control] essentially nothing of the broader system . . . we had to have adapters to SAP . . . into financial systems and for order fulfillment.’’ Given that much of the product architecture was outside of their control, firms were frustrated at their inability to provide what they considered the best product for the customer. As one executive summarized succinctly, ‘‘There are balances to what we do . . . We end up sacrificing performance to accommodate an industry standard more and more of the time.’’ Firms also found it challenging to differentiate their products given the constraints of the commonly shared interfaces imposed on them. Consumers sometimes had difficulty distinguishing products of different firms, depending upon how much MODULARITY AND PRODUCT DEVELOPMENT J PROD INNOV MANAG 2005;22:303–321 311 Table 2. Operating in an Interfirm Modular Environment: Concerns and Potential Solutions Identifieda Concerns Raised Potential Solutions Identified Product Definition Product definition and design are constrained by interfaces that 1. Attempt to influence key parts of the system architecture to your other firms or institutions control advantage Invest in firms producing complementary products Firms cannot always optimize performance of their modules Share personnel on product development teams with other firms Differentiating products is difficult given the need to Focus on ‘‘neglected’’ parts of the architecture such as testing accommodate common standards standards The location of value/intelligence in the system becomes an area of Reintegrate across modular boundaries in order to optimize performance negotiation. 2. Given an existing architecture, differentiate within ‘‘hidden’’ module subindustries Develop intellectual property Bundle value-added services Interdependency Management Even though the system architecture is modular, hidden 3. Develop fluid task structures and emphasize experimentation interdependencies remain Changes made by other firms can affect the performance of a 4. Create a modular organization within the firm that is flexible and given firm’s product accommodates change throughout the development process Changes made by other firms can occur in the middle of a given firm’s development project and impose changes in product 5. Use internal–external teams with longer lead times to plan module design upgrades Coordination of systemic change—upgrading multiple modules simultaneously—is difficult Relationship Management Decentralized multiparty networks of relationships are highly 6. Recentralize control over formal product development alliance complex and difficult to coordinate management External firms simultaneously hold both convergent and divergent interests 7. Categorize and manage relationships by tiers Firms often have multiple concurrent, yet uncoordinated, relationships with a single external firm Multiple interfaces exist at all levels of the organizational hierarchy Individuals often lack knowledge of other relationships Relationships are both formal and informal a The solutions do not correspond line by line with the concerns but rather section by section. Solutions are numbered for expositional purposes in the conclusion. unique value could be communicated. Despite efforts to clarify product differences in the UBS Flash Drive market, Lexar found that many consumers simply categorized the product as a ‘‘floppy disk replacement’’ not understanding the additional features that flash drives could offer. Product Definition: Potential Solutions An important decision for firms in this environment was which of two strategic paths they should follow: should they attempt to influence at least part of the overall architecture to their advantage, or should they simply optimize given the existing industry architecture? Firms with more market power, such as Motorola in the paging domain, attempted to influence the overall architecture, whereas smaller firms, such as Lexar Media in removable storage media, were more inclined to accept external architectures. The first response was basically an attempt to move from the ‘‘Development web’’ quadrant closer to the ‘‘Single-apex’’ quadrant by controlling key parts of the architecture. This strategy is discussed in some depth in the literature on complementary products and standards (McGahan, Vadasz, and Yoffie, 1997; Sengupta, 312 J PROD INNOV MANAG 2005;22:303–321 1998; Shapiro and Varian, 1999). Intel’s Architecture Lab encouraged other firms to develop products compatible with Intel’s products (Gawer and Cusumano, 2002), while NEC cultivated relationships with thirdparty developers in the Japanese PC industry (Methe, Toyama, Miyabe, 1997). Similarly, the firms in the present sample attempted to influence adoption of their preferred architectures. Red Hat invested in early-stage ventures that they thought would help deploy Linux, and Motorola and Adobe Systems both created corporate venture capital funds that searched out and made investments in potential complementary technologies. Lexar influenced the oft-neglected third element of architecture, system compliance and testing. Lexar was able to help establish ‘‘click to click’’ time as the appropriate speed measure for Compact Flash used in a digital camera. When their primary competitor advertised a speed advantage using a different metric, Lexar filed and won a false advertising lawsuit. Thus, even focusing on what might be considered a less critical element of architecture created value. Adobe Systems’ actions driving the genesis of desktop publishing provide a comprehensive example of how multiple mechanisms can be utilized to influence an architecture. Adobe exploited a number of external collaborative relationships to exert some control over the architecture of desktop publishing systems and to establish the Postscript printer page description language as a de facto interface standard between applications software and laser printers. Postscript technology facilitated the emergence of desktop publishing by enabling the use of typeset quality fonts and the integration of text and graphics in documents created and printed with desktop equipment. Previously, professional publishers and printers were needed to generate such documents. In promoting Postscript, Adobe engaged producers of each complementary module: Apple (which provided the Mac computer and Laserwriter printer), Aldus (which provided Pagemaker software), and Linotype-Hell (which provided typeset quality font designs). These relationships helped to ensure a large number of Postscript users early on, avoiding the chicken–egg problem often encountered when starting a feedback process. To facilitate further adoption, Adobe also provided continual support to application developers and printer manufacturers, making the integration of Postscript in their products easier. Adobe would often provide one of its own employees to participate on the product development team of a printer manufacturer that was incorporating Postscript technology. N. STAUDENMAYER, M. TRIPSAS, AND C. L. TUCCI When firms could not influence the overall architecture, they often attempted to regain some control by moving from the ‘‘Development web’’ quadrant of Figure 1 to the ‘‘Partial integration’’ quadrant, by rebundling modules in the system, sometimes in collaboration with other firms. Ideally the combined modules would have higher performance when bundled but could still function with other firms’ products if needed. For instance, Lexar Media worked with cameramaker Nikon to develop an interface that would optimize the speed of Lexar’s flash memory when it was used with an Nikon digital camera. While Lexar Media’s flash memory would still work with the cameras of other manufacturers, and Nikon digital cameras could use flash memory from other firms, performance was significantly improved with the Lexar/Nikon combination. Although Lexar maintained broad compatibility, this strategy could be dangerous if the newly bundled combinations had proprietary interfaces and were no longer compatible with products from other firms. Another option when firms could not influence the overall system architecture was to identify means for differentiation within the hidden module. They often attempted to move functionality and bring value into the module(s) that they did control. While the protection of intellectual property is important in many industries, it was particularly key in this situation where so much of the product definition was standard across firms. For instance, even though the interface for Compact Flash was defined by an industry association, Lexar Media developed a patented controller that enabled its Compact Flash to store images from a digital camera more quickly than competitive products. Another option was to bundle the product with value-added services. For example, Red Hat bundled consulting and distribution services with the commodity Linux package that one could, theoretically, download and install for free over the Internet. Interdependency Management: Challenges An interfirm modular environment is by definition supposed to be characterized by interfaces that eliminate interdependencies. Without eliminating interdependencies and without agreeing upon design rules, the products of different firms would not work together well enough for consumers to be satisfied doing their own mixing and matching. So in a ‘‘perfect’’ setting, firms producing the different components of MODULARITY AND PRODUCT DEVELOPMENT an interfirm modular system would have no interdependencies. Ironically, this study found that the large number of firms involved in the creation of these systemic products, combined with the imperfect ability of individuals to predict interdependencies, resulted in a situation where firms experienced more, rather than fewer, interdependencies with external firms in this environment. The systems were indeed ‘‘nearly decomposable’’ (Simon, 1962). Clean lines could not always be drawn, both within and between firms. Many interdependencies were not necessarily identifiable in advance or visualizable by employees; instead, they emerged unexpectedly over time. Given the shifting patterns of structures, relationships, and tasks, and the presence of multiple-party effects where firms communicated with one or two partners who in turn were interdependent with other parties in the network, more and more interdependencies remained ‘‘hidden’’ or beyond the conscious awareness or purview of all team members. These hidden interdependencies could pop up unexpectedly and required flexible management responses. Events such as the emergence of new firms or changes in competitive strategy often created new interdependencies mid-process (Tjosvold, 1986). When other firms made changes to interdependent modules, that change might simply affect the performance of other modules or might require that a firm respond by changing its product design. So despite the appearance of independence, even firms that had no formal relationship experienced interdependencies in this environment. Other emergent interdependencies resulted from cognitive limits at the individual level. It was simply impossible to unearth all the interdependencies in advance in very large system products. Some interdependencies emerged over time as a result of increased experience and knowledge of tasks. For instance, one executive in the present sample lamented that ‘‘there are some examples where we have gotten really far down the road and then we decide, you know, we can’t do that fundamental thing that you need us to do.’’ Product requirements and features therefore were continually evolving and were driven as much by external as internal factors. Interdependencies also created frustration when making systemic changes and upgrades. Upgrading a module could be a difficult task as could reacting to a partner’s proposed upgrade. Systemic shifts were especially challenging since coordination among the decentralized group of firms in different subindustries J PROD INNOV MANAG 2005;22:303–321 313 was difficult. Since changes to multiple modules involved coordinating multiple parties, it was often difficult to accomplish. In addition, the incentives of firms around whether to make a break with the past and to announce an incompatible upgrade differed, with firms that held a strong position preferring to maintain upward compatibility to exploit their installed base and with other firms preferring to make a break and attempt to level the playing field (Dhebar, 1995). Interdependency Management: Potential Solutions Existing thinking on interdependency management has largely emphasized up-front minimization or outright elimination of interdependencies, particularly with those outside the team (Galbraith, 1973; Thompson, 1967). The goal has been to identify the tasks, to define the interdependencies between them, and then to redistribute the tasks and/or to redefine the product architecture to minimize contingencies (Eppinger et al., 1994; Van de Ven, Delbecq, and Koenig, 1976; von Hippel, 1990). For example, if two different software development teams needed to access to and change the same software module and their actions conflicted, resulting in a failed system integration or system ‘‘crash,’’ the recommended solution would be to resegment the contingent tasks so that only one team was responsible for both sets of changes. In contrast, this study found that for firms in this sample, systematic management of interdependencies, as opposed to attempted up front elimination of them, was more effective. Unfortunately, as discussed previously, in many situations identifying all interdependencies ex ante was not possible; some interdependencies emerged midstream despite efforts to minimize them. Managers therefore focused on coordinating portfolios of interdependent development relationships holistically, attempting to recognize and to adjust to changing patterns as they arose rather than eliminating or minimizing interdependencies through initial structures (Bensaou, 1999; de Figuerido and Teece, 1996; Miller, 1993; Staudenmayer, 1997). For example, Lucent’s software development group, encompassing both internal employees and external collaborators, developed a process for bringing interdependencies out ‘‘into the light’’ when they occurred. If two different teams or individuals caused a system ‘‘crash’’ upon reintegration, they 314 J PROD INNOV MANAG 2005;22:303–321 communicated directly with each other to negotiate a solution—all dependencies were to be pursued person to person and not via proxy or third party. Only if an interdependency cropped up several times was there an attempt to ‘‘sever’’ it by consolidating control in one team. Interestingly, Lucent found that most interdependencies were only sporadic and not consistent. This more flexible approach, while difficult at first in that it required more frequent interactions, also enabled the various teams to adjust to ongoing changes more effectively. Building flexibility and experimentation into the design process also became extremely important (cf. Thomke, 1997; Thomke, von Hippel, and Franke, 1998) to deal with unpredictable change imposed by other firms changing their modules. In the firms studied here, evidence was found of fluid organizational structures and task assignments where team members’ roles evolved throughout the project. Task assignments within the team were no longer distinct. As a manager of a new wireless project at Lucent observed, ‘‘Everyone is responsible for everything.’’ Boundaries defining who was inside versus outside the team were also becoming fuzzy (cf. Bleeker, 1994; Davidow and Malone, 1992; DeSanctis, Staudenmayer, and Wong, 1999). Flexibility also came through forming a modular internal organization to increase adaptability and responsiveness (Sanchez and Mahoney, 1996). In attempting to overcome the problem of coordinating systemic product upgrades, some firms found it useful to create a special track of advanced long-term development projects that did not follow typical release cycles (e.g., a release every year or six months). The advantage of such an approach was that for major changes, the firm’s partners and complementors knew what to expect and were better able to buy into the long-term plan. The relative lack of time pressure also enabled enlarging the scope of the team, with heavy representation from external stakeholders. For products that were coming to market immediately, it would have been impractical to work with so many constituencies. For example, Red Hat had a team associated with Red Labs that was working on products/features destined for two to three years in the future. This contrasted with a typical Red Hat product development team that was working under a four-month release cycle. The Red Labs teams typically had 5–10 internal developers, 10–20 active external developers representing different firms’ (and even individuals’) interests, and 100–200 less active external members whose primary roles were N. STAUDENMAYER, M. TRIPSAS, AND C. L. TUCCI tracking what was happening, providing limited input, and developing small pieces of the products/ features. Relationship Management: Challenges The nature of product development is such that it has always involved contact with players outside the boundaries of the firm. Producers source components from suppliers and integrate those components into their products. The management of these supplier relationships has been found to yield accelerated product development through early, on-site supplier involvement during the design, development, and production of new products, services, or processes (Clark and Fujimoto, 1991; Dyer, 1996; Imai, Ikujiro, and Takeuchi, 1985; Takeuchi and Nonaka, 1986). Similarly, customers provide important input regarding their needs and sometimes become involved in the design process (Athaide, Meyers, and Wilemon, 1996; von Hippel, 1990). Understanding whether new products satisfy user needs, perform unique tasks, or have unique features is a strong predictor of new product success (Cooper, 1985). There thus exists considerable empirical evidence that external collaborations, resources, and relationships constitute an important part of the development process (Brown and Eisenhardt, 1995, 1998), although formal collaborative relationships in this domain have been traditionally managed as a discrete set of external dyads, with individuals playing key roles such as a ‘‘gatekeeper’’ (Allen, 1977; Roberts and Fusfeld, 1982), ‘‘scout,’’ or ‘‘ambassador’’ (Ancona and Caldwell, 1990, 1992). In an interfirm modular environment, however, especially in the ‘‘Development web’’ quadrant, the network of external firms with which a given firm might interact was on a different scale in terms of both the diversity and complexity of relationships. Rather than a set of dyadic relationships, firms were embedded in a broader, interrelated web of organizations including customers, suppliers, complementary producers, competitors, and institutions (Bensaou, 1999; Singha and Cusumano, 1991; Tidd, 1995). Firms had a combination of formal and informal relationships with other organizations, and in addition, these organizations also had relationships with each other. Firms in the present study’s sample therefore found that the coordination of relationships was much more difficult, with communications among multiple players particularly challenging. MODULARITY AND PRODUCT DEVELOPMENT To begin with, the scale and variegated nature of the relationships was overwhelming. In addition to some formal joint development projects, firms had less welldefined strategic alliances, multifirm research consortia, as well as arms-length relationships that occurred through trade associations and standards bodies. Coordinating this portfolio of relationships without sacrificing development time and quality was challenging. As a Gale Group product manager told us, ‘‘The number of partners that you have to have to be viable these days has grown exponentially, so overall you are putting much more energy into it.’’ Another interviewee lamented, ‘‘Just keeping track of [partner relationships] is hard, isn’t that embarrassing?’’ Given the proliferation of contacts, people needed to know much more about the tasks of individuals outside the firm’s boundaries, including individuals working for organizations with which a firm had no formal relationship. For example, a manager of one project commented, ‘‘From one to twenty people have to know what you are doing every minute while you are doing it, which affects how fast you can do it.’’ Coordinating firms with divergent interests was another challenge. As a manager at Red Hat put it, ‘‘In the a.m., you can be my partner, but in the p.m. you can be my competitor. And in a lot of cases, you are both at the same time.’’ A related challenge was how to control indirect effects where the results of one dyadic collaboration spilled over into other current or future relationships. For example, Red Hat had an equity relationship with Intel and a revenue relationship with Dell. Dell was also a partner of Intel. So whatever happened in the Red Hat/Dell relationship could potentially have an impact on Red Hat’s relationship with Intel. Thus, one critical issue was how to account for one’s collaborator’s collaborators. The complex network of external collaborations also fundamentally changed the nature of internal interdependencies and communication. Whereas earlier dyadic firm relationships generally had a single point of contact, respondents spoken to in this study noted that external collaborations now had multiple points of contact at different levels of the hierarchy, requiring higher levels of internal coordination in order to effectively manage them. This shift could be in part due to the general trend toward greater decentralization and autonomy (Hitt, Keats, and DeMarie, 1998). For example, a strategic alliance with one partner might yield multiple projects, thus requiring numerous project managers in the firm to coordinate closely. Integration therefore was multidimensional, occur- J PROD INNOV MANAG 2005;22:303–321 315 ring across levels, functions, and business units or divisions in a dynamic time frame. In fact, it was not uncommon for different groups within the company to be unaware that they were simultaneously coordinating with the same partner. As one executive noted, ‘‘We literally have cases where different people are talking to the same outside partner on projects that other people don’t know about . . . and [other business units] may have past relationships with other companies that you may or may not know about.’’ Motorola’s involvement in wireless Internet service provided one example of relationship challenges. A broad range of wireless information devices was emerging, ranging from laptop computers to ‘‘communicators,’’ (e.g., PDAs with wireless connections) to enhanced mobile phones. Motorola was involved with the development of a key component of these devices: the computer chips that enable the service. Motorola also manufactured the cellular phones themselves. To compete in this emerging arena, Motorola had to coordinate with a number of external firms. For instance, in 2000 Motorola was a member of the Symbian alliance (along with Psion, Ericsson, Nokia, and Matsushita), a group that was jointly developing Epoc, an operating system for mobile phones. At the same time Nokia, one of the partners of Symbian, also was working directly with Palm Computing to develop a competing cell phone operating system. In addition to these stakeholders, Motorola also had to work with both private and public organizations involved in developing complementary standards, including Bluetooth (a connectivity standard as discussed already), Java, and the Wireless Application Protocol (an information services standard). Lucent Technologies provided another example of some of these challenges. Digital switch software products were primarily designed to provide new services such as Caller ID or Call-Return. Yet these new functions had to be integrated into the existing software system infrastructure architecture, most of which was designed and implemented 25 ago. Lucent therefore faced the challenge of managing a ‘‘legacy’’ of external collaborations with customers that utilized these older systems in addition to developing and coordinating with a broad range of new partners. Teams involved internal Lucent engineers working on the new product, internal developers maintaining and creating interfaces with the old system, marketing representatives, external standards-making boards, computer hardware manufacturers, phone manufacturers, the corporate customer, end users, 316 J PROD INNOV MANAG 2005;22:303–321 and international governmental and institutional structures. The development and subsequent deployment of new products/services therefore involved hundreds of people both inside and outside the firm, spanning multiple time periods, with many organizational and functional affiliations. Relationship Management: Potential Solutions To manage the diversity of relationships with which they were confronted, some of the firms in the present sample centralized control over relationship management. They created a new role of ‘‘relationship manager,’’ an individual who typically reported to the division head or chief executive officer (CEO) and was responsible for coordinating strategic collaborations across the firm. This individual kept track of all external relationships across multiple development projects to ensure that the entire portfolio of relationships was balanced. A relationship manager could keep track of not only the firm’s own dyadic relationships but also relationships among other firms in order to know ‘‘who was talking to whom and about what’’ according to a manager at the Gale Group. Thus potential conflicts could be identified and resolved before they materialized in a significant way. When a firm had multiple relationships with a single partner, a relationship manager served as a control point, coordinating and communicating across projects. For example, the IBM relationship manager at Red Hat focused on ensuring that the messages being sent to IBM were consistent and coordinated. Establishing a relationship manager also helped to establish long-term relationship memory in the firm. As explained by one Gale Group executive, ‘‘It seems to me that most of the partnerships that we pursue, we pursue for a larger aim than any one particular project.’’ The expectation of future interactions combined with the knowledge that there was organizational memory could help to control potential opportunistic behavior among partners. Some of the sample firms also categorized relationships by tiers, defining distinct ‘‘mini webs’’ within their full set of external collaborators. These subsets of firms, typically categorized based on the function of the relationship (e.g., equity, revenue-generating), were coordinated intensely. For example, at the time of the interviews, Red Hat had 10 revenue-generating collaborations. Some of these were called ‘‘Tier One’’ relationships because they were the firm’s most im- N. STAUDENMAYER, M. TRIPSAS, AND C. L. TUCCI portant and influential partners. At any given time, only about half of those were actively and intensively involved in developing the latest rounds of new products, features, and services. Other firms within the sample identified a focal partner for external contact that served as a gateway to other producers. The firms communicated often with that primary contact, and that contact was in turn responsible for communicating with and coordinating more peripheral partners, both of which are examples of so-called ‘‘nesting’’ coordination techniques—webs within webs (Mintzberg and Van der Heyden, 1999). Matching Solutions with Types of Interfirm Modular Environments The concerns outlined led to experimentation on the part of the firms in the sample as they searched for solutions to help them cope with their environments. While performance data is not known, the solutions have been identified that appeared to be most appropriate for each quadrant of the matrix (see Figure 2). Thus, as managers apply the solutions identified in this study, it is suggested that they first evaluate which quadrant of the matrix they are currently located in and then identify appropriate action. Since firms in the Appliance quadrant are not experiencing interfirm modularity, most of the solutions suggested would not apply. The one exception is the application of flexible, modular design principles to the organization. If a firm is using modular design principals for its internal product development efforts, then it has been suggested that a parallel modular organization improves performance (Sanchez and Mahoney, 1996). In fact, a more flexible, modular organization form is suggested for all quadrants of the matrix. In the ‘‘Single-apex’’ quadrant, given that a firm is not the one controlling the architecture—and assuming that that the firm is unable to exert control over the architecture, at least in the short term—it is suggested that differentiating and adding value within the hidden modules is the best course of action. The need for extensive external relationships in product development is more limited in this context since firms are dealing with a single entity that controls the architecture, and thus relationship management solutions were deemed relatively less important and are not included here. When multiple firms are involved in defining the architecture, as in the ‘‘Partial integration’’ quadrant, MODULARITY AND PRODUCT DEVELOPMENT J PROD INNOV MANAG 2005;22:303–321 1. Attempt to influence architecture to advantage 2. Aim to differentiate within module 3. Develop fluid tasks structures and experiment 4. Create a flexible modular organization 5. Use internal-external teams with long lead times High 6. Centralize control over PD alliance management 7. Categorize and manage relationships by tiers Number of organizations involved in defining the system architecture Partial Integration 4. Create a flexible modular organization 317 3. Develop fluid task structures, experiment 4. Create flexible modular organizations within firm 5. Use internal-external teams with long lead times 6. Centralize control over PD alliance management 7. Categorize and manage relationships by tiers Development Web 2. Aim to differentiate within modules 4. Create flexible modular organizations within firm Low Appliance Single-apex Cluster Low High Number of firms involved in producing the “systemic” product Figure 2. Proposed Solutions and How They Fit with the Types of Interfirm Modularity then external relationships become important, so solutions such as centralizing relationship management become relevant. But since firms in this quadrant were generally combining previously distinct modules in order to add value, differentiation within hidden modules was not suggested. Finally, in the ‘‘Development web’’ quadrant all of the solutions suggested by the firms in the sample would be appropriate. Conclusions and Implications Summary This article explored the implications of interfirm modularity for the product development process. While a great deal of research has examined the organization of product development, relatively little work has focused on the context of interfirm modularity in particular—a context that is becoming more relevant across a number of industries. Interfirm modularity allows the products of different firms to work together in a decentralized system, often configured by the user. While some products such as stereo sys- tems have had this characteristic for a long time, others have only recently moved in this direction. Two dimensions of interfirm modularity are identified that are key: whether multiple firms produce products in a system and whether multiple firms/institutions control the overall architecture of the system. In the most extreme environment, called a Development Web, firms are confronted with a highly decentralized environment where large numbers of organizations have influence over different parts of a system’s architecture, and multiple firms produce the products that make up the system. Through an exploratory field-based study of product development processes at seven firms, both challenges and potential solutions to the challenges raised by this context were documented, making a number of contributions to understanding of the product development process. Challenges fell into three broad domains. First, firms were frustrated at their lack of control over the definition of their own products. Standard external interfaces imposed constraints on what their own products could deliver. Second, while an environment of interfirm modularity should in 318 J PROD INNOV MANAG 2005;22:303–321 theory eliminate interdependencies among firms since interfaces between products are defined ex-ante, it was found, ironically, that interdependencies were ubiquitous. Interdependencies continually emerged throughout the product development process, despite efforts to limit them. Third, firms found that the quantity and variegated nature of external relationships made their management exceedingly difficult. The sheer complexity was daunting, given both the size of the external network as well as the number of ties per external collaborator. Partners with whom control over the architecture was shared often had divergent interests, or at least not fully convergent interests. The solutions observed in this sample of firms were creative and in many cases ran counter to established wisdom. For instance, research has suggested many ways for a firm to influence interface standards, especially in markets with network externalities (e.g., Shapiro and Varian, 1999). While the firms in the present sample followed some of this advice, they also focused on a more neglected aspect of architecture, the compliance and testing standards that accompany modules and interfaces. By concentrating their efforts in a different area, even smaller firms in the sample were able to have some influence. Similarly, existing research has focused on how to eliminate interdependencies via task partitioning, modularization of tasks, or structured coordination mechanisms (Galbraith, 1973; von Hippel, 1990). In the context of interfirm modularity, however, it was observed in the present research that instead of focusing on the elimination of interdependencies, firms benefited from concentrating on the management of interdependencies as they emerged. Finally, while layers of management and ‘‘bureaucracy’’ are often viewed as unproductive, even the relatively young firms in the sample found that adding structure, through positions such as relationship manager, was highly beneficial in handling the coordination and control of a wide range of external relationships. In addition to enhancing understanding of the product development process, this study also helps extend understanding of change and coevolution across levels of analysis. The present work demonstrates that change at the industry level in the form of increased modularity and standards can lead to profound changes at the organization, team, and individual levels. But as organizations experiment with new technological and organizational solutions to cope with this changing environment, their actions in turn contribute to the environment within which they op- N. STAUDENMAYER, M. TRIPSAS, AND C. L. TUCCI erate (Rosenkopf and Tushman, 1994). Changes at the level of the environment and firm level relationships therefore are related to new organizational forms and processes and vice versa, creating a reciprocal feedback loop (between individuals and firm, industry and firm, and among interacting firms). Implications for Future Research A number of interesting questions for future research arise from this work. Previous scholars have addressed the issue of ‘‘product proliferation’’ as more and more firms develop different versions of a core design (Barnett and Freeman, 1999). The present work suggests a parallel problem of ‘‘collaboration proliferation.’’ Thus, the presence of more and more external collaborative relationships means that firms must confront the inevitable issue of when and how to dissolve relationships. Clearly, there exists an upper limit in terms of just how many collaborations a firm can reasonably maintain and get value from over time due to cognitive and coordination limits. Furthermore, opportunities are continually arising where a new partner with greater expertise or industry power, or both, may be preferred over an existing one. Yet what are the criteria for disengaging, and how can a firm do so gracefully without limiting future opportunities for interaction? Collaborations, especially founding ones, involve an intense emotional component. How could (or should) a company abandon a small firm that played an important part in their success? The challenge of collaboration management in a modular industry environment involves knowing not only how to form and maintain, but how to dissolve such relationships. It involves balancing and managing personal and emotional issues just as much as strategic ones (DeSanctis, Staudenmayer, and Wong, 1999). Previous scholars have primarily concentrated on the ‘‘courtship’’ and ‘‘marriage’’ of external collaborating (e.g., Jemison and Sitkin, 1986). The present work would imply that there needs to be equal attention to issues of ‘‘divorce’’ or deliberately ending collaborations. Another question concerns the generalizability of the present findings. While the relevance of this research is clear for traditional high technology complex systems industries already characterized by interfirm modularity, it also has implications for historical ‘‘appliance’’ industries that are becoming more networked. For instance, the home automation industry takes a number of previously distinct product MODULARITY AND PRODUCT DEVELOPMENT categories and creates interfaces among them in order to deliver an overall system. The management of intrafirm and interfirm networks in this evolving industry is thus an important element of product development (Tidd, 1995). While there is a temptation to think of modules and interfaces in terms of a physical product, one also can imagine knowledge-based industries for which interfirm modularity is relevant. The drug development process and the biotechnology and pharmaceutical industries, for instance, share many of the characteristics described here, with interorganizational networks of learning playing a key role (Powell, Koput, and Smith-Doerr, 1996), even if the end product, a drug, is not a systemic product. Some of the suggestions described here for relationship management therefore also may be relevant for firms in these industries, while concerns about product definition may be less salient. Lastly, interfirm modularity is becoming increasingly important in industries historically categorized as low tech due to the adoption of the Internet across industries (Afuah and Tucci, 2003; Brews and Tucci, 2003). Publishing, for instance, is in the midst of radical change as more and more material is published real-time over the Web without the need for typesetting and printing of batched material. The present authors look forward to research that examines the impact of interfirm modularity on product development in these additional industries. Given increased levels of flexibility and fluidity, it seems that firms undergoing industry changes toward modular environments may want to reconsider the manner in which they track and communicate both internal and external collaborative relationships, depending on the type of modular environment their industry is evolving toward. For example, traditional organization charts, which treat people and units as boxes with clear lines of connection, primarily represent vertical chains of authority. Not only are such representations unable to keep pace with the level of change inside the firm, they also somehow fail to help employees struggling to understand how their companies work. What are the relationship ties and configurations? What parts connect to one another in web-like interdependencies? How should processes and people come together? Organizational charts’ relevance therefore may be diminishing in situations where traditional forms of hierarchy are being replaced by new organizational forms. Elements such as chains, hubs, and webs represent what an organi- J PROD INNOV MANAG 2005;22:303–321 319 zation is and what it does: relationships and processes as opposed to names, titles, and reporting relationships (Mintzberg and Van der Heyden, 1999). The present research indicates that firms are desperately searching for adequate ways of representing new organizational forms that capture the rich complexity of work relationships today but are also simple and flexible enough for employees to understand—an area demanding future research. Indeed, numerous examples were seen where firms were experimenting with solutions to the coordination of collaborations such as the role of a relationship manager, nesting coordination techniques, new metrics, and databases to track partner value. Further theoretical and empirical work is needed to establish the basic conceptual problem and test the robustness of these managerial solutions. The context outlined in this article foreshadows new managerial challenges and new theoretical and empirical research opportunities that will increase understanding of work in contemporary organizations. References Afuah, A. and Tucci, C.L. (2003). A Model of the Internet as Creative Destroyer. IEEE Transactions on Engineering Management 50(4):395–402. Aldenderfer, M.S. and Blashfield, R.K. (1984). Cluster Analysis. Thousand Oaks, CA: Sage, Quantitative Applications in the Social Sciences Series No. 44. Allen, T.J. (1977). Managing the Flow of Technology. Cambridge, MA: MIT Press. Ancona, D.G. and Caldwell, D.F. (1990). Beyond Boundary Spanning: Managing External Dependence in Product Development Teams. Journal of High Technology Management 1(2):119–135. Ancona, D.G. and Caldwell, D.F. (1992). Bridging the Boundary: External Activity and Performance in Organizational Teams. Administrative Science Quarterly 37(4):634–665. Athaide, G.A., Meyers, Patricia W. and Wilemon, David L. (1996). Seller–Buyer Interactions during the Commercialization of Technological Process Innovations. Journal of Product Innovation Management 13(5):406. Baldwin, C. and Clark, K.B. (2000). Design Rules: The Power of Modularity. Boston: Harvard Business School Press. Barnett, W. and Freeman, J. (1999). Too Much of a Good Thing? Product Proliferation and Organizational Failure. Working Paper. Stanford University. Bensaou, M. (1999). Portfolios of Buyer–Supplier Relationships. Sloan Management Review 40(4):35–44. Bleeker, S.E. (1994). The Virtual Organization. The Futurist March– April, 9–14. Brews, P. and Tucci, C.L. (2003). Building Internet Generation Companies: Lessons from the Front Lines of the Old Economy. Academy of Management Executive 17(4):8–22. Brown, S.L. and Eisenhardt, K.M. (1995). Product Development: Past Research, Present Findings, and Future Directions. Academy of Management Review 20(2):343–378. 320 J PROD INNOV MANAG 2005;22:303–321 Brown, S.L. and Eisenhardt, K.M. (1998). Competing on the Edge: Strategy as Structured Chaos. Boston: Harvard Business School Press. Clark, K.B. and Fujimoto, T. (1991). Product Development Performance. Boston: Harvard Business School Press. Clark, K.B. and Wheelwright, S.C. (1992). Organizing and Leading ‘‘Heavyweight’’ Development Teams. California Management Review 34(3):9–28. Cooper, R.G. (1985). Selecting Winning New Product Projects: Using the NewProd System. Journal of Product Innovation Management 2(1):34–44. Cusumano, M.A. and Nobeoka, K. (1998). Thinking beyond Lean. New York: Free Press. Davidow, W.H. and Malone, M.S. (1992). The Virtual Corporation. New York: Harper Business. de Figuerido, J.M. and Teece, D.J. (1996). Mitigating Procurement Hazards in the Context of Innovation. Industrial and Corporate Change 5(2):537–559. DeSanctis, G., Staudenmayer, N. and Wong, S.S. (1999). Interdependence in Virtual Organizations. In: Trends in Organizational Behavior. C. Cooper and D. Rousseau (eds.). New York: Wiley and Sons. Dhebar, A. (1995). Complementarity, Compatibility, and Product Change: Breaking with the Past? Journal of Product Innovation Management 12(2):136–152. Dyer, J. (1996). Specialized Supplier Networks as a Source of Competitive Advantage: Evidence from the Auto Industry. Strategic Management Journal 17(4):271–292. Eisenhardt, K.M. (1989). Building Theories from Case Study Research. Academy of Management Review 14(4):532–550. Eppinger, S.D., Whitney, D.E., Smith, R.P. and Gebala, D. (1994). A Model-Based Method for Organizing Tasks in Product Development. Research in Engineering Design 6(1):1–13. Galbraith, J.R. (1973). Designing Complex Organizations. Reading, MA: Addison-Wesley Publishing. Garud, R. and Kumaraswamy, A. (1995). Technological and Organizational Designs for Realizing Economics. Strategic Management Journal 16(special issue):93–109. Gawer, A. and Cusumano, M.A. (2002). Platform Leadership: How Intel, Microsoft, and Cisco Drive Industry Innovation. Boston: Harvard Business School Press. Glaser, B. and Strauss, A.L. (1967). The Discovery of Grounded Theory. Chicago: Aldine. Griffin, A. (1992). Evaluating QFD’s Use in U.S. Firms as a Process for Developing Products. Journal of Product Innovation Management 9(3):171–187. Henderson, R. and Clark, K.B. (1990). Architectural Innovation: The Reconfiguration of Existing Product Technologies and the Failure of Established Firms. Administrative Science Quarterly 35(1): 9–30. Hitt, M.A., Keats, B.W. and DeMarie, S.M. (1998). Navigating in the New Competitive Landscape: Building Strategic Flexibility and Competitive Advantage in the 21st Century. Academy of Management Executive 12(4):22–42. Iansiti, M. (1998). Technology Integration. Boston: Harvard Business School Press. Iansiti, M. and Levien, R. (2004). The Keystone Advantage. Boston: Harvard Business School Press. N. STAUDENMAYER, M. TRIPSAS, AND C. L. TUCCI Kahn, K.B. (1996). Interdepartmental Integration: A Definition with Implications for Product Development Performance. Journal of Product Innovation Management 13(2):137–151. Langlois, R.N. and Robinson, P.L. (1992). Networks and Innovation in a Modular System: Lessons from the Microcomputer and Stereo Component Industries. Research Policy 21(4):297–313. Larson, E.W. and Gobeli, D.H. (1988). Organizing for Product Development Projects. Journal of Product Innovation Management 5(3):180–190. Matutes, C. and Regibeau, P. (1988). Mix and Match: Product Compatibility without Network Externalities. RAND Journal of Economics 19(2):219–234. McGahan, A.M., Vadasz, L.L. and Yoffie, D.B. (1997). Creating Value and Setting Standards: The Lessons of Consumer Electronics for Personal Digital Assistants. In: Competing in the Age of Digital Convergence. D.B. Yoffie (ed.). Boston: Harvard Business School Press, 227–264. Methe, D.T., Toyama, R. and Miyabe, J. (1997). Product Development Strategy and Organizational Learning: A Tale of Two PC Makers. Journal of Product Innovation Management 14(5):323–336. Meyers, M. and Lehnerd, D. (1997). The Power of Product Platforms. New York: Free Press. Miller, D. (1993). The Architecture of Simplicity. Academy of Management Review 18(1):116–138. Mintzberg, H. and Van Der Heyden, L. (1999). Organigraphs: Drawing How Companies Really Work. Harvard Business Review 77(5):87–94 (September–October). Motorola Incorporated (2003). News Releases for Q1 2003. Available at: http://www.motorola.com/mediacenter/news/detail/0,1958,2349_ 1920_23,00.html (accessed March 2003). Nelson, S.A., Parkinson, M.B. and Papalambros, P.Y. (1999). Multicriteria Optimization in Product Platform Design. Proceedings of the ASME Design Engineering Technical Conference, Las Vegas September, 12–16. O’Mahoney, S. (2005). Non-profit Foundations and Their Role in Community Software Development. In: Perpsectives on Free Source and Open Software. F.B. Feller, J. Fitzgerald, S. Hissam, and K. Lakhani (eds.). Cambridge, MA: MIT Press. Papalambros, P.Y. (1995). Optimal Design of Mechanical Engineering Systems. Journal of Mechanical Design 117(4):55–62. Powell, W.W., Koput, K.W. and Smith-Doerr, L. (1996). Interorganizational Collaboration and the Locus of Innovation: Networks of Learning in Biotechnology. Administrative Science Quarterly 41(1):116–145. Roberts, E.B. and Fusfeld, A.R. (1982). Critical Functions: Needed Roles in the Innovation Process. In: Career Issues in Human Resource Management. R. Katz (ed.). Englewood Cliffs, NJ: Prentice-Hall. Rosenkopf, L. and Tushman, M. (1994). On the Co-evolutionary Dynamics of Organizations. In: Evolutionary Dynamics of Organizations. J. Singh and J.A.C. Baum (eds.). New York: Oxford University Press. Sanchez, R. and Mahoney, J.T. (1996). Modularity, Flexibility, and Knowledge Management in Product and Organization Design. Strategic Management Journal 17(special issue):63–76. Schilling, M.A. (2000). Toward a General Modular Systems Theory and Its Application to Interfirm Product Modularity. Academy of Management Review 25(2):312–334. Sengupta, S. (1998). Some Approaches to Complementary Product Strategy. Journal of Product Innovation Management 15(4):352–367. Imai, K., Ikujiro, N. and Takeuchi, H. (1985). Managing the New Product Development Process: How Japanese Companies Learn and Unlearn. In: The Uneasy Alliance: Managing the Productivity– Technology Dilemma. K.B. Clark, R.H. Hayes and C. Lorenz (eds.). Boston: Harvard Business School Press, 337–375. Shapiro, C. and Varian, H.R. (1999). Information Rules. Boston: Harvard Business School Press. Simon, H.A. (1962). The Architecture of Complexity. Proceedings of the American Philosophical Society 106:467–482. Jemison, D.B. and Sitkin, S.B. (1986). Corporate Acquisitions: A Process Perspective. Academy of Management Review 11(1):145–163. Singha, D.K. and Cusumano, M.A. (1991). Complementary Resources and Cooperative Research: A Model of Research Joint Ventures among Competitors. Management Science 37(9):1091–1106. MODULARITY AND PRODUCT DEVELOPMENT Staudenmayer, N. (1997). Managing Multiple Interdependencies in Large Scale Softward Development Projects. Unpublished Ph.D. diss. The Sloan School of Management, Massachusetts Institute of Technology. Swink, M.L., Sandvig, J. Christopher and Mabert, Vincent A. (1996). Customizing Concurrent Engineering Processes: Five Case Studies. Journal of Product Innovation Management 13(3): 229. Takeuchi, H. and Nonaka, I. (1986). The New New Product Development Game. Harvard Business Review 64(1):137–146. Terwiesch, C., Loch, C. and Niederkofler, M. (1998). When Product Development Performance Makes a Difference: A Statistical Analysis in the Electronics Industry. Journal of Product Innovation Management 15(1):3–15. Thomke, S., von Hippel, E. and Franke, J. (1998). Modes of Experimentation: An Innovation Process and Competitive Variable. Research Policy 27(3):315–332. Thomke, S.H. (1997). The Role of Flexibility in the Development of New Products: An Empirical Study. Research Policy 26(1): 105–119. J PROD INNOV MANAG 2005;22:303–321 321 Thompson, J.D. (1967). Organizations in Action. New York: McGraw Hill. Tidd, J. (1995). Development of Novel Products through Intraorganizational and Interorganizational Networks: The Case of Home Automation. Journal of Product Innovation Management 12(4):307–322. Tjosvold, D. (1986). The Dynamics of Interdependence in Organizations. Human Relations 39(6):517–540. Ulrich, K. and Eppinger, S.D. (2000). Product Design and Development. New York: Irwin McGraw-Hill. Van de Ven, A.H., Delbecq, A.L. and Koenig, R. (1976). Determinants of Coordination Modes in Organizations. American Sociological Review 41(3):322–337. von Hippel, E. (1986). Lead Users: A Source of Novel Product Concepts. Management Science 32(7):791–805. von Hippel, E. (1990). Task Partitioning: An Innovation Process Variable. Research Policy 19(5):407–418. Yin, R.K. (1984). Case Study Research. Beverly Hills, CA: Sage. Yoffie, D.B. (1997). CHESS and Competing in the Age of Digital Convergence. Boston: Harvard Business School Press.
© Copyright 2026 Paperzz