Annotated Bibliography Needed

We are indebted for support of this project to Doshisha Business School, to the Institute for Technol- ogy, Enterprise and Competitiveness (ITEC), Doshisha University and to the Center for Japanese Studies, UC Berkeley. We particularly benefited from the comments of Eva Chen, Process Excel- lence Leader, Intuit Inc. and Noriaki Kano, Prof. Emeritus, Science University of Tokyo. Neither is responsible for the uses we have made of their observations.

Too Much of a Good Thing? QUALITY AS AN IMPEDIMENT TO INNOVATION

Robert E. Cole Tsuyoshi Matsumiya

77CALIFORNIA MANAGEMENT REVIEW VOL. 50, NO. 1 FALL 2007

Q uality professionals stress the positive relationship between a firm’s quality focus and its innovative activities. Trailblazing empirical results include Noriaki Kano’s work in analyzing Konica’ s development of new camera features in the 1970s

in which he showed how a firm’s effort to understand customers’ latent needs could lead to product innovation.1 This research highlights the meaning of qual- ity as not simply freedom from deficiencies (e.g., reliability, durability, and con- formance to specifications) but product or service features that meet the needs of customers.2

The belief of quality professionals in the positive link between quality and innovation can be seen in the content of the proceedings of the 2007 annual conference of the American Society of Quality.3 Of the 82 individual presenta- tion titles, twenty-one had some version of the term innovation in their title (slightly over 25%). This prevalence suggests that quality professionals seek to link the subject of quality to the positively desired corporate value of innovation. Moreover, almost without exception, the presentations implicitly or explicitly assume a positive relation between quality and innovation. Finally, no distinc- tion is typically made between incremental and radical innovation. It is our hypothesis, and a major theme in this article, that while a firm’s pursuit of qual- ity is often quite compatible with incremental innovation, its relationship with radical or disruptive innovation is a good deal more problematic.

In contrast, academic researchers working on innovation have shown scant interest in quality. In a recent compendium of research designed to provide a holistic understanding of innovation, there are relatively few mentions of quality.4 When quality is mentioned, it is described as a trigger for innovation with pressure coming from either the demand or supply side. Also discussed is the role of quality ladders, in which new innovations of higher quality supersede those of lower quality and finally there is a discussion of total quality control as an organizational innovation. What little research that has been done on the topic of quality’s relationship with radical or disruptive innovation, however, has been conducted by academics outside the quality field.5

Japanese manufacturing firms are widely acknowledged to have been leaders in creating the modern approach to quality improvement. They are often exemplars of best practice, displaying strong quality cultures. Thus, we have chosen to use examples drawn from the recent experiences of Japanese firms and industries, though we will also introduce some U.S. cases. It is also widely acknowledged that Japanese firms have drawn strong competitive advan- tage for their high-quality performance and reputation combined with compe- titive prices. Most notable is the success story represented by the Japanese automobile industry, a very visible adopter, user, and promoter of modem qual- ity improvement methodologies. Its high-quality performance, especially relia- bility, has given an enormous competitive boost to the Japanese auto industry.

Successive process improvements (continuous improvement), built on a foundation of repetitive and relatively stable processes, play a major role in

determining success in the automotive industry. Technology changes quite slowly. In particular, safety considerations require that technology be fully tested before being implemented. This takes time and favors incremental innovation. Indeed, most of the empirical research purporting to show that quality improvement is consistent with innovation turns out to focus on incremental innovation.6 We have no reason to question these findings. It makes intuitive sense that quality improvement activities, properly

managed, could lead over time to successive incremental innovations, which, in turn, can make major contributions to competitive advantage.

What about industries in which radical/disruptive technological and mar- ket changes have characterized the industry? We will examine three such cases from the Japanese high-tech sector. The first is based on recent Japanese scholarship on Japan’s loss of the Dynamic Random Access Memory (DRAM) chip industry to Korean firms; the second is based on the author’s study of the Japanese network equipment industry and its adoption and deployment of the Internet; and the third is based on the authors’ study of NTT DATA, the largest Japanese software firm, specializing in system integration.

Too Much of a Good Thing? Quality as an Impediment to Innovation

UNIVERSITY OF CALIFORNIA, BERKELEY VOL. 50, NO. 1 FALL 200778

Robert E. Cole is Professor Emeritus at the Haas School of Business, UC Berkeley; Visiting Fellow at the Institute for Technology, Enterprise and Competitiveness (ITEC), Doshisha University, Kyoto, Japan; and also serves as Executive Director of LEAP, a joint program between eTQM College, Dubai, and the MOT Program at UC Berkeley. <[email protected]>

Tsuyoshi Matsumiya is a Graduate Student at Doshisha Business School, Kyoto, Japan. <[email protected]>

There is another benefit to pursuing these empirical cases. While other scholars have suggested that quality improvement can inhibit innovation, they have been vague on key issues. Sitkin, Sutcliffe, and Schroader have provided a conceptual analysis using a contingent model.7 They stress that control and vari- ance reduction have successfully characterized approaches to quality manage- ment in slow moving industries with repetitive processes but these approaches are not suited to exploring innovation where variance increase is important and to conditions of high task uncertainty. This is certainly consistent with our theme that the challenge of quality management may be quite different in the high-tech sector where radical/disruptive changes are more common and conse- quently task uncertainty high. Sitkin and associates’ analysis, however, operates at a high level of abstraction, stressing, for example, that non-repetitive error is desirable when exploring new solutions in highly uncertain situations, in con- trast to the traditional role of quality management in suppressing error. True enough, but their analysis doesn’t tell us anything about the specific mechan- isms and practices involved. Nor is there any analysis of how innovation- inhibiting activities are linked to competitive outcomes.

Benner and Tushman conducted an empirical study of innovation in the paint and photography industries to see the effects on innovation of the adop- tion of the process improvement initiative ISO 9000 (first version).8 They found that the greater the number of ISO certifications, the fewer the number of origi- nal patents. They concluded that a focus of firm resources on process improve- ment activities and variance reduction crowded-out exploratory innovations that would lead to more original patents. They did not explicate the organiza- tional mechanisms that produce these outcomes in the two industries but instead relied on the broad scholarly literature for their explanations. Thus, they claimed that the crowding-out process derives from an organizational cul- ture focused on incremental improvement and refining existing capabilities and routines. There is also no systematic analysis of the implications of their findings for competitive outcomes in the two industries studied. We extend their work by analyzing the following three cases in terms of the light they shed on the crowd- ing-out hypothesis, on still different mechanisms, and the knowledge they pro- vide of competitive outcomes resulting from an excessive inflexible quality emphasis.

Finally, there is the recent popular management literature that suggests that the tools of quality improvement, as reflected in the current popular use of Six Sigma methodologies, are not suited to innovation. The reasoning is that the Six Sigma culture of quality improvement and cost reduction, with its rigorous analysis, constricts choice, and as such does not mesh well with the more free- thinking and risk-taking culture required for germinating new ideas.9

Japanese Loss of the DRAM industry

Two recent studies of the Japanese loss of the DRAM industry to the Koreans inform this analysis. The first is by Takashi Yunogami and the second

Too Much of a Good Thing? Quality as an Impediment to Innovation

CALIFORNIA MANAGEMENT REVIEW VOL. 50, NO. 1 FALL 2007 79

by Chuma and Hashimoto.10 Japanese DRAM producers’ global market share plummeted from its high of almost 80% in 1989 to 10% in 2004.11 A commonly cited explanation is that the Japanese producers failed to keep pace with the aggressive capital investments of the Korean producers. True enough, but this explanation begs the question of why these investments were not made and whether the Japanese producers knew what were the right product and technol- ogy strategies. Japanese pre-eminence in the DRAM industry was cemented in the 1970s and 1980s as they supplanted American leadership. At this time, their DRAM production was heavily focused on its use in mainframe computers. Yunogami notes that at an early date, they sought to differentiate their products through high reliability and durability. Japanese producers were said to aim for 25 years’ durability. This was in keeping with the primary use of DRAM chips for mainframes at the time, with the national telephone company, NTT, being a major customer and primary driver of Japanese DRAM producer standards. Their success in differentiation through high reliability was confirmed by a test carried out in 1980 by Hewlett Packard and publicly reported by them.12

The success of Japanese DRAM producers created a mindset among employees that equated rising competitiveness with improved quality. Past practices set standards used for succeeding generations of DRAM chips. The pro- duction of higher and higher quality DRAM chips became the norm for which engineers aimed with strong support and guidance from powerful quality departments. As a result, more specialized equipment, more “masks,” more inspection steps, and overall more steps in the process flow were required. Along with higher and higher quality came added cost and time.13

The market for DRAMs shifted increasingly from mainframes to PCs in the early and mid-1990s. This radical change had major consequences for mar- ket demands. Speed and cost became more important. Korean producers entered the market focused on the PC market without any mainframe legacy practices. Chuma and Hashimoto document two turning points, the first when the market for DRAMS shifted to PCs and the second came with Samsung’s design and release of the 128Mbit DRAM in the year 2000.14 The 128Mbit DRAM was destined for vendors of low-priced personal computers and servers. It was a product admirably suited to meeting emergent user needs at the time. Its success cemented Samsung’s position as the dominant player in the DRAM market. This development coincided with the additional shift in DRAM use from PCs to con- sumer products (e.g., DVD players, cameras), bringing on still more changes in the device itself as well as interface specifications and packaging, all of which had significant implications for meeting user needs. The combination of these developments, for which management of the Japanese firms was not prepared, led most Japanese producers to withdraw from the DRAM market. Samsung, by leading the market, was able to build on its initial success by acting before others to gather valuable knowledge on emergent user needs. Through aggressive capi- tal investment and product engineering, it was able to incorporate that knowl- edge quickly into subsequent designs and products.15

Too Much of a Good Thing? Quality as an Impediment to Innovation

UNIVERSITY OF CALIFORNIA, BERKELEY VOL. 50, NO. 1 FALL 200780

There is an organizational dimension to these issues as well. The key Japanese players (NEC, Hitachi, and Toshiba) were semiconductor divisions in systems companies. As such, the DRAM engineers had to respond to the unique technical specifications (including those designed to raise quality levels) desired by internal customers. This interfered with them getting a correct sense of the rapidly changing user needs in the broader market. Companies such as Samsung and Micron were freestanding units without internal customers imposing demands on them, and thus they were able to adapt at a faster rate to changing market demands.16

In summary, the Japanese producers built an organizational culture that led to a management failure to adapt and innovate in response to the dramati- cally different market requirements of the new PC and later consumer product eras. They had designed into their development process and organizational cul- ture practices to insure high levels of quality, which, in the PC era, proved exces- sive and led to higher costs and longer development times. Like any other longstanding set of organizational practices, there had grown up norms, values, incentives, and political power to support their persistence. They were making incremental innovations but not responding to the more expansive meaning of quality. Specifically, they failed to adjust sufficiently fast—relative to competi- tors—to the dramatic changes in user needs produced by the rise of PC and later the shift to consumer products technology.17

We can see this case as one mode of the “crowding-out” explanation in which a focus on continually improving existing practices absorbs all resources and energy at the expense of a recognition and response to radical technology and market changes. More broadly, it is a dramatic example of a dialectical process in which the practices that make for success at one time and in one set of circumstances then trap firms and contribute to their downfall at a later time under a new set of circumstances.18

Challenges in the Shift to the Internet Era and the Network Equipment Industry

Our second case builds on an analysis of the evolution of the network equipment industry with the arrival of the Internet.19 NTT traditionally has served as the dominant technology trend setter for Japanese high-tech; and viewed from a competitive perspective, it was slow to embrace the Internet. Other major national telephone companies such as AT&T were also slow to embrace the Internet.20 The reasons for their initially negative response were quite similar.

NTT displayed powerful internal institutional rigidities. Above all, the digital Internet protocol challenged the major Western and Japanese telephone companies’ massive investment in analog telephone technology.21 The Internet was based on totally different principles. It is a best effort network based on packet technology. As a best effort network, the network does not provide any guarantees that data is delivered or that a user is given a guaranteed quality of

Too Much of a Good Thing? Quality as an Impediment to Innovation

CALIFORNIA MANAGEMENT REVIEW VOL. 50, NO. 1 FALL 2007 81

service level or a certain priority. In a best effort network, all users obtain best effort service. TCP/IP, the Internet protocol, drops many messages (packets do not get delivered on first try); its solution for these dropped messages is to resend them. In the language of quality, this is an undesirable rework solution, but in this case, one in which the cost of resending is almost zero.22

In the early years of the Internet, this best effort network did not match the traditional quality benchmarks of telephony; in the network area, these are known as Quality of Service (QoS) benchmarks. In particular, it was deficient in providing sufficient bandwidth guarantees as well as in insuring reliability (cor- rectness of data transfer). These weaknesses were huge in the eyes of NTT researchers and executives.

NTT had a powerful reliability culture that stressed progressive elimina- tion of error. With great effort, they had built a highly dependable system. NTT decision makers believed that the only way to get high-quality connections was based on a dedicated connection between sender and receiver—something that the Internet protocol, TCP/IP, did not offer. As late as the mid- and late-1990s, many high-ranking NTT executives still did not see TCP/IP protocol as serious technology. Internally, many derisively referred to it as a toy.

The choice for NTT was made even more difficult because they, like many national telephone companies in the West, were committed to developing an alternative technology for networking services. Moreover, NTT was more com- mitted than most because they believed they were leaders in developing this new technology and thus stood to benefit most. This technology was Asynchro- nous Transfer Mode (ATM) and it evolved from telephony. NTT began research on ATM in the mid-1980s. It was a natural extension of the existing public tele- phone network. So it would not require the destruction of NTT’s massive invest- ment in analog technology and infrastructure—as would the Internet. ATM also held more promise for QoS than TCP /IP. By emphasizing the active configura- tion of QoS parameters, ATM was expected to insure high quality and reliability. This, in turn, was important because of its implications for the reliability of real time communications. Ultimately, however, ATM was found to be too slow and unable to compete with the rapid deployment of TCP/IP in the marketplace.

Japanese high-tech firms were slow to embrace the Internet and to develop network equipment supporting its deployment, as was the case among many incumbent firms in the United States. Distinctive to the Japanese case, however, the leading electronic firms, as long-term suppliers to NTT, were accus- tomed to following NTT’s technology lead. As a result, they were mostly working on ATM and were slow to adopt TCP/IP. These long-term stable trust-based rela- tionships, known as “relational contracting” were held up as model helping to explain Japan’s success in the 1980s.23 However, in the context of rapid change and uncertainty, where the leaders make the wrong strategic technology choice, these relationships limit options and can be a recipe for failure. In this case, they greatly slowed the deployment of Internet technology by the leading electronic companies, giving Cisco a huge opportunity.

Too Much of a Good Thing? Quality as an Impediment to Innovation

UNIVERSITY OF CALIFORNIA, BERKELEY VOL. 50, NO. 1 FALL 200782

The delayed response by Japanese electronic firms gave first-to-market advantages to Cisco in its development of network equipment for the Internet. Cisco was then able to slowly learn what customers needed and then get their new products (such as routers) deployed in the marketplace. They then exploited the experience curve and made their operating system software, IOS, the standard software platform for Internet networking. This created strong bar- riers to entry for late-arriving Japanese firms.24 American firms, such as Cisco and Juniper Networks, continue to dominate the Internet network equipment market. A major element triggering this chain of events was NTT’s initial hostil- ity to TCP/IP, which, to a significant extent, grew out of NTT’s strong reliability culture and the dependency of the major electronic firms on NTT’s leadership.

Quality and Innovation in the Systems Integration Market

Our third case involves an examination of shifts in the system integration market as a result of radical technology and market change. The analysis is based on our case study research of NTT DATA. NTT DATA traditionally has been strong in the public sector and grew up in the mainframe era. Quality (specifi- cally, reliability) emerged as a key source of competitive advantage and market differentiation for the firm. Relative to the private sector, customers in the public sector are on average less knowledgeable about IT. Those from engineering fields with strong qualifications in IT are specifically prohibited from holding policy making positions in the national government, thereby excluding them and their expertise from IT decisions.25 This is an indication of the failure of the Japanese government to understand the strategic significance of IT. As a consequence, it made it easier for NTT DATA to tell customers what they needed with less fear of contradiction. Over the years, this bred a certain arrogance. It did not prepare the firm to dealing with knowledgeable private sector customers who are much more assertive about their requirements. Thus, we have the anomaly of a firm with a reputation for high quality that has not been very sensitive to customers’ perceived needs.

The explosive growth in IT applications in the private sector since the mid-1990s created huge growth opportunities. However, NTT DATA’s sales and market share have not kept pace with the rapidly growing private sector. Private sector customers value reliability, but they also value, equally and sometimes more, speed and innovation (new and improved functionality). The firm’s “no bug” culture, however, breeds risk aversion and slowness in introducing new features and technologies. It drives out alternative approaches to winning cus- tomers.

The firm uses a modified version of the “waterfall process” for its software development methodology. The waterfall method is a linear process of develop- ment (developed for mainframes) in which the firm insures against failure through fully specifying customer requirements before writing any code. The key to success in this model is accurately specifying customer requirements up front. The use of waterfall methodology makes programmers more productive,

Too Much of a Good Thing? Quality as an Impediment to Innovation

CALIFORNIA MANAGEMENT REVIEW VOL. 50, NO. 1 FALL 2007 83

technically speaking, since it allows them to make fewer changes downstream, thereby reducing rework. This approach assumes that the main design specifica- tions can be completely laid out in advance and that the firm’s system engineers are capable of extracting these “secrets” from customers. This methodology is consistent with the quality principle that quality should be built into the design of product and services as early as possible as part of a “prevention of error” strategy.

Market needs, however, changed greatly with the arrival of the PC era. In response to rapidly changing markets and technology, customer needs often change during the course of a project’s development, which can range from six months to a number of years. Thus, the main design specifications cannot be completely laid out in advance. If a firm uses a strict waterfall methodology in this environment, it runs the risk of delivering a highly reliable, bug-free prod- uct without the functionality needed by customers. What is called for instead is some sort of iterative development process that allows firms to incorporate changing user requirements during the development process. For a variety of reasons this is very difficult to develop, not least because of the common prac- tice, not only in Japan, of outsourcing the writing of code. In summary, follow- ing the quality principle of pushing quality prevention upstream into the design stage inhibits the innovations needed to respond to changing customer needs during the course of software development.

Implications of our Analysis of the Three Cases

Now we turn to what can be learned from our three cases. With respect to the Internet protocol, TCP/IP, we saw that initially it did not match quality benchmarks of traditional universal voice service. Those engineers developing TCP/IP, however, were able incrementally to add new features and improve reliability as one after another of its technical problems were gradually solved. Neither the ability to solve these problems nor the rate at which these problems would be solved, were knowable in the early years of the Internet. However, some researchers persevered.

Under these circumstances engineers in incumbent firms, steeped in a strong quality culture, initially found this technology unacceptable and were slow to adopt it. High-reliability firms (such as telephone companies, in particu- lar) have a great deal of difficulty of responding positively to technologies with these kinds of trajectories. They underestimated their potential for improve- ment. This left the initiative, and gave competitive advantage, to those start-ups that pushed ahead with TCP/IP.

The TCP/IP case is not unique. The many low QoS and reliability features exhibited by the early Internet are a common feature of disruptive technologies according to the research of Clayton Christensen.26 Incumbent market leaders, he argues, have difficulty responding positively to these technologies, which are often introduced in new and down markets where margins are low. Instead,

Too Much of a Good Thing? Quality as an Impediment to Innovation

UNIVERSITY OF CALIFORNIA, BERKELEY VOL. 50, NO. 1 FALL 200784

incumbents are geared to supplying their existing customers with improved products containing more and more features and functionality.

He cites many industry examples such as the response of incumbent U.S. steel leaders to the introduction of thin slab casting at steel mini-mills in the early 1990s. The initial steel produced by this new methodology had low surface quality. The incumbent steel companies could not use it with their exist- ing customers (e.g., appliance makers). Nucor, a start up and the leader in its introduction, successfully targeted less quality-sensitive markets.27 The leading firms were not interested in these markets with their lower profit margins. Over time, however, engineers at new firms such as Nucor were able to incrementally improve reliability as they solved many of the new technology’s problems. With this improvement, the use of thin slab casting products gradually moved up market, challenging the incumbent steel makers in these markets.

The challenge that disruptive technology poses for incumbents is univer- sal. Yet, it appears to have been applicable with special force in Japan. Japanese manufacturing firms are especially strong in producing high-quality products based on capturing user needs. User-led innovation is a notable feature of the Japanese manufacturing sector relative to the United States.28 Japanese top managers themselves often refer to this characteristic. Consider the 2004 state- ment by Machida Katsuhiko, President of Sharp Corp.

“Under the conventional approach to manufacturing, that is, user-oriented prod- uct development that seeks to make improvements and fix complaints, it will be extremely difficult to come up with new hit products. Consumers are seeking products that give them a sense of wonderment and create a positive impression.”29

When we speak of user-led innovation, we typically refer to current cus- tomers and the kind of innovation that is likely to result in incremental innova- tion. By being so focused on meeting user needs of current customers, however, Japanese firms are more likely to miss opportunities presented by radical/disrup- tive technologies that meet the needs of different sets of customers, create new markets, and offer totally new functionalities.30

Still another challenge arises out of Japanese industry’s overall success in raising its quality performance. This success owes much to the shift from downstream inspection of production processes to upstream prevention. This has been one of Japan’s great contributions to the global quality movement. Kaoru Ishikawa, the recognized leader of the postwar Japanese quality move- ment, describes the evolutionary process as follows:

Because of the limitations of the process control oriented quality assurance, . . . industry started to build quality into its products by performing careful evaluation at every stage of product development from new product planning through design to pilot production and by using the QC approach to investigate reliability in its broadest sense.31

There is little doubt that this new approach made a powerful contribution to quality assurance. As intended, it reduces risks associated with investing in

Too Much of a Good Thing? Quality as an Impediment to Innovation

CALIFORNIA MANAGEMENT REVIEW VOL. 50, NO. 1 FALL 2007 85

products that have a high probability of not meeting prevailing reliability, dura- bility, and performance standards. It is also not hard to see, however, that this approach, strenuously implemented, over time would make many companies more risk-averse in approaching innovation. One consequence of the upstream approach to quality is that more departments (such as production, business planning, and quality) participate earlier in the development process. They press demands for stability in the production process and meeting customer needs through the production of a product that will have few warranty claims. The product development and production engineering people feel a pressure to be responsive to these considerations.

These pressures increase the probability of creating a risk-averse organi- zational culture. As a result innovative product features and materials are often delayed or sacrificed because of the uncertainty they create. As was the case with TCP/IP, the improvement potential of new technology is often underesti- mated. One analyst observes that NTT DATA, with its zero bug culture, devel- oped a “corporate DNA” that places the highest priority on creating system stability. This leads the firm to give preferential treatment to products and tech- niques that have achieved satisfactory results in the past at the expense of the latest technologies.32 Again, we see the genesis of a risk-averse culture in which a firm chooses stability over innovation.

There is an additional challenge posed by Japan’s remarkable success in building its quality brand. Firms can become prisoners of their positive reputa- tions and then their reputation risks becoming a negative factor. One of the troublesome consequences of having a well-earned reputation for high-quality products is that firms may find it difficult to introduce to the market those prod- ucts that lack the reliability or functionality they normally contain. They behave this way because of fear that such products would damage their quality brand. Many of the growth markets of the future, however, are in the BRIC countries (Brazil, Russia, India, and China). To attack their mass markets often requires stripped down basic products that lack the value-added quality featured by Japanese firms.

Japanese auto firms, for example, find it hard to think of introducing cars to these markets without airbags. Apart from the ethical issues (which could be handled by making such equipment optional and leaving it to customers to choose), they find it hard to break the mode of ever-increasing quality. There is another sense in which Japanese high-tech firms may be prisoners of their qual- ity achievements. The Japanese market is particularly demanding (both individ- ual and business customers). As one manager said to us, “they (their customers) don’t tolerate any failure.” When these exacting standards are in sync with the global market, meeting these standards can be a great launching pad for Japan- ese exports. When domestic customers are more demanding of their radically innovative products than overseas customers, however, it may make Japanese companies conservative and slow to introduce them to the global market.

Too Much of a Good Thing? Quality as an Impediment to Innovation

UNIVERSITY OF CALIFORNIA, BERKELEY VOL. 50, NO. 1 FALL 200786

Possible Solutions: Strategies and Tactics

We have described a variety of challenges that quality improvement may pose for radical innovation. A strong quality improvement culture may lead a firm to be unresponsive to technology and market developments that shift demand to reduced quality requirements, as in the case of the DRAM industry. A strong quality improvement culture with a focus on user-led innovation for current customers may blind firms to new technologies and product features that would be attractive in new markets. In particular, a firm with a powerful reliability culture may not be receptive to new disruptive technologies that ini- tially display poor reliability, as was demonstrated by NTT and the leading elec- tronic firms’ negative responses to the Internet protocol, TCP/IP. We also saw that a “no bug” culture, apparent at NTT DATA, can breed risk aversion and slowness in introducing new technologies. Building in quality upstream early in the product planning stage, a powerful formula for improved quality, never- theless accentuates risk aversion when evaluating the potential of new products and materials. Locking in quality requirements early can be a special problem when there is rapid change in technology and markets during the product devel- opment process. User requirements may not be met under these circumstances. Overall, there is a danger that some Japanese high-tech firms could become “prisoners” of their quality brand.

While we won’t propose strategies and tactics to deal with all these challenges, we can make some observations. One can see it as a reasonable tradeoff for Japanese high-tech firm to be leaders in incremental improvement and innovation while conceding the introduction of radical innovation to others. Japanese export-oriented manufacturers have done quite well competitively emphasizing incremental user-led innovation over radical innovation. One man- ager of a major electronics firm we interviewed acknowledged that his firm would concede the introduction of radical innovation to others, arguing, how- ever, that with their strong stock of technology, they are fast followers. This may indeed be true when the firm is dealing with incremental innovation, but radical innovation often makes prior technology stocks outmoded, thereby rendering the fast follower model unusable. While researchers and managers continually search for ways that firms can be “ambidextrous,”33 there may be fundamental and intractable contradictions between the capabilities and practices required for incremental versus radical innovation.34

That said, in the spirit of those scholars and managers who advocate the need to build ambidextrous organizations that can both incrementally improve and truly innovate, we offer the following considerations.

Earlier, we discussed Japan high-tech firms as potential prisoners of their quality reputation and the challenge this posed for Japanese firms trying to attack the rapidly growing BRIC markets. It is a challenge born of their very success in creating a high-quality brand for their products. It is a particularly acute challenge because of new competitors from Taiwan, Korea, and above all China, which often appear more nimble and sensitive to user needs in these markets. One possibility is for Japanese firms to use subsidiaries, or develop

Too Much of a Good Thing? Quality as an Impediment to Innovation

CALIFORNIA MANAGEMENT REVIEW VOL. 50, NO. 1 FALL 2007 87

separate brands, for developing more basic products for these markets and in this fashion meet market demand without diminishing their own quality reputa- tion. These basic products typically have lower functionality. Contrary to popular view, these products are not necessarily of lower quality since they are fit for use for the targeted customers in question.

The novelty of such solutions in Japan is illustrated by a recent decision of Daihatsu, the Toyota group firm minicar maker. Luxury passenger cars have developed frontal collision prevention systems based on millimeter wave radar system. The use of such expensive systems in the mincar market would drive up the price to unacceptable levels. In response to this dilemma, Daihatsu engineers developed a laser-based system at about half the price of the radar-based system. Their new approach, however, has one serious disadvantage—it doesn’t function in rainy conditions. It rains a lot in Japan. They decided to release it and, not- withstanding its limitations, it has been popular among older drivers. In this case, then, Toyota clearly felt its reputation would not be damaged by the release of a lower-quality collision prevention system. That it was the subject of an arti- cle in the leading business newspaper, the Nikkei Shinbun, suggests the novelty of such a solution in Japan.35

Still another option could be found in Japanese high-tech firms making stronger use of beta testing as a way to ease into the market earlier without abandoning their high quality standards. Beta testing can occur in the phase just prior to mass production and distribution; it is a way of allowing the company to test the market and engage selected customers over the most desirable product features and functionality. This allows for further refinement prior to product release. Beta testing has become quite popular in U.S. over the last twenty years. It began in the computer industry, then spread to semiconductors and software by the late 1980s. By 1994, it was estimated that 50 percent of Fortune 500 com- panies had used beta testing and 20% used it regularly.36 It is used, however, only modestly in Japan and it is not discussed in the context of strategy. Yet, for many consumer and business products, customers see great value in getting the new product as early as possible. For individuals, it might be to show off to their friends that they have the latest technology or to participate in shaping the prod- uct, and for firms so it can be to get access to a competitive weapon as early as possible or to shape the emerging product. Properly done, customers are told of the risks up front; this makes them less likely to blame the manufacturer for problems that might arise. This approach minimizes the risk of damaging a firm’s quality brand resulting from too early release, while enjoying a number of bene- fits. For the firm, beta testing can provide early market intelligence about cus- tomer needs that helps shape the ultimate product. It is about intentionally exposing selected sophisticated users to error, an approach that doesn’t come naturally to Japanese firms. Beta products can also serve as a marketing tool to create a cachet about the product among early adopters.37

A stronger version of beta testing involves releasing early versions of products to selected markets. Sharp Corp. is reported to release its LCD panels for commercial production when its yield (complex measure of both quality and

Too Much of a Good Thing? Quality as an Impediment to Innovation

UNIVERSITY OF CALIFORNIA, BERKELEY VOL. 50, NO. 1 FALL 200788

productivity) reaches around 60%. Samsung is said to release some of its LCD panels to selected markets when its yield reaches 20%. A Japanese analyst explained to us that it is only at 60% that quality standards are met and the product is economically viable (can produce profits). Is Samsung therefore behaving irrationally releasing its product at 20% yield to selected markets like China? Not necessarily. By releasing the product early to selected markets, Sam- sung minimizes damage to the brand when problems surface. However, it also gathers—just as with beta testing—valuable market intelligence that allows it to adjust the product to better meet user needs as well as to solve particular quality problems, such as what level of defect is noticeable by viewers of LCD displays for specific applications. This early information gives them an edge on later entering competitors. They will be quicker to move to second-generation prod- ucts that incorporate the early feedback they have received. We see this ability to gather market intelligence by using select customers to shape the product prior to widespread release as part of a continuum. Key is choosing customers who will be representative of the target markets. On one end of the continuum would be a product like Wikipedia, where the user community totally designs the product and has a sense of ownership of that product once it is released; at the other end are beta releases.

Lastly, one can also pursue an approach that changes the focus of the quality attributes that are emphasized. With many innovative high-tech prod- ucts, performance and features will be more important than reliability and dura- bility in attracting customers at the beginning of the product cycle. While all are quality attributes, managers often see the latter as quality issues but not the former. Thus, this approach is easier to advocate than to implement.

The First to Market Challenge

Researchers have long noted that at the beginning of a new technology’s product cycle, early adopters are most interested in new functionalities (enabling them to do something they couldn’t do before) rather than reliability or durabil- ity per se.38 This may be especially true in the high-tech sector.39 Customers understand that there will be some problems with new products. Kume Hitoshi, a noted Japanese quality expert, points out that profit not quality is the goal early in the product cycle.40 Over time, competitors enter the market and match the functionality of the innovator’s product or service. Gradually the basis of competition shifts to price and quality. Customers will begin to choose products that are more reliable from vendors that are deemed reputable. Firms begin to design products that meet the changing needs of existing customers, They focus on meeting user needs, especially lead users, and a trajectory of quality improve- ments is set in motion as firms improve their products and supporting services from one generation to the next.41

Many large Japanese manufacturing firms would never think explicitly about trading off some quality for being first to market. Strict adherence to the simultaneous achievement of QCD (Quality, Cost Delivery) and “quality first”

Too Much of a Good Thing? Quality as an Impediment to Innovation

CALIFORNIA MANAGEMENT REVIEW VOL. 50, NO. 1 FALL 2007 89

mantras limit strategic choices. Quality is seen as something that completes innovation, as one Japanese consultant put the matter to us, not something to be traded off. Yet, in dynamic high-tech markets, as we saw in the case of Cisco and Samsung, a firm can get tremendous competitive advantage if they are able to get the right products deployed first in the marketplace, absorb and act on user feedback, exploit the experience curve and set industry standards to deter late entering firms. The former CTO of one the most successful and innovative U.S. software firms in Silicon Valley told us that he “didn’t know of any project [at his company] that ‘hasn’t’ sacrificed quality in an effort to get functionality into a customers hands, particularly if a competitor is on the horizon. There is always a trade-off and that trade-off is always reasoned about informally.” We see two very different approaches, perhaps emblematic of U.S. and Japanese high-tech executive thinking.

To be sure, there are circumstances in which the prudent decision would be to withhold a new product from the market until its quality problems are solved. Consider the recent case of Affymetrix. The company invented and dom- inates the commercial gene chip market. In late 2005, it unveiled a new genera- tion chip, but the product was released without sufficient testing and had flawed software. It took five months, a new software download, and delivery of new chips to fix the problem.42 This disaster provided an opening for a fast-charging start up, Illumina, which as a result gained market share at Affymetrix’s expense. Affymetrix countered to regain market share subsequently with a large price reduction, thereby reducing its profit margins. Clearly, it was a disas- ter on Affymetrix’s part to have released this under-tested product so early. They took dangerous risks and damaged their reputation. Could they not have released a beta product to a less important customer to assess the possible costs of such an early release? There can be high costs associated with letting a rival beat you to market with a new product, but there may be even higher costs for you beating them to market with a faulty unusable product.

On the other side, Apple allowed just 6 months product development time for its release of the first iPod in 2001, normally a product of this nature would have a one-year development time.43 Many firms were working on simi- lar products and Apple wanted to be first to market with what it thought was a novel solution—and to be able to take advantage of the upcoming Christmas season. The initial quality was not good. Notwithstanding, stories surfaced of consumers dropping and breaking the product, but instead of swearing off Apple, they just went out and bought another iPod. There are some parallels here to consumer responses to the recent X-box failures.44 After almost 5 years, Apple reports that failure rates were still running at somewhat “less than 5%.”45

This is not an enviable achievement. Yet, it is clear that the new functionality, ease of use and “cool” aesthetics of the new product, have more than out- weighed the product’s durability problems. They were the first to get the key things that mattered to users right. Moreover, the iPod’s durability has gradually improved with each new model and surveys show that users’ satis- faction remains above those of competitors.46 Thus, Apple’s decision to rush

Too Much of a Good Thing? Quality as an Impediment to Innovation

UNIVERSITY OF CALIFORNIA, BERKELEY VOL. 50, NO. 1 FALL 200790

development, before having worked out all the iPod’s durability problems, was justified as a business decision.

How does a firm know if they have Affymetrix’s genotyping equipment or an Apple iPod situation on their hands? That is the strategic challenge. There is no data to suggest that Apple intentionally downgraded reliability or durabil- ity. Nevertheless, in their rush to be first to market with a different kind of MP3 player, they ended up sacrificing durability. Yet, they survived the resultant bad publicity.

What if a firm, and we are not referring to just Japanese companies at this point, could determine in advance the benefits of being first to market relative to its competitors? Apart from short-term revenue benefits, will the firm be able to use early user feedback, and the experience curve to maintain its lead over com- petitors? Does the firm fully understand the quality deficiencies of its product and can it estimate expected costs of these deficiencies if it goes to market early. Can the firm estimate the probability of failure occurrences times the severity of consequences for the customer as measured in dollars and time lost? In other words, can they assess how much pain they would inflict on customers by going to market early?

If both the revenue and experience-curve benefits of going to the market early are large and the costs of weaker than “normal quality” are judged to be manageable, a firm might well consider taking the risk of going to market early. Success using this approach depends on the firm being able to accurately learn about and assess these benefits and costs. Moreover, is it able to accurately assess its learning capabilities relative to competitors? Are they a fast learner relative to competitors? David Apgar calls these non-random “learnable risks.”47 The chal- lenge is not just to learn about risks—involving customers and technologies— but to understand a firm’s risk-assessment capabilities, both of the risk occurring and the likely costs and benefits should it occur. Key is whether the firm can learn about these risks and deal with them faster than competitors. Unfortu- nately, too often, arbitrarily selected launch dates set months, even years in advance, dictate release dates, not careful analysis. The benefits of abiding by the sacred launch date, notwithstanding marketing commitments and the like, is way overrated.

We have explored the ways in which the quality culture of Japanese high-tech firms poses a challenge for innovation. The various mechanisms we have identified go well beyond what has been previously highlighted in the lit- erature. In particular, we called attention to the risk-averse culture that may be created, damaging in turn the potential to develop or respond to radical innova- tion. We also offered up some exploratory views on how Japanese firms might meet this challenge. Our analysis of the first to market challenge suggests that high-tech firms might benefit from thinking more strategically and flexibly about the role of quality at the early stages of the product cycle. As it stands now, Japanese high-tech firms display a strong tendency to overestimate the risks of going to market early with radically new products of less than stellar reliability or durability. U.S. high-tech firms may display the opposite tendency.

Too Much of a Good Thing? Quality as an Impediment to Innovation

CALIFORNIA MANAGEMENT REVIEW VOL. 50, NO. 1 FALL 2007 91

Notes

1. Noriaki Kano, “TQC as Total Quality Creation,” International Conference on Quality Control 1987, Tokyo, Nihon Kagaku Gijutsu Renmei, 1987.

2. J.M. Juran, Juran’s Quality Control Handbook, fourth edition (New York, NY: McGraw-Hill, 1988).

3. American Society of Quality, Proceedings of the World Conference on Quality and Improvement, Orlando, FL, April 30-May 2, 2007.

4. Jan Fagerberg, David Mowery, and Richard Nelson, The Oxford Handbook of Innovation (New York, NY: Oxford University Press, 2005).

5. See, for example, Sim Sitkin, Kathleen Sutcliffe, and Ronald Schroeder, “Distinguishing Control From Learning in Total Quality Management: A Contingency Perspective,” Academy of Management Review, 19/3 (July 1994): 537-564; M.J. Benner and M. Tushman, “Process Management and Technological Innovation: A Longitudinal Study of the Photography and Paint Industries,” Administrative Science Quarterly, 47/4 (December 2002): 676-706.

6. See, for example, Paul Adler, Barbara Goldoftas, and David Levin, “Flexibility versus Effi- ciency? A Case Study of Model Changeovers in the Toyota Production System,” Organization Science, 10/1 (January/February 1999): 43-68; Eitan Naveh and Miriam Erez, “Innovation and Attention to Detail in the Quality Improvement Paradigm,” Management Science, 50/11 (November 2004): 1576-1586.

7. Sitkin, Sutcliffe, and Schroeder, op. cit. 8. Benner and Tushman, op. cit. 9. Brian Hindo, “At 3M, A Struggle Between Efficiency and Creativity,” Business Week, June 11,

2007, <http://www.businessweek.com/magazine/content/07_24/b4038406.htm>. 10. Takashi Yunogami, “Technology Management and Competitiveness in the Japanese Semi-

conductor Industry,” in D. Hugh Whittaker and Robert E. Cole, eds., Recovering from Success, Innovation and Technology Management in Japan (Oxford: Oxford University Press, 2006); Hiroyuki Chuma and T. Hashimoto, “Nihon wa Naze DRAM de Sekai ni Yabureta no ka sono Haiin no Konkan wo Kenshou suru (1) [Japan Looks into the Basic Reasons Why They Lost the World Wide Competition with DRAM (1)],” Nikkei Microdevices (March 2007a), pp. 41-4; Hiroyuki Chuma and T. Hashimoto, “Nihon wa Naze DRAM de Sekai ni Yabureta no ka sono Haiin no Konkan wo Kenshou suru (2) [Japan Looks into the Basic Reasons Why They Lost the World Wide Competition with DRAM (2)],” Nikkei Microdevices (April 2007b), pp. 43-50.

11. Chuma and Hashimoto (2007a), op. cit., p. 43. 12. Robert E. Cole, Managing Quality Fads (New York, NY: Oxford University Press, 1999), p. 51. 13. Yunogami, op. cit., p. 80. 14. Chuma and Hashimoto (2007a), op. cit., p. 43-44. 15. Chuma and Hashimoto (2007b), op. cit., p. 44-45. 16. Communication from Prof. David Hodges, former Co-Director of The Competitive Semicon-

ductor Manufacturing Program, August 9, 2007. 17. Chuma and Hashimoto (2007a, 2007b), op. cit. 18. Cf., Veblen’s “trained incapacity” and March’s “competence trap.” 19. Robert E. Cole, “The Telecommunication Industry: A Turnaround in Japan Global Presence,”

in D. Hugh Whittaker and Robert E. Cole, eds., Recovering From Success: Innovation and Technol- ogy Management in Japan (Oxford: Oxford University Press, 2006).

20. John Naughton, A Brief History of the Future (Woodstock, NY: Overlook Press, 1999), p. 117. 21. Ibid., p. 100-107. 22. Cf., <http://en.wikipedia.org/wiki/Best_effort_delivery>. 23. Ronald Dore, Taking Japan Seriously (Stanford, CA: Stanford University Press, 1987), pp. 173-

192. 24. Cf., Annabelle Gawer and Michael Cusumano, Platform Leadership (Boston, MA: Harvard

Business School Press, 2002), pp. 175-178. 25. Interview with Prof. Kaneda Shigeo, College of Engineering, Doshisha University, November

13, 2006. 26. Clayton Christensen, The Innovator’s Dilemma (Boston, MA: Harvard Business School Press,

1997). 27. Clayton Christensen, Innovation and the General Manager (Boston, MA: Irwin McGraw-Hill,

1999), p. 43. 28. Edwin Mansfield, “Industrial Innovation in Japan and the United States,” Science, 241 (Sep-

tember 30, 1988): 1769-1774, at p. 1771.

Too Much of a Good Thing? Quality as an Impediment to Innovation

UNIVERSITY OF CALIFORNIA, BERKELEY VOL. 50, NO. 1 FALL 200792

29. <http://www.watch.impress.co.jp/av/docs/20040108/sharp.htm>. 30. Cf., Christensen (1999), op. cit., pp. 96-99. 31. Kaoru Ishikawa, Introduction to Quality Control (Tokyo: 3a Corporation, 1990), p. 15. 32. Nikkei Computer,” Chijimu Koukyou Sodatanu Wakete NTT DATA no ‘Naiyuu Gaikan’”

[NTT DATA troubled both at home and abroad by shrinking public and unsatisfactory devel- opment of younger employees], August 22, 2005, pp. 42-45.

33. Michael Tushman, Philip Anderson, and Charles O’Reilly, “Technology Cycles, Innovation Streams, and Ambidextrous Organizations: Organization Renewal Through Innovation Streams and Strategic Change,” in Michael Tushman and Philip Anderson, eds., Managing Strategic Innovation and Change (New York, NY: Oxford University Press, 1997); Linda Argote, Organizational Learning (Norwell, MA: Springer Publishers, 2005).

34. James March, “Exploration and Exploitation in Organizational Learning,” Organization Science, 2/1 (February 1991): 71-87.

35. “Daihatsu Makes Safety Affordable,” The Nikkei Weekly, May 14, 2007, p. 13. 36. J. Daly, “For Beta or Worse,” Forbes ASAP, December 5, 1994, p. 37. 37. Juliet Chung, “For Some Beta Testers, Its About Buzz, Not Bugs,” New York Times, July 24,

2004. 38. James Utterback, Mastering the Dynamics of Innovation (Boston, MA: Harvard Business School

Press, 1994), pp. 92-102. 39. Geoffrey Moore, Crossing the Chasm (New York, NY: Harper, 1991). 40. Hitoshi Kume, “Business Loss and Quality Management,” Quality Progress, 21 (July 1988):

40-43. 41. Christensen (1999), op. cit., pp. 96-99. 42. Marilyn Chase, “Price War Between Gene-Scanners Could Spur Research,” Wall Street Jour-

nal, December 21, 2006, p. B1. 43. Steven Levy, The Perfect Thing (New York, NY: Simon and Shuster, 2006), pp. 105-106. 44. Matt Richtel, “Xbox 360 Out of Order? For Loyalists, No Worries,” The New York Times,

August 13, 2007, pp. C1, C7. 45. Nick Wingfield, “When iPods Die,” New York Times, December 6, 2006, p. Dl. 46. Ibid. 47. David Apgar, Risk Intelligence (Boston, MA: Harvard Business School Press, 2006).

Too Much of a Good Thing? Quality as an Impediment to Innovation

CALIFORNIA MANAGEMENT REVIEW VOL. 50, NO. 1 FALL 2007 93