Real-World R&D: Jumping the Product Generation Gap

In most companies, the R&D process follows a well-trodden, familiar path. First comes basic research, in which the scientists in charge explore a new concept—say, a new polymer. Next, scientists with specific knowledge of the research area improve the concept until they identify an application for the polymer, such as a new insulating material. Finally, they hand off the job of actually developing a commercial product and its manufacturing process to engineers down the line. I think of this traditional R&D approach as a series of successive refinements, one group of experts after another adding its contribution to the developing product. Such a linear approach tends to compartmentalize specific knowledge; a particular part of the R&D process may even be restricted to a single researcher. Certainly the traditional R&D pipeline, based on the assumption that the greatest challenge to developing new products lies in making scientific discoveries, has led to significant innovations, such as the transistor and the color television. Yet I will argue that many of the best high-tech companies, particularly in the computer industry, have evolved a fundamentally different approach to new product development, one that is much more efficient and better suited to today’s R&D complexities. This new approach, which I call system focus, integrates the entire R&D process, rather than just shooting projects down a narrow pipeline. Based on a study of the R&D organizations of 12 mainframe computer companies—AT&T, Bull, DEC, Fujitsu, Hitachi, IBM, ICL, Mitsubishi Electric, NEC, Siemens, Toshiba, and Unisys—my research associates and I have determined that system-focused companies achieve the best product improvements in the shortest time and at the lowest cost. This is no small feat in the high-tech industry, where rolling out new, technically challenging products at the right time has been the key to success for more than a decade. We compared the approaches these R&D organizations took during the 1980s in developing new technology for a particular product: the multichip module. Mainframe multichip modules house and connect the computer’s most crucial integrated circuits, affecting the entire system’s speed and reliability. They thus present an extremely complex and technically challenging development task, precisely the type of R&D hurdle most companies now face, whether they produce computers, cars, or pharmaceuticals. The different approaches these companies took toward the multichip module demonstrate how a product, its manufacturing process, and the needs of its users constitute a system and should be developed as such. Any change in this product system—for example, the use of a more reliable material in the production of the module’s substrate—will change the entire design and manufacturing process. In turn, these changes can lead to development dead ends, longer product rollouts, and many wasted hours if not accounted for from the very beginning. Therefore, the goal of new product development shifts from simply incorporating a powerful new element to optimizing the whole system. High-tech companies are most affected by volatile markets and the rapid pace of technological change. But while industries that have traditionally been considered science-based, such as aerospace and semiconductors, may feel the most pressure to adapt their R&D approach, the challenge of incorporating new technical developments into commercial products is now a reality of competition in many other industries as well. From the application of laser-welding in the Lexus LS 400 automobile to the use of composite materials in the construction industry, new technical concepts have fueled innovations in a wide range of products. Improving the speed and efficiency of product development has, indeed, become a major competitive weapon. The Integration Team When it comes to R&D resources, principally the time and people involved, the most striking characteristic of a system-focused approach is the central role of technology integration. By this I mean the integration of the R&D process within a company, not the “fusion” of different technologies to create new products, although technology fusion often does result in complex components like the multichip module. System-focused companies form a core group of managers, scientists, and engineers at the earliest stages of the R&D process. This integration team investigates the impact of various technical choices on the design of the product and the manufacturing system. The team’s main objective is to balance new research from the lab with the manufacturing system’s existing capabilities. A good integration team adapts new technologies to what a company already knows how to do. And more than that, it enhances the current system to take advantage of those new ideas. A good integration team adapts new technologies to what a company already knows how to do. In fact, all of the system-focused companies in our study established integration teams in the early stages to handle technology integration for the multichip module. Some of the companies went so far as to build pilot plants in order to experiment with new technologies, a sizable but worthwhile investment since it often provided unambiguous information on future production yields and manufacturing costs. What follows is a typical profile for a successful integration team. In general, the members are the foundation of a system-focused approach to R&D. They possess a T-shaped combination of skills: they are not only experts in specific technical areas but also intimately acquainted with the potential systemic impact of their particular tasks. On the one hand, they have a deep knowledge of a discipline like ceramic materials engineering, represented by the vertical stroke of the T. On the other hand, these ceramic specialists also know how their discipline interacts with others, such as polymer processing—the T’s horizontal top stroke. Note that in the system-focused companies we studied, the fact that team members had this combination of skills was no accident. Team leaders considered their most important job to be assigning projects that would develop individuals by providing learning opportunities in other areas. According to one of the project managers we interviewed, “Each engineer is responsible for lots of different parts of the project. We choose the pieces to stretch their knowledge.” Successful integration teams specify and design both the product and the manufacturing process, leading the way in choosing new equipment. The team works with both system and component designers (for example, both mainframe designers and chip developers), facilitating a clearer understanding of how individual product components will interact. In fact, the integration team is usually in daily contact with the manufacturing plant. Team members often deal with major production problems on ongoing product lines, which allows them to evaluate the impact of new technology on production. And as a development project progresses, the team physically moves to the site of major technical difficulties, such as the pilot or manufacturing facility. Perhaps most important of all, the integration team works on a stream of related products, forming a cohesive unit of engineers who develop from project to project. Retained over multiple product generations, team members become the company’s repository of technically integrated “system” knowledge. While various research groups continue to develop and present new options, it is the integration team that turns new ideas into useful work by conceptualizing new products and providing continuity. For projects that jump several product generations, system focus can end up saving hundreds of millions of dollars. Such continuity over product generations pays off. Even when it comes to individual research projects, system-focused companies can save many years of staff effort and development time (see the graph “Finding the R&D ‘Sweet Spot’”). Even one less engineer working on a project saves roughly $ 100,000 a year; for projects like the multichip module, which extend over a decade and across product generations, the difference between traditional R&D and system focus can amount to hundreds of millions of dollars. Finding the R&D “Sweet Spot” Of course, an emphasis on technology integration should not cut into the status of a company’s research organization. I believe that for system focus to succeed, basic researchers must provide the integration team with a broad array of technical possibilities. Most integration teams will have a natural bias toward using older approaches to materials and manufacturing, because that’s what team members are familiar with. However, research (whether conducted internally or by outside suppliers) must offset this potential inertia by championing a variety of alternatives. Though they frequently used suppliers to provide additional technical options, the most successful companies in our study had vibrant internal research organizations of their own. Research: System Focus vs. the Traditional Pipeline When the traditional research pipeline and system focus are examined side by side, their differences are clear. Let’s start by comparing them during the research phase, when scientist-engineers investigate new technical possibilities for upcoming product generations. The following case histories of “Traditional Company A” and “System-Focused Company B,” loosely based on companies in our study, illustrate the differences. Note that the multichip module provides a good example of development complexity and its inherent research challenges (see the sidebar “Development Complexity and System Focus”). For instance, the latest IBM multichip module, which makes up the core of its new ES9000 mainframe computer, contains more than 65 stacked layers of electrical circuits that total about one mile of wiring—all packed on a five-inch-square flat piece of ceramic. Development Complexity and System Focus When the entire R&D process becomes system-focused, companies are in a much better position to wrestle with the development complexity inherent in the creation of today’s new products. Development complexity refers to the complications of the R&D process itself, not what the end product looks like or how difficult it is to manufacture. For example, the development of drugs involves integrating many different research decisions based on knowledge of pharmacology, toxicology, physiology, relevant body systems, regulatory requirements in dozens of countries, and differences in medical practice and custom in those countries. But ironically enough, the pill that results from such a complicated set of R&D decisions is often so simple it can be manufactured in a home laboratory. This illustrates how bringing successful drugs quickly to market depends much more on integrating the knowledge of all R&D factors at every point in the process than on pure scientific prowess. Of course, the development complexity that characterizes an industry can change over time, often in response to competitive pressures for products that deliver higher performance. Twenty years ago, automobiles were assembled from discrete and largely independent parts. But today’s cars can be more accurately described as integrated systems rather than assembled parts. Consequently, a tight fit between individual design decisions is essential to satisfying customer expectations. Now automobile manufacturers frequently “tune” the noise characteristics of a car to match the expectations of the market it is designed to serve. That means a high-revving sports car has a much different sound quality—as well as overall ride—than a luxury car. Such specific R&D challenges require a focus on the entire system, since decisions in basic engine design must be linked to the structural properties of the chassis, right down to the acoustical properties of the felt used in floor mats. In such complex projects, the research phase involves more than coming up with a new material or production technique. It also includes technology integration in the form of a technical concept: a detailed specification of how the complete set of technical options will combine to provide the new product with good quality and low cost. While the research group controls technology integration in traditional companies, in a system-focused company the integration team takes charge. Traditional Company A. The research group at Company A explored a new ceramic material composition that could conduct a lot of heat. Using scaled-down equipment in the lab, these researchers fabricated small quantities of the material. They surveyed the literature, communicated intensively with scientists at universities, and performed many small-scale experiments aimed at characterizing the new material’s properties, including its detailed microstructure. All of their research led them to believe that this new ceramic material could increase the speed and reliability of their company’s multichip module. Obviously, these researchers had much to gain if their ceramic was adopted, and they aggressively championed this material. They subsequently won: senior managers allowed them to develop their concept for the entire module system. Therefore, the same scientists who had launched the initial investigations were then responsible for integrating their new materials into a functional system. Yet their sole criterion for success was feasibility: they had only to fabricate a small number of prototypes using lab equipment. The group eventually produced several partially functional modules, which showed that it ought to be possible to build a real module based on the new ceramic. Once the technical concept had proved feasible, technology integration, driven by the research group, was complete. The basic technologies and the product design were set—which is fine, if both hold up in the real world of manufacturing. But in this case, the technical concept became a foregone conclusion without an active investigation of the challenge of future high-volume production. In fact, the activities of the research group ended at this point, and the module project was deemed ready for development. System-Focused Company B. Here an integration team, formed at the very beginning of the R&D process, led the multichip module project. These 20 scientists and engineers monitored the basic investigations conducted by the research group as well as those contracted to several outside ceramic materials suppliers. As the results from initial research came in, the integration team began selecting the most promising of the various new techniques and worked on combining the elements into one technical concept. They investigated several possible technical concepts in parallel, based on different combinations of materials, such as aluminum and glass ceramic substrates. This stage took much more time and involved more resources than the explorations of Company A (see the chart “From Concept to Product: Slimming Down Development”). As the integration team learned more about the systemic effects of the various alternatives, some were dropped, others postponed for future generations, and others refined and kept for further study. Of course, this weeding of alternatives contrasted sharply with the process at Company A, where the initial research group championed their ceramic material and focused on demonstrating its feasibility. From Concept to Product: Slimming Down Development In fact, Company B’s integration team did not identify feasibility as an explicit requirement for their technical alternatives. They gradually selected the most promising concepts, using a steady, ongoing selection process designed to solve problems largely ignored by Company A’s researchers: manufacturability, yield, and reliability. And as the integration process steadily continued, other groups became more involved. Equipment suppliers, for example, worked directly with integration team members to specify the production tooling for the new system. Engineers from various development groups and the manufacturing facility discussed the details of the production process. It was only at this point that the integration team finally committed to a single new concept and froze the basic design of their company’s multichip module. Competing Philosophies. Clearly, Company A and Company B represent fundamentally different approaches to the exploration and selection of new technology. Under the traditional R&D philosophy of companies like A, researchers assume that discovering new scientific possibilities is the critical challenge and involves the identification and early exploration of new ideas. While the traditional R&D pipeline certainly allows for freedom in early investigations, its main aim is to select a concept that includes the technical possibilities with the maximum theoretical impact on future product characteristics, as proven “feasible” in the lab. The critical weakness of the traditional method, however, is that it does not characterize the actual impact of the technical concept on the product and manufacturing process before selecting a winning concept. This is particularly a problem when a given research project is part of a much larger and complex “product system” like the multichip module. The research group first optimizes the technology chosen, and the development group is left with the problem of making the system work. As one frustrated development engineer put it, “By the time we got involved, the basic technologies were more than 90 % established.” One frustrated development engineer said: “By the time we got involved, the basic technologies were more than 90 % established.” In contrast, the system-focused philosophy emphasizes discovering and capturing knowledge about the interactions between new research in the lab and the company’s existing product and manufacturing systems. At Company B, the goal was joint optimization of system and technology, even though that initially consumed more time and resources. An integration team encouraged unbiased investigation of many alternatives, but its selection was greatly influenced by the details of the existing system. One engineer at a system-focused company in our study said, “The most essential part of choosing a new technology is to establish its impact on the production process. And we still never get it quite right.” That in no way means that system-focused companies like B settle for less aggressive results. In our study, we repeatedly observed that traditional companies like A often developed individual elements that were superior to B’s—for example, ceramic materials with better electrical and thermal properties. Yet system-focused companies still achieved superior overall performance because technology integration from the start of the R&D project more than compensated for apparently inferior materials. In other words, the traditional R&D pipeline adds up to a whole that is less than the sum of its parts, while system focus produces a whole greater than the sum of its parts. Development: How System-Focused Companies Get Results Although comparing the research phases in Company A and Company B shows the importance of system focus, it’s in the development of new products that the approaches yield dramatically different results. The “Multichip Module Project Maps” provide graphic evidence of why system-focused companies ultimately save on both time and money in rolling out new products and their subsequent generations. Multichip Module Project Maps For most companies, a project really begins in earnest at the concept approval stage. The introductory research phase is considered quite cheap compared to the resources allocated for development Company A. When it came time for Traditional Company A to implement the new ceramic material for the multichip module, the resources devoted to the project mushroomed. As you can see in the project map for Company A, a large development group was necessary because of the difficulty of increasing production yields and product reliability. Many of the details of the winning technical concept turned out to be extremely difficult for Company A to implement. As the head of development for a traditional company in our study observed, “We completely underestimated what it would be like to ramp up.” After extensively redesigning the product and manufacturing process, the development group transferred the new multichip module system to the production facility. But pilot production also proved difficult, requiring additional design changes. After many false starts, targeted yields were finally achieved, volume production began, and the development group moved on to the next generation of module, taking over their end once more from the research group. Due to promotions and individual career choices, many of these development engineers shifted to unrelated projects. In all, it took Company A over 61/2 years of development time and about 800 person-years of engineering and scientific activity to complete the development phase. Company B. In contrast, System-Focused Company B completed the development phase in less than 41/2 years and 300 person-years. The chart “How Many Engineers Does it Take?” indicates the big difference in average resources used during development. In Company B, the integration team, which had been responsible for the basic conceptual design, remained in charge of the project during development. A few team members worked part-time on integration of the next generation. Most were deeply involved with development, working directly at the plant or with materials suppliers. How Many Engineers Does it Take? As development progressed, integration team members gradually shifted to work on the next generation; yet the team still led the process until the plant had achieved full production yields. Many members continued to be responsible for production yields even after the product had been introduced, and they were called into the plant when major problems occurred. In fact, the same people who were responsible for improving production yields of the current generation also often worked on specifying the technical concept for the next generation. Organizational Learning. System-focused companies like Company B capture knowledge about the different elements of the entire R&D process and feed that knowledge back in at the technology integration stage. At Company B, involving engineers in the integration of several product generations allowed them to transfer valuable knowledge all along the line. This upstream transfer of information provides not only continuity from generation to generation but also continual learning about the impact of new technology on the complex production capabilities of an organization. And this transfer of knowledge goes beyond good communication. Organizational learning promotes individual learning, especially for integration team members. As one team leader noted, “At this point, one person is product engineer and production engineer at the same time. Basically, we no longer have the luxury of spending a lot of time communicating.” The sidebar “A System-Focused Résumé” illustrates how an engineer gains integration knowledge over several product generations. A System-Focused Résumé Mr. Furube 1991–1992 Engineer Currently designing chip set for gen 3 mainframe. 1989–1990 Massachussetts Institute of Technology: M.S. in electrical engineering. 1987–1989 Worker and Trainer Developed gen 2 module assembly process, working with suppliers. Designed, ordered, and set up production machines for microchip carrier. 1985–1986 Worker Improved production process for gen 1.5 module. Conducted feasibility study for new gen 2 module concept. Worked on basic conceptual design of gen 2 module; applied for patent. 1983–1984 Trainee Developed production process for LSI modules for gen 1 mainframe; set up specialized machines and trained workers on pilot production line. Designed production process for gen 1.5 module. 1979–1983 Tokyo University, degree in mechanical engineering Traditional companies like Company A devote substantial attention to the transfer of knowledge but only downstream—through the narrow pipeline of research to development to manufacturing. Under this approach, there is no mechanism for transferring knowledge back upstream so that it can improve the next round of technology selection and integration. Moreover, the compartmentalized knowledge typical of traditional R&D pipelines often disappears as scientists, engineers, and product developers shift from one project to another or take other jobs. Given this distinction in organizational learning, it’s not surprising that the system-focused companies in our study were faster and more productive on individual projects and that the performance gap between traditional companies and system-focused ones actually increased over time. We found a substantial difference between resources used in subsequent product generations for traditional and system-focused companies. The resources used by the system-focused companies each decreased dramatically as integration teams learned more and more about the product and its manufacturing system. When it comes to jumping the product generation gap, efficiently transferring knowledge is essential. In contrast, companies that relied on the traditional R&D pipeline decreased in productivity over successive product generations. Competitive pressures to keep up with increasing technological innovations led them to increase the resources they sunk into technology development. Yet without fundamental changes to the whole R&D process—namely, the use of integration teams to facilitate organizational learning—their R&D became even less efficient than before. Integrated Problem Solving The companies I’ve been calling system-focused depart from the traditional R&D pipeline both in philosophy and organizational design. They also differ significantly in their approach to solving problems. Our study included 61 different problem-solving attempts. While the traditional companies took a relatively narrow approach to solving problems, the system-focused companies were remarkable for the breadth of theirs. Consider the following example. Problem. The core of this multichip module contains more than 40 ceramic layers, each one carrying a complex circuit. The layers first are patterned, then stacked, and finally fired together at a high temperature. However, as the ceramic bakes, it shrinks. Therefore, the challenge is to achieve uniform shrinkage without buckling so that the module retains its shape and has a smooth surface for connecting integrated circuits. Solution A. Scientists at Company A identified the buckling problem early on in the research phase. At this point, they worked with a simplified model of the multichip module core, using scaled-down equipment in the lab. After making a number of adjustments to the ceramic composition and the firing process, they succeeded in eliminating the buckling. Later, during the development phase, a more representative model of the multichip module was used on a new pilot production line, which closely—though not exactly—represented the manufacturing conditions of volume production. The buckling problem reoccurred. Development engineers spent much time and energy adjusting the ceramic material and the production process in order to fix the problem again. At the end of the development phase, the new product was moved to the actual manufacturing plant. Once again, the buckling reoccurred, causing additional delays in the production schedule. After a great deal of effort, the third and final round of reengineering succeeded in eliminating the problem for good. Solution B. For scientists at Company B, the problem wasn’t narrowly defined as “how to eliminate all buckling” but instead as “how to get the entire system to function effectively.” The integration team focused earlier on a prototype of the product and the manufacturing process, using a pilot production line with equipment representative of actual conditions but flexible enough to allow experimentation. Developing a module prototype that was representative of high-volume production conditions allowed the team at Company B to characterize precisely the extent of the buckling. And by asking a broader question at the start and drawing on a broader knowledge base, the integration team found a faster and cheaper solution: controlling the buckling and coating the ceramic substrate with a polymer to smooth the surface. When the project moved from research into development, the buckling problem did not reoccur. In fact, the system-focused companies in our study were only slightly better at identifying problems early on; they identified 74 % of the significant problems in multichip module development, but the more traditional companies found 61 % of their own early problems. The most striking difference came in achieving real rather than apparent solutions. Only 40 % of the early fixes made by traditional companies stood up to the requirements of later project phases. Yet 77 % of the early solutions discovered by system-focused companies actually worked in the long run. A New R&D Philosophy System focus is a philosophy rather than a specific technique, one that underpins and reinforces the importance of technology integration: the mutual adaptation of new technology, product design, manufacturing process, and user needs. An engineer from a system-focused company said, “We get together with the semiconductor and systems group people to discuss future possibilities. Everyone talks about this, and negotiations occur throughout. There’s lots of give-and-take.” Compare this remark with that of an R&D manager from a traditional company: “The strategy is always to take a piece of the technology and set up a group to own it. If coordination problems exist, we set up a task force.” These two remarks reveal fundamentally different assumptions about what R&D is all about. Most companies that live by a system-focused philosophy emphasize the work of an integration team during all phases of a project. Yet the substantial advantages of technology integration don’t come without investment and the commitment of senior managers as well. It takes time to develop the skills of integration teams. The new approach may appear slow and cumbersome at first. And even after good results start to roll in, team leaders and senior managers still may need periodically to redirect the work of individual team members, helping them to fight inertia. Still, the most effective companies in our study demonstrated the value of system focus in an environment that is both complex and changeable. Their experience shows that an integration team can build a solid and powerful base of knowledge about the interactions between the most critical decisions in the design of a new product. Of course, while similar in purpose and character, integration teams of various organizations will develop different focuses depending on the nature of the technical environment. In high-performance computer processor design, the link between material choice and manufacturability presents the toughest challenge. Therefore, successful integration teams, like those responsible for the multichip module projects in our study, will emphasize retaining detailed knowledge of the impact past materials choices have had on manufacturing. In contrast, the pharmaceutical industry provides a very different set of R&D challenges. Integration teams there would find that the most complex interactions are between the chemical formulation of a new drug and its safety and efficacy, as perceived by both users and regulatory agencies. In this case, these would replace manufacturing difficulties and history as the foundation of a rich R&D knowledge base. But regardless of the industry, the traditional R&D pipeline is not up to managing technology integration in any environment that is characterized by development complexity. A company marketing cosmetics, for example, had to postpone rolling out a new product because its novel chemical formulation proved incompatible with the planned packaging. A forest products company barely avoided the complete failure of a new venture in engineered wood products; it discovered in the nick of time that there were subtle inconsistencies between the preset production process (which had been optimized for northern pine) and the properties of southern pine. And a pharmaceutical company failed to obtain approval for a new cancer drug due to poor integration of regulatory requirements and the development process. A competitor with a more system-focused approach integrated its regulatory activities and drug formulation accordingly and was able to roll out the new drug, gaining sole access to a very large market. With sophisticated customers who demand greater performance and new, aggressive, and often subtle product characteristics, developing successful products requires managing an increasing number of complex design decisions. Compared with old-line industries, which carefully nurtured deep knowledge of narrow specifics, today’s shifting markets call for a flexible breadth of experience, backed by the organizational and technical ability to integrate. For system focus to work, then, the company must have a consistent technology strategy and view the whole R&D process as a continuous stream of competence-building projects, not as a series of isolated efforts. Successful companies will target core technical areas and gradually build technology integration in those areas. Such a consistent approach, driven by the long-term commitment of senior managers, will allow integration teams to acquire the knowledge, tools, and procedures necessary for the efficient integration of new technology, ultimately producing the innovative products that customers want.

Real-World R&D: Jumping the Product Generation Gap

Partner Center