Until the AT&T break-up, Bell Laboratories maintained a rigorous Engineering Complaint System through which it collected technical problems and inquiries from its (customer)operating companies. These inquiries had to receive an action response within a month;that is, Bell Labs had to specify how it could solve or would attempt to resolve the customer's problem and notify the operating company in writing of its action. Motorola, established its 10X program to cut the cycle time for innovation by a factor of ten to improve its response to identified customer problems or when introducing new innovations. For example, it reengineered its personal pager operations so that customers could design their own pagers from the most current available combination of features. It flexibly automated its pager production line so that a customer's order would go into production within twenty minutes of receipt of the order itself, for completion within two hours. Its goal was overnight delivery of a custom-designed product.
Many companies have tried to use formal market research to target their innovation activities. Most have found that although such guidance works extremely well for incremental product line extensions or improvements on existing lines, it has severe limitations in guiding early-stage research or innovation. And formal market analyses are frequently quite misleading when applied to radical innovations. Among many classic examples are the following:
Studies said Intel's microprocessor "would never sell more than 10 per cent of the mini computer market." But within two years, microprocessor-installed power was greater than the combined power of mini-and microcomputers. Sony's marketing researchers initially said its hugely successful transistor radios, mini-television sets, and Walkman concepts "would fail in the marketplace" because the public and always associated small size with low quality. Conversely, many highly touted, superb technical concepts (like hovercraft, bioengineered interferon, Pilkington's 10-11 shatterproof glass, and IBM's Josephson-effect devices never fulfilled their carefully estimated market potentials. Many of industry's largest failures were actually over-planned and over-researched on paper (like Ford's Edsel, IBM's FS System, and Federal Aviation Administrations supersonic transport) rather than interactively developed with genuine customer feedback.
Frequently neither innovators, market researchers, nor users can quite visualise a totally new product's real potential. Initial market success often depends on highly opportunistic interactions that uncover unspecified or unsuspected needs.
James Gosling of Sun Microsystems originally wrote Java as part of a failed effort for Sun to pioneer equipment for interactive TV. The Global Positioning System (GPS) was created to locate military forces, not direct autos, emergency units, and farm tractors or locate hijacked or lost trucks and pleasure craft. Recording English-language programs for Japanese schools and songs in Japanese bars provided the first commercial markets for Sonys audiotape recorders. The Hovercraft became profitable only when its original air cushion and amphibian use concepts were abandoned because ferryboat operators insisted that rigid sideskirts and direct power drives be extended into the water.
Innovative companies try to structure interactive user participation as early as possible in the innovation process. They both improve results and lower risks by incorporating some of the users innovative ideas and by foreseeing what the product's actual use, abuse, and desirability characteristics may be.
Sam Walton said that 90 per cent of his company's best ideas came from contact personnel's suggestions. Hewlett-Packard, 3M, Microsoft, Sony, Wal-Mart, Rubbermaid, Raychem, and Nintendo-among the most successful and sophisticated diversifying innovators we encountered-generally introduce radical new products through small teams that closely participate with users, learn from them, and rapidly modify designs and entry strategies around this information. In the past, these interactive contacts have been largely on a personal basis. Increasingly, critical information is captured through groupware and intranets, which allow instant and complete documentation of questions, problems and solutions.
To meet its customers customisation needs, Toshiba has created seven software factories. Each develops and produces software for one of seven application areas. Each factory produces a variety of programs for customers using standardised procedures, tools, and components reusable across different customer projects. Interestingly, Toshiba found that innovative successes mainly depended on quality; customers and profits were not much influenced by the impact of "efficiency" improvements in the software development process itself. However, by concentrating on the quality interface, Toshiba was able to increase efficiency, flexibility, and customer value simultaneously. Electronic networks and virtual workspaces have created entirely new ways in which customers can be active team members of innovation.
Multiple approaches
All innovation is probabilistic. No one knows at first whether the desired result can be achieved at all or which of several possible approaches may work best in practice. Generally no one knows precisely when the result will be achieved, what competitors will accomplish in the meantime, or how customers and potential suppliers will react to these competitive offerings. Finally, no one can predict quite how customers will use the innovation when it occurs or which approach will dominate the field as it develops. Initial insights can come from almost anywhere, and the right insight can create a whole new industry or offer dominance to its discoverer for years. Several well-known innovations provide examples:
A gust of wind, blowing mould on Fleming's cultures, created the antibiotic age. Sir Alastair Pilkington was washing dishes when he got the critical insight that led to float glass. The microcomputer was born because Ted Hoff happened to work on a complex calculator for a Japanese client just when DEC's PDP8 architecture was fresh in his mind Carl Djerassi's group at Syntex was not looking for an oral contraceptive when it created 19 norprogesterone, the precursor to norethingdrone. which became the active ingredient in most early contraceptive pills. DARPA developed the Internet primarily to allow laboratories to share information and to utilise large-scale computers more efficiently.
Lucky accidents are involved in almost all major technological advances. A soldering error led to the n junction transistor. A researcher looking for a blocker for angiogenesis induced blindness discovered the most promising potential intervention for preventing, and perhaps curing, cancer currently available. And so on. In fact, Murphy's Law works because scientists work within their paradigms and engineers design for what they can foresee; hence what fails is what theory did not predict or what human error introduced. Even the most sophisticated testing may not define all the interactions of components and subsystems over the complete performance envelope of a complex device's anticipated operations.
Elaborately designed ring seals failed on the Challenger space shuttle, thoroughly tested jet engines failed in an Iranian sandstorm during the US attempt to rescue its Embassy's personnel held as hostages, and patients were horribly burned when a minor software change was made during X-ray tomography treatments.
While software models now enable innovators to manipulate and handle many more such variables, ultimately there must be tests of the full system under multiple actual operating conditions and provision for feedback and correction of errors and use problems at all stages of innovation.
Recognising these characteristics of development, most innovative companies in our sample consciously started multiple component and prototype programs in parallel, thinning down the number of alternatives only as probabilities of success approached certainty or cost became prohibitive.
Sony in its videotape recorder programme pursued ten different major options, each with two or three subsystem alternatives. It now creates technological pools of tested subsystem and component options it can introduce quickly in response to new market demands or innovative concepts. Having multiple options available allows it to introduce a new Walkman every two weeks and a new camcorder every ninety days.
Planned redundancy and flexibility clearly can help cope with some of the inherent unpredictability in technical develoment and early market introduction. Multiple development approaches, especially those that can be implemented and tested in software, also encourage testing of more options interactively with customers. Constructive competition of ideas helps motivate groups, shorten cycle times, and improve the quality and reliability of results.
Objective performance shoot-outs
To maximise the motivational and informational benefits of multiple competing approaches, many companies consciously structure performance "shoot-outs" between the competing approaches of various teams after their projects reach advanced prototype stages. Many shootouts can be performed in software. But physical prototype comparisons are needed to test subjective or tactile elements like taste, smell, feel, sound, visual, entertainment, physical access, or ergonomic features. Ultimately, of course, the product or process must be given a physical test under actual use conditions. Because of their expense when used, physical tests should be designed to maximise the level of objective data available for decision purposes by allowing teams to critique each others approaches, minimise risks by making tests as close as possible to the actual marketplace, and stimulate each multifunctional team's commitment to the point that it is ready to move ahead immediately if its option wins.
Although anathema to many who worry about presumed efficiencies in R&D, multiple approaches, by allowing greater effectiveness in choosing a right solution with commitment behind it, can easily outweigh any duplication costs, especially when early-stage shoot-outs occur in software. Multiple prototyping and performance shoot-outs are likely to be warranted when the market rewards significantly higher performance or when large volumes justify increased technical or cost sophistication. Under these conditions, competing approaches can improve probabilities of success, decrease development times, and decrease costs in both the short and the long run.
Properly rewarding and reintegrating losing teams is perhaps the most difficult and essential skill in managing competing projects. If the total company is expanding rapidly or if a successful project creates a substantial growth opportunity, competing team members can generally find another interesting program or sign on with the winner as it moves toward the marketplace. for the shoot-out system to work continuously, however, executives must create a climate that honours high performance, whether the performers' specific approach wins or loses; reabsorbs people quickly into their technical specialities or onto other projects; and accepts and excepts rotation among tasks and groups.
Smaller, more flexible skunk works units
Most successful mid-level innovation managers in our studies tried, whenever possible, to use relatively small units and flat organisations to stimulate commitment, interactiveness, and rapid response to user needs. Most thought the optimum number of key players per team was from three to seven, although it might have to be much larger for highly complex systems requiring a variety of skills. This small size generally seemed to provide a critical mass for accessing needed skill bases, foster maximum communications and flexibility, and allow reasonable latitude for individual creativeness and commitment. Although software facilitates larger team sizes, higher numbers tend to interfere with personal communications.
The first skunk works, named after "Kelly" Johnson's successful group at Lockheed, were merely small teams of engineers, technicians, and designers working together with no intervening organisational barriers. Now companies may put innovative production, marketing, worker, supplier, customer, corporate staff, and post-sale service people on the team, as Ford did in designing the original Taurus-sable. But vastly increased software support is needed to maintain communications and technical cohesion as team sizes expand. Even in the strong planning-oriented cultures of Europe and Japan, the most innovative groups we observed used skunk works to improve their innovation potentials. Few were as thoroughly developed as Ford's original Team Taurus. Nor, interestingly, was Japan's much publicised ringi decision-making process evident in their skunk works situations.
Sony's founder, Ibuka, and his technical successor, Kihara, participated directly with their design groups and made rapid on-the spot decisions at key junctures, as did successor CEOs Morita and Ohga. Nintendo's CEO, yamauchi, personally makes the decisions on which games and features are acceptable for his company. Soichiro Honda was known for working with his engineering and design groups and often emphasising his own views by shouting. Face-to face discussions among different technical and management levels in these companies are common, with constructive confrontation encouraged at the design level.
Why do such direct interactions work so well? Progress is technology is largely determined by the number of successful experiments made per unit time. The biggest cause of delays in large enterprises are bureaucracies and hierarchical organisations' effects on decision cycles. Skunk works help flatten organisations, eliminate bureaucratic delays, allow fast and unfettered communications and permit the quick turnarounds and decisions that multiply experiments and stimulate rapid advance. Groupware, virtual workspace, or modelling software that enables diverse and remotely located team members to interact around a common problem can increase advance rates even more.
Flat, externally leveraged organisations
To encourage fast, effective decision making, innovative middle managements consciously try to create very flat organisations with as few organisational layers between bench participants and top decision makers as possible. By keeping total division sizes below 400 people, they enforce a maximum of two intervening decision layers to the top. In units much larger than this, people quickly lose touch with the total concept of a product or process, bureaucracies grow, and projects must go through more and more formal screens to survive. The improved communications and discipline that software imposes on design groups enables flat organisations to function more effectively than ever. With proper organisational software, a number of small teams can operate in parallel on a single large, co-ordinated project, as they did on Boeing's (2800 engineering location) 777 aircraft or Ford's (400 subgroup) first Taurus-Sable project. Before attempting such wide-span projects, middle managers should make sure that appropriate software, culture, performance tracking, and incentive systems are firmly in place.
The most innovative enterprises also consciously tap into multiple external sources of technological capability. No company or research group can spend more than a small portion of the world's (approximately $300 billion) R&D budget. Yet scientific knowledge and technological advances from outside sources (and seemingly unrelated fields) often interact to create totally new concepts or opportunities important to the enterprise. Recognising this, many concerns have active strategies to develop information for trading with outside research or technology groups, and special teams and software to tap these sources. Often these programs, typically implemented through gatekeepers of technical exchange, are responsible for initiating a very high percentage of successful innovations in constantly innovative organisations. In a textile company we examined, one man found and introduced 90 per cent of its major innovations; in a smaller chemical company over 70 per cent; and so on. At a minimum, such strategies ensure
that the company stays intimately in touch with its most important technical and user communities.
Increasingly, well-designed software is a key element in monitoring and communicating with these sources. Such leveraging relationships have expanded substantially in recent years, especially in larger companies that participate (as joint ventures, consortium members, limited partners, guarantors of first markets, venture capitalists, or spin-off equity holders) in a variety of projects. Such leveraging relationships now rival the imagination and variety of structures used in entrepreneurial start-ups.n
Innovation explosion
By James Brian Quinn, Jordan J Baruch and Karen Anne Zien
Published by The Free Press, New York
Distributed by India Book Distributors
Pages 432; Price: $28.
