The problem with such approaches is that they cannot convert an inherently uncertain process into a predictable one. The analogy with manufacturing breaks down because, with manufacturing processes, the list of components and the specific steps are known in advance. For routine innovations, utilizing familiar technical and business concepts, that may be true. The job of the phase-gate or similarly scripted process is to make sure the organization executes what it already knows how to do. But, once we enter the realm of transformative innovation, where the underlying “ingredients” may be unfamiliar, then no amount of scripting the process removes the underlying uncertainty. In fact, imposing a structured process on an inherently uncertain one will only make matters worse by providing a false sense of control.
Because integrating diverse streams of knowledge is inherently uncertain, it requires a fluid process designed around rapid experimentation, iteration, and learning. Experimentation, iteration, and learning (from failure) are prominent themes in cases of both transformative technological innovation and business model innovation. You may recall from Chapter 5 the “suspend belief” principle at Flagship Pioneering. But, at Flagship, initial venture hypotheses are tested relatively quickly with rigorous experiments. The point of the experiment is partly to assess validity but partly to point the way to different ideas, what Afeyan described as converting a “what if” to “something that sounds more like ‘it turns out that you can do this.’” This theme of iteration appears in other highly innovative organizations. Ed Catmull highlighted the central role of iteration at Pixar in his book Creativity, Inc.: “Creativity has to start somewhere, and we [at Pixar] are true believers in the power of bracing, candid feedback and the iterative process—reworking, reworking, and reworking again until a flawed story finds its through-line or a hollow character finds its soul.”7
Fleming’s case study of Hewlett Packard’s creation of the first commercially viable ink-jet printer, a very significant innovation, highlights how the fluidity of HP’s labs’ process enabled rapid exploration of diverse technical concepts. Early prototyping was critical to testing which combinations of materials and concepts were technically viable. One participant he interviewed explained how a lead engineer on the program was “able to very quickly explore a multitude of prototypes by working informally with his friends in the integrated circuit laboratory.”8 By being able to test prototypes rapidly, the team was able to iterate through more potential design solutions to critical technical problems.
Business models also evolve through iteration. There is always a temptation to think that great entrepreneurs see the future perfectly and “nail it” with their first business plan. Nothing could be further from the truth. Amazon’s business model has evolved immensely since the company’s humble founding in 1994. Amazon’s first attempt to create a market for third-party sellers failed. As Jeff Bezos recalled about that site, “I think seven people came, if you count my parents and siblings.”9 Its second attempt, an auction site called zShops, also failed (“again, no customers,” in Bezos’s words). It was not until the company began to let third-party retailers compete on the Amazon site that volume began to accelerate. Likewise, at Google’s founding, the initial business model was not entirely clear to Larry Page and Sergey Brin (or their investors). The model evolved through experiments, experience, and, yes, some mistakes too.
Companies often squash iteration through processes that essentially require you to “meet your plan.” In one pharmaceutical company I consulted for, each therapeutic area for drug R&D (cancer, cardiovascular disease, etc.) had to create a five-year plan detailing the kind of programs it wanted to pursue, the critical science it wanted to explore, the talent it required, and the external partnerships that were necessary for success. There is nothing wrong with asking questions about these issues or requiring a rigorous degree of planning for the future. Every two years, the heads of these groups were required to report to executive leadership on their progress. Again, there is nothing wrong with interim reviews, check-ins, and progress reports. The problem was that the groups were evaluated on how well they were tracking their plan. To not pursue the projects, technology, or partnership detailed in the original plan was considered a problem. This process assumes that you can plan everything up front—correctly—and that course corrections are a sign of failure, rather than the natural result of tackling highly complex and uncertain challenges. In essence, the process left no room for learning. It is hard to come up with transformative innovations this way.
Organizational Structure: Create Permeable Boundaries
There is a theory in the innovation literature that the design of a product mirrors the design of the organization that created it.10 Physically inspect the product (or deconstruct the service), and you can pretty much figure out the structure of the organization that created it. You have probably endured products where the components just do not fit together either aesthetically or functionally. If you ever found yourself asking about a product, “Did the person who designed part A (say, your car’s steering wheel) ever talk to the designer of part B (say, the instrumentation panel)?” the answer is probably no. Companies that create barriers between people from diverse technical specialties, functional expertise, experience, and knowledge—like Sony or the financial services company mentioned above—snuff out opportunities for the kinds of creative synthesis that lead to transformative innovations.
Restructuring an organization is one approach to overcoming these barriers. Bell Labs is a great example of how structuring to intentionally drive cross-disciplinary integration can produce big benefits. Starting with the design of its new campus in Murray Hill, New Jersey (which opened in 1942), everything about Bell Labs was organized to drive communication across scientists and engineers from different disciplines and functions.11 The physical layout of the facilities was designed in a way that made it almost impossible for people from different departments to avoid one another.12
After World War II, the newly appointed head of Bell Labs, Mervin Kelly, replaced the lab’s discipline-oriented department structure with cross-disciplinary groups focused on specific technologies or problems.13 The solid-state electronics programs that ultimately gave birth to the transistor was one of the new programs created from this restructuring. The solid-state group contained theoretical physicists like William Shockley and John Bardeen, experimental physicists like Walter Brattain and Gerald Pearson, chemists, materials scientists, metallurgists, electrical engineers, and technicians.14 Constant exchange of ideas and spontaneous meetings to discuss problems and hypotheses were the norm. The transistor group was clearly endowed with some of the best minds in twentieth-century science—Shockley, Bardeen, and Brattain would win Nobel Prizes for their work; Bardeen would go on to win a second Nobel Prize in physics for superconductivity (he is so far the only person in history to win two Nobel Prizes in physics); and Pearson would later invent the photovoltaic cell. It is hard to imagine the transistor being invented under an organizational structure that divided the work across separate discipline-based groups. Having the right talent on board is important, of course, but meshing that talent together is what is often needed for transformative innovation.
Of course, organizational design involves trade-offs.15 No group can include every possible capability, technical discipline, or domain of expertise that might prove useful at some point. Attempting to do so would simply produce a bloated, unwieldy organization that can’t get anything done. As in building architecture, choices must be made about where to put up “walls.” And those choices will shape what the organization can do particularly well. They will shape the specific combinations of capabilities and technologies that are likely to be tried. As a result, they will influence the kinds of innovations that get created.
The problem is that, in dynamic technological or market contexts, the most attractive combinations that potentially yield transformative innovations change frequently. Putting some combination of disciplines together a
t one point in time (say, physicists, chemists, material scientists, and metallurgists, as they did at Bell Labs) might be a good idea, but what if new combinations become relevant to future innovations (say, electrical circuit engineering and process engineering, which were critical to the invention of the integrated circuit ten years after the transistor)? An organization can find itself laden with an obsolete structure. How can you prevent this at your organization?
One approach is to use “soft structures” like project teams or temporary organizations focused around specific innovation opportunities. These are like those temporary office walls or partitions than can be moved around to reconfigure space as needed. Temporary structures like project teams can be a more fluid way to respond to changes in the technological or market landscape.
The birth of Amazon Web Services is a great example of this approach.16 Today it is a $12 billion business and the leading provider of cloud computing services.17 This was a significant business model innovation for both Amazon (which, until that time, had been purely a retailer) and for the IT world. The concept, though, started innocuously enough in 2002 as a small experiment inside the company’s retail operation (Amazon.com). The initial idea was to give companies that sold products on the Amazon.com website greater control over how they presented their goods (the kinds of display photos they used, the information they provided, etc.). The thesis was that, because sellers understood their own products best, giving them control over their presentation should drive greater sales. But the only way to achieve this goal was to give third-party developers access to Amazon’s web infrastructure and the application interfaces used to display products. So Amazon decided to run a small experiment and temporarily open up part of its web infrastructure to see what third-party developers could achieve.
The results of this experiment were both surprising and dramatic. Amazon learned that developers were very interested in leveraging Amazon’s platform. As Andy Jassy, who went on to run Amazon Web Services, recounted:
It caused us to step back and wonder if something broader was going on. If developers would build applications from scratch using web services, and if a broad array of Web services existed (which we believed would be the case), then the Internet would become the operating system. We asked ourselves if the Internet became the operating system, what would the key elements be. Which had already been built, and which would Amazon be best-equipped to provide for the community? At the time we were looking at it in 2003, none of the key elements of the Internet operating system had been built. When we thought about Amazon’s strength as a technology company that had simply applied its technology to the retail space first, and what Amazon had done well over the last decade, we realized we could provide a lot of the key building blocks.18
In many organizations, this idea would have died right here. After all, what did “the Internet as operating system” or “web services” have to do with the “core” business of retailing? And, structurally, Amazon was not well positioned to exploit this opportunity. Amazon had no services division or any other divisional home for it. But this is where the soft-structure approach came into play. Early in the process, the idea was escalated to the company’s most senior management, who discussed it at a strategic planning meeting later in 2003. Amazon senior management decided to explore the concept and appointed Jassy (who had joined Amazon in 1997) to run a newly formed web services team (which was separate from retail). The team formed by Jassy was a vehicle for synthesis. It integrated sources of information and know-how from multiple sources. It tapped Amazon’s technical community to understand the kinds of problems it faced when building applications. It also gathered twenty business and technical leaders from throughout Amazon to understand which Internet applications existed and which ones should exist. It spoke to outside developers who had been actively involved in the original experiment. For each candidate service (e.g., storage), Jassy formed an autonomous subteam that combined both technical and management talent pulled from various parts of Amazon and from external sources.
Notice the contrast to my earlier description of the financial services company (or perhaps what you have seen in your own organization). Amazon used small, nimble teams that were given a high degree of autonomy (and resources) to explore and integrate ideas from different sources and to develop and test concepts. Input and talent came from various parts of Amazon. Boundaries were fluid.
I want to stress that the kind of teams we saw at Amazon are infinitely different from the kind of limp “matrixed” teams that are so common in many larger companies. Matrix teams are cross-divisional or cross-functional teams built on the underlying divisional (or functional) structure of the company. The idea behind matrix organizations is to alleviate the trade-offs inherent in the company’s structural design. So, for instance, if the company is organized around market-facing business units, it might create matrix teams for different kinds of cross-divisional integration efforts. In theory, this is not a bad idea. In practice, though, these teams usually have no real teeth. They lack budgets and authority—those usually rest firmly within the iron grasp of the existing business units. They lack dedicated people: everyone has two jobs in a matrix—the divisional or functional job and the team job, so in theory everyone has two bosses. (In reality, of course, everyone knows that the person who sets your pay is your boss.) Instead, the matrix team’s job is pretty much limited to making sure information is shared across business units. Jassy’s team was not a classic matrix team. It had authority, budget, and full accountability. It had its own dedicated personnel (picked by Jassy). Fluid does not have to mean weak.
An extreme example of using soft (but powerful) structures to drive transformative innovation through synthesis is how the Defense Applied Research Projects Agency (DARPA) organizes research.19 DARPA is a government agency (part of the Department of Defense) responsible for developing technologies with potential military applications. It is well known for seeding such transformative technologies as the Internet, very large-scale integrated circuits (VLSI), global positioning systems (GPS) for navigation, carbon composites, and others. In many ways, DARPA has become a modern-day Bell Labs, but its structure is completely different. DARPA’s current annual budget is about $3 billion. If DARPA were a company, it would rank in the top fifty of global R&D spenders. But DARPA lacks something that every major corporate R&D spender has. DARPA has no R&D lab of its own, and it employs a relatively small number of people directly. It organizes all its programs around temporary project teams composed of networks of external collaborators from both industry and academia. Each team is headed up by a DARPA program leader, a highly talented scientist responsible for assembling the right mix of collaborators and orchestrating their efforts. The program leaders own the program—they control the budget, manage relationships, deal with problems, and make all the critical decisions.
Every project has a fixed timeframe (usually two to five years), and the tenure of a DARPA program leader lasts only as long as the project. The network is fluid—external collaborators are brought in and out of the project as needs change. According to two former heads of DARPA, Regina Dugan and Kaigham Gabriel, the temporary nature of DARPA employment contracts and contracts with external collaborators enables a high degree of flexibility in pursuing innovation opportunities as both needs and technologies change:
The DARPA model allows a company to alter its portfolio of projects faster and at a much lower cost than a conventional internal research organization can. During our recent tenure at the agency, we were able to shift significant investment from programs in space and large air and ground systems to programs in cybersecurity, synthetic biology, and advanced manufacturing in less than a year.20
Again, with DARPA, as with Amazon, there was nothing weak about the soft structure. Program leaders are leaders; they are not coordinators or facilitators. They own the critical decisions, and they control budgets. Creative synthesis—pulling ideas and capabilities from different sources—i
s not easy and requires strong leadership and decision-making authority. Weak teams are not capable of the kinds of creative synthesis that leads to transformative innovation.
Which approach is right for your company? Structure your organization to integrate the right mix of skills, capabilities, and experience like Bell Labs? Or opt for more fluid soft structures like project teams like Amazon and DARPA? The answer is that both can be complementary tools in the pursuit of transformative innovation. An organizational structure is like the foundation of a house. The choices you make about its design will shape and constrain what you can do with what rests on top of it. If transformative innovation is part of your strategy, then you should think about creating structures that actively catalyze important connections among people with different technical and functional expertise, experience, and commercial perspectives. About your current organizational structure, you should ask, Does it impede the kinds of cross-disciplinary or cross-market collaborations that are critical to our future success? If so, then consider restructuring. But also recognize that you will never have the perfect structure. A shifting technological and market landscape will produce opportunities and threats that require you to quickly integrate new combinations of capabilities and skills both from inside and outside your organization. Amazon and DARPA show that soft structures—when properly resourced and led and endowed with the same authority as hard structures—can be very powerful.
Creative Construction Page 17