It may seem paradoxical, but the truth is that the very process of innovation can stand some innovation. Readers of this article will learn why large corporations especially have something to learn and how applying those lessons will lead to a better record on innovation and a greater commitment from employees in the lab.
Two models dominate innovation activities in the Western world: corporate R&D and the venture-backed start-up. However, the past several decades have seen numerous changes in the nature of emerging technologies and the ways that innovation is pursued. Underlying these changes are the widespread understanding and use of the Internet and the explosion in computing power. These facts alone might lead us to wonder whether institutions developed in the 19th or mid-20th centuries are still the right answer for developing innovations.
This question is particularly critical when it comes to corporate R&D. Even today, the lion’s share of R&D spending continues to take place, in industrial laboratories, not in startups or with entrepreneurs, as many might guess. This was true in 1953, when private industry performed 70 percent of the research, and is still true, at least it was as of 2008, when it accounted for 74 percent of the spending.[i] While there have been changes over this period — for instance, academic research has sharply surpassed government-performed research in this period —the basic pattern has been one of stasis rather than of change.
Research today, as during the 20th century, is still dominated by the very largest of firms.[ii] Corporations with over 10,000 employees still account for more than half the research spending. Those with fewer than 500 employees, a traditional definition of small a business, account for less than one-fifth of the total expenditures.[iii] These same patterns hold internationally.[iv]
However, since the 1970s there has been a growing dissatisfaction with the “bang for the buck” approach that corporations have enjoyed with respect to their research facilities. Some firms have responded by changing the way research laboratories were organized and managed. Many corporations started emphasizing divisional laboratories and alliances in favor of central facilities. But in many cases, these efforts have merely shifted the locus of the problem of disappointing research productivity rather than introducing fundamental reforms, such as linking the compensation of corporate researchers more closely to their success. This is the main reason why it is natural to ask whether corporate R&D can be made more effective. In particular, can the corporate research laboratory learn from the venture capital sector and elsewhere? In this essay, I argue that the answers to these questions are a resounding ”Yes.”
Why venturing does it better
One of the key lessons from the venture sector has to do with staying power. The commitment that an institutional investor makes to a venture fund is a binding one. Even if the limited partner contributes a small amount of the total capital promised at the time of closing, there is an expectation that the total amount promised will be provided. Even during the depths of the financial crisis, it was rare for investors to walk away from these commitments.
Now contrast this experience with that of corporate venturing. Companies have been all too fickle in their commitment to new innovation initiatives. Often, simply the accession of a new senior officer—a replacement CEO, chief financial officer, or R&D head—has been enough to trigger the abandonment of earlier efforts: it is almost a corporate ritual to discard the pet projects of one’s predecessor!
This historical lack of commitment in the corporate venture domain has, of course, real consequences. Employees are less likely to join a corporate venturing group they fund, entrepreneurs are reluctant to accept their funds, independent venture funds are hesitant to syndicate investments with these groups, and corporate funded start-ups find collaborations harder to arrange. In each case, the very real possibility that the rug will be pulled out from under the corporate venture initiative leads others to be reluctant to work alongside them.
Any suggestion that a corporation should make a formal commitment to a venturing initiative might seem counterintuitive to many. Isn’t the flexibility to stop a program good thing? Why make it difficult to change strategy? Moreover, in some cases, opportunistic entrepreneurs have ended up suing their corporate venture backers when their ventures failed, even if the failure simply reflected the technological challenges encountered along the way. These suits often seem motivated by a cynical calculation that the deep-pocketed corporation would rather settle than undergo a bruising and uncertain fight. Commitments like the ones recommended here might leave the corporation more vulnerable to such actions. But despite these costs, the benefits from these steps are likely to be substantial. By highlighting their commitment to these programs, corporations can help address the concerns engendered by the faddish behavior in the past, and attract a higher caliber of partner to the effort.
A second area where corporations should borrow ideas from the world of venture funds relates to the design of compensation schemes. Corporations face a major challenge when shaping reward structures for those who undertake and supervise innovation activities. Traditional rewards that link payments to short-run performance may lead to unwillingness to take chances. But ultimately, given the difficulty of supervising innovative projects, having the power of long-term incentives seems essential. Moreover, in an era like today’s, where high-powered incentives for innovators are increasingly available elsewhere, an organization that does not hold out the possibility of substantial rewards would likely lose its best innovators.
A closely related point where the world of venture capital should provide a model for corporations is in its willingness to embrace (the right kind of) failures. Venture-backed projects have a wide range of outcomes, and worthy failures are celebrated in Silicon Valley as an essential rite of passage for an aspiring entrepreneur.
Far too often in corporations, on the other hand, any hint of a reversal is shunned. A corporate venturing team that has operated for several years will proudly boast that they have not yet shut down any of the firms in their portfolio. While their corporate superiors may interpret this fact as a sign of success, it is highly likely that they are funding some moribund firms—the “living dead,” in venture parlance—whose funding should have been cut off years before.
New models for innovation
Numerous models ensure the combination of short-term security (to ensure experimentation) and long-term rewards (to motivate hard work). Certainly, the traditional venture capital process embodies many of these features. Corporations would do well to model their internal venturing efforts on the approaches of independent groups.
A more general problem faced by corporations is figuring out how to offer appropriate rewards for those within research laboratories. Designed incorrectly, the wrong rewards can distort behavior. The ideal reward structure will both encourage progress in areas critically important to the firm and ensure that if multiple team members are necessary for working on a project, all contributors will be rewarded.
The key is removing some of the stigma from the failure for corporate innovators–something that is easier said than done. Certainly, the R&D department of a firm where researchers are forgiven for every misstep and lack accountability will be eventually be dominated by malingers and incompetents. This is why linking rewards to performance is a critical step, but over a time span, which allows room for experiments, some of which are almost certain to be unsuccessful.
One crucial, though often neglected, point is that such tolerance for failure requires not just a rethinking of compensation schemes, but also of how projects are selected and funded. One of the reasons why failure is not an option in many corporate laboratories is that group leaders are loath to endanger continuing funding for their projects. A question which would reward both further research by economic theorists and real-world exploration is how to induce “truth telling” when evaluating high-risk innovative projects. Several examples exist of ways to address this problem, from venture groups who employ a “devil’s advocate” to make the case why a proposed investment should not be undertaken, to corporations who rely heavily on outside experts when making project funding decisions.
One fascinating example is the changes with Glaxo SmithKline’s laboratories beginning in 2008.[v] The company—one of the world’s leading pharmaceutical firms—dramatically restructured its traditional slow-paced, bureaucratic system of R&D to emulate the relatively fast pace and entrepreneurial system of biotech companies.
These changes were a reaction to the centralization that had characterized pharmaceutical R&D in recent decades. By breaking the research efforts into autonomous, highly accountable, and extremely specialized units consisting of 7 to 8 researchers, firms sought to encourage more collaborative thinking and interactions than the traditional centralized research facility would allow.
In setting up this program, the firm also instituted a much tougher review process for deciding which project to fund. In the new model, each small lab-scale unit would “pitch” its program to a board composed of both internal senior R&D executives and outsiders, including a banker, a leading scientist, a venture capitalist, and biotech CEO. At the end of a three-month review process, successful units would be awarded three years’ investment money, a sum that often totaled hundreds of millions of dollars. In the ensuing years, the review board would track each unit’s progress against various performance and value-creation metrics and could terminate funds if progress was lagging.A related change was to push the units to rely heavily on outside resources, essentially creating competition between GSK’s internal units specializing in formulation of new drugs, animal studies, and the like, and external service providers. While the long lead times associated with drug development mean that a definitive assessment of these changes can only be made in the future, preliminary indications are promising.
The importance of knowledge transfer
Another important lesson relates to the importance of investing in knowledge transfer. Since the pioneering work of Wes Cohen and Dan Levinthal,[vi] students of innovation have understood that “absorptive capacity,” or an ability to learn from others, is a critical success factor. But far too often, the individuals running a corporate venturing unit or a business development unit may learn valuable information, but it does not get conveyed back to the operating units in a timely or usable manner. Because of this neglect, many of the most important potential gains from these efforts go to waste.
The barriers that prevent such transfers of knowledge are easy to understand. The corporate venturing and business development groups may be located far from the firm’s operations. The former groups are likely to be dominated by young MBAs, while the operating managers may be far more seasoned engineers. The fact that corporate life—particularly in the current era of austerity—overwhelms managers with day-to-day responsibilities may dull the focus on building ties with others in the organization as well. Finally, many of the technologies being developed by the alliance or venture partners may be sufficiently protean that their applicability within the corporation is hard to discern.
Corporations have tried several approaches to overcome these barriers. One approach has been to put an operating manager on the board of each portfolio firm. The rationale was that these individuals, by engaging with the younger firms on a regular basis, would become conduits of knowledge back into the firm. Alas, such efforts have typically failed. A manager running a 2,000-person refrigerator assembly plant is unlikely to have much time or the needed skills to add value to a ten-person start-up!
A more successful approach has been to create a dedicated unit devoted to transferring the knowledge from the alliances and corporate investments back into the firm. Probably the best illustration along these lines is the Central Intelligence Agency’s venture capital program, In-Q-Tel.[vii] Founded in 1999 to acquire greater access to novel technologies on behalf of the U.S. intelligence community, the fund primarily made equity investments in cutting-edge technologies developed by young firms. Often, these firms had developed products for the private sector without considering their national security applications.
But the knowledge transfer challenges were particularly challenging here. It can be difficult for an outsider to identify which technologies the intelligence community would find useful. Once the relevant technologies are identified, they often need substantial adaptations. Moreover, communications between the start-up executives and the product developers within the Agency were severely constrained by limits on communicating classified information.
To address this challenge, In-Q-Tel adopted a bipartite structure. The firm’s Silicon Valley-based venture team consisted of approximately 12 employees, closely mirroring a traditional group where general partners and associates scouted deals, performed due diligence, prepared term sheets, and shepherded portfolio companies. At the same time, the company’s 20-employee technology team worked in Arlington, Virginia. The latter group focused their efforts on assessing new technologies, testing the appropriateness of portfolio firms’ technologies for the Agency, and interacting with intelligence officials.
The problems that In-Q-Tel faced may be extreme, but they underscore a fundamental and often-neglected principle: knowledge flows from start-ups to large organizations are far from automatic. If corporations really want to enjoy the benefits from outside research, they cannot leave the capturing of knowledge flows, sometimes called spillovers, to chance. Only by investing in people who focus on capturing knowledge is learning likely to happen.
Underlying many of the arguments above is a final key point: the need for organizational experimentation. Far too often, the corporate world does a poor job of learning from the past: earlier initiatives, if unsuccessful, are forgotten about, and the key architects dismissed or exiled to the Kazakh subsidiary. The outcome is highly predictable: many firms seem destined to repeat the past, making the same mistakes in pursuing innovation again and again. An illustration is the experience of General Electric, which repeatedly began corporate venturing efforts, only to abandon the efforts after the venture teams left due to frustration over the disconnect between their broad responsibilities and modest compensation levels.[viii]
The basic message is a very simple: despite the academic and real-world insights of recent years, many aspects of the innovation process remain poorly understood. Rather than sticking to one time-honored route of pursuing new ideas, exploring the impact of different organizational structures—whether internal skunk-works or formal corporate venturing initiatives—is likely to be a recipe for success.
To the extent possible, comparisons of new and old approaches should rely on randomized designs: that is, they should compare the innovative output from one set of projects (which employ one approach) to another group that employing an alternative structure. For instance, one set of software programs may be written in house while another set is developed by outside programmers in a contest setting. The subset of projects that receive each approach would be randomly chosen. A simple randomized design like this is the scientifically most reliable method, and it is often the most cost-effective method for estimating the impact of a new organizational structure, as it allows us to learn the maximum amount by affecting only the smallest number of researchers. This is the reason that randomization is so commonly used for testing new drugs and educational approaches where important interventions are evaluated. There is no reason that randomized approaches — if they are used to evaluate the safety of innovative products and processes — cannot be used to assess the ways that innovation itself is organized.
Adapted and reprinted by permission of Harvard Business Review Press. Excerpted from The Architecture of Innovation: The Economics of Creative Organizations. Copyright 2012 Harvard Business School Publishing Corporation. All rights reserved.
[i] While we don’t have comprehensive information, the role of industry R&D was about the same magnitude before World War II: estimates by David Mowery and Nathan Rosenberg suggest that corporations undertook at least two-thirds of the R&D in the United States during the 1930s, and even more beforehand. Technology and the Pursuit of Economic Growth, New York: Cambridge University Press, 1991. Technically, they compute (unlike the earlier data) who provided the funding for the R&D, not who undertook it. Because federal funding of corporate research was quite modest before World War II, however, these two shares should be quite similar.
[ii] Based on U.S. National Science Board, op. cit., Appendix Table 4-12, and similar tables in earlier editions.
[iii] This analysis looks only at the spending of non-Federal funds, as the corresponding data with all research spending has a number of omissions to preserve confidentiality that preclude year-to-year comparisons.
[iv] This table is based on corporate securities filings in Standard & Poor’s, “Compustat North America,” http://www.compustat.com, Thomson Reuters “Worldscope,” http://thomsonreuters.com/products_services/financial/financial_products/a-z/worldscope_fundamentals/, and only includes public firms (all sites accessed July 18, 2011).
[v] Sarah Houlton, “Small Is Beautiful for GSK Drug Discovery,” Chemistry World, July 25, 2008, http://www.rsc.org.ezp-prod1.hul.harvard.edu/chemistryworld/News/2008/July/25070805.asp (accessed September 17, 2011); Toby Stuart and James Weber, “GSK’s Acquisition of Sitris: Independence or Integration?,” Harvard Business School Case no. 809026, 2009; and Jeanne Whalen, “To Innovate, Glaxo Brings Biotech Precepts In-House,” Wall Street Journal, July 1, 2010, A5.
[vi] “Absorptive Capacity: A New Perspective on Learning and Innovation,’ Administrative Science Quarterly, 35 (1990), 128–152. For a perspective on how the understanding of this phenomenon has evolved in recent years, see Kwanghui Lim. “The Many Faces of Absorptive Capacity: Spillovers of Copper Interconnect Technology for Semiconductor Chips,” Industrial and Corporate Change, 18 (2010), 1249–84.
[vii] This account is based on Kevin Book, Felda Hardymon, Ann Leamon, and Josh Lerner, “In-Q-Tel,” Harvard Business School Case no. 804146, 2005; Business Executives for National Security, Accelerating the Acquisition and Implementation of New Technologies for Intelligence: The Report of the Independent Panel on the Central Intelligence Agency In-Q-Tel Venture, Washington, BNES, 2001; and Steven Levy, “Geek War on Terror,” Newsweek, 143 (March 22, 2004), E6-E12.
[viii] This account is drawn from, among other sources, Stephen Barr, “Bright Lights, Big Paychecks,” CFO, 16 (March 1, 2000), 115-17; G. Felda Hardymon, Mark J. DeNino, and Malcolm S. Salter, “When Corporate Venture Capital Doesn’t Work,” Harvard Business Review, 61 (May/June 1983), 114-120; and Venture Economics, “General Electric,” Corporate Venture Capital Study, Unpublished manuscript, 1986.