Consider the two faces of the global innovation movement. Company A, having grown through acquisition, produces multiple brands for multiple markets and operates a worldwide network of research and product development centers. Each of its R&D sites was initially responsible for its own brands and local market, but with globalization these distinctions have lost their importance.

Company B, on the other hand, was built largely through internal growth and has two global brands. It operates one primary R&D center supported by a handful of special-purpose sites around the world. This comparatively sparse network has helped Company B win wide admiration for the efficiency of its engineering.

Because expanding the number of nodes in a network exponentially increases its complexity, it is not surprising that Company A’s R&D structure is more expensive to operate. Company A has considered closing some sites, but has resisted doing so because it fears losing capabilities and insights, and roiling local markets. Meanwhile, incremental budget cuts have chipped away at engineer and supplier morale. Having built its network to maximize the value associated with market access, it is now forced to manage the network for cost.

Most global innovation networks look like Company A’s — and suffer the same problems. Company B’s R&D structure is clearly more productive, but it is not necessarily ideal either. Its network might be too compact, limiting access to knowledge that could maximize performance. Thus, to identify principles and practices for creating a truly well-designed innovation network, Booz Allen Hamilton and INSEAD, the international business school, surveyed R&D leaders in 186 companies from 17 industry sectors in 19 nations in 2005. The survey results, and our own experience, suggest one central truth: Organizations benefit when they configure their innovation networks for cost and manage them for value. (For an in-depth look at the survey results, see www.strategy-business.com/media/file/global_innovation.pdf.)

The survey respondents, who together account for nearly 20 percent of global corporate R&D expenditures, or $76 billion, clearly understand the problems that arise in overseeing a bloated, competitively disadvantaged innovation network. They named what they view as the primary R&D challenges: assessing the value of new knowledge, encouraging cross-site and cross-functional collaboration, managing the complexity of global projects, and optimizing innovation footprints. They also emphasized that having a well-managed R&D network is becoming particularly advantageous as companies expand R&D beyond their home turf. Between 1975 and 2005, the survey found, the share of R&D sites located outside the markets of their corporate headquarters has risen from 45 percent to 66 percent. That share is likely to increase, with 77 percent of the R&D sites planned over the next three years slated for China or India.

Several factors have contributed to the dispersion of corporate R&D sites. Rising costs in the West, rapid growth of markets in developing nations, advanced information technology, a scarcity of engineers and scientists, and the opening of markets in China and India have each encouraged companies to globalize their R&D efforts. Our survey suggests that future R&D sites in Western Europe, the United States, and Japan will be selected primarily because they offer value such as proximity to technology or research clusters, to markets or customers, or to qualified workers commensurate with their higher cost. Locations in the developing world will be chosen primarily to gain access to local markets, to decrease costs, and, particularly in India and Eastern Europe, to tap into a pool of highly qualified workers.

But choosing the right location isn’t an easy task for R&D leaders. Because it’s critical that leaders be able to justify the high cost of knowledge access in a developed market location and of operational efficiency in a developing market location, the process must be managed strategically. Without fully understanding the calculated benefits of a potential site, R&D executives are likely to incur an unjustified increase in structural costs.