Kodak and Boeing are just two of the more recent examples of once dominant companies that failed to adapt to market changes. Kodak excelled at analog photography but hasn’t been able to make the leap to digital cameras. Boeing, a longtime leader in commercial aircraft, has experienced difficulties in its defense-contracting businesses and has recently stumbled in the face of competition from Airbus.

The failure to achieve breakthrough innovations while also making steady improvements to an existing business is so commonplace – and so fascinating – that it has become a battleground of management thought. For decades, scholars have spun theories to explain the puzzle and offered advice on how to solve it. Some have argued that there’s no way out of the conundrum – that established companies simply lack the flexibility to explore new territory. Some have suggested that big companies adopt a venture capital model, funding exploratory expeditions but otherwise staying out of their way. Others have pointed to cross-functional teams as the key to creating breakthrough innovations. Still others have claimed that a company may be able to shift back and forth between different organizational models, focusing on exploitation for a period and then moving into exploration mode.

Some companies have actually been quite successful at both exploiting the present and exploring the future, and as we looked more deeply at them we found that they share important characteristics. In particular, they separate their new, exploratory units from their traditional, exploitative ones, allowing for different processes, structures, and cultures; at the same time, they maintain tight links across units at the senior executive level. In other words, they manage organizational separation through a tightly integrated senior team. We call these kinds of companies “ambidextrous organizations,” and we believe they provide a practical and proven model for forward-looking executives seeking to pioneer radical or disruptive innovations while pursuing incremental gains. A business does not have to escape its past, these cases show, to renew itself for the future.

Companies tended to structure their breakthrough projects in one of four basic ways. Seven were carried out within existing functional designs, completely integrated into the regular organizational and management structure. Nine were set up as cross functional teams, groups operating within the established organization but outside the existing management hierarchy. Four took the form of unsupported teams, independent units set up outside the established organization and management hierarchy. And 15 were pursued within ambidextrous organizations, where the breakthrough efforts were organized as structurally independent units, each having its own processes, structures, and cultures but integrated into the existing senior management hierarchy.

We found that the organizational design and management practices employed had a direct and significant impact on the performance of both the breakthrough initiative and the traditional business.

When it came to launching breakthrough products or services, ambidextrous organizations were significantly more successful than the other three structures. While none of the cross-functional or unsupported teams and only a quarter of the functional designs produced real innovations, more than 90% of the ambidextrous organizations achieved their goals. (An exception was breakthrough innovations intended to directly substitute for existing products; in these instances, functional designs performed as well as ambidextrous designs.) The superiority of ambidextrous designs became even more apparent when we examined eight cases in which a company originally organized its breakthrough initiative around functional designs, cross-functional teams, or unsupported teams and then shifted to an ambidextrous design. In seven of the eight cases, the initiative’s performance increased substantially after the change. In contrast, three companies started from an ambidextrous design and then moved to one of the others; performance decreased substantially in two of these cases.

When we measured the effects of all 35 initiatives on their existing businesses, we found that ambidextrous organizations were again clearly superior. In almost every instance in which an ambidextrous structure was used, the competitive performance of the existing product either increased or held steady. By contrast, the results of the traditional operations frequently declined where functional designs, cross-functional teams, or unsupported teams were employed. At a theoretical level, it’s easy to explain why ambidextrous organizations would outperform other organizational types. The structure of ambidextrous organizations allows cross-fertilization among units while preventing cross-contamination. The tight coordination at the managerial level enables the fledgling units to share important resources from the traditional units – cash, talent, expertise, customers, and so on – but the organizational separation ensures that the new units’ distinctive processes, structures, and cultures are not overwhelmed by the forces of “business as usual.” At the same time, the established units are shielded from the distractions of launching new businesses; they can continue to focus all their attention and energy on refining their operations, improving their products, and serving their customers.