Cyert and March said actual or expected failures induce a firm to make changes in its goals or forecasted outcomes. If these changes prove inadequate to allow forecasts of success, the firm searches its environment for new alternatives. Similarly, Sitkin argued that moderate failures draw a firm's attention to potential problems and stimulate it to search for possible solutions for these problems. He set forth criteria for ‘intelligent failure’ (p. 243), noting that people should plan their actions (a) to yield diagnostic information, (b) to limit the costs of failure, (c) to generate feedback quickly, and (d) to focus on familiar domains so that they will be able to analyze what happens. He also listed some organizational properties that foster intelligent failure. Burgelman (1994) also saw moderate failures as sources of information. He analyzed Intel Corporation's development of strategies as evolution in which Intel's top managers create rules for selecting among alternative courses of action. For such rules to work, the selected actions must match pressures in the firm's environment. This matching depends both on middle managers generating appropriate strategic initiatives and on top managers having “strategic recognition capability.” Failures enable top managers to learn how Intel's “distinctive competence” relates to “the basis of competition” (p. 52).
However, this idea that failure stimulates experimentation contrasts with research findings from studies of individual people indicating that unpleasant consequences generally stop existing behaviors without stimulating efforts to behave differently. As well, Hedberg (1981) pointed out that organizational inertia often delays or counteracts problem solving. Milliken et al. (1992) interpreted data about two industries as showing that most poorly performing firms do not alter their strategic orientations. Dendrell (2003) and Levinthal and March (1993) have argued that failures are undersampled both by firms and by management research.
In fact, there have been very few empirical studies of firms’ responses to failure, as both researchers and managers have been prone to record and advertise successes. For that reason, Baumard and Starbuck (2005) studied 14 strategic failures that occurred at ‘Eurocom’, a very large European company. With help from executives in the firm, they compiled data about seven large failures and a matched sample of seven small failures. Eurocom had experienced many more than seven small failures, but the researchers wanted to contrast the large failures with small ones in similar competitive environments.
The idea of comparing large failures with small ones arises from observations made in previous studies. Normann (1971), Rhenman (1973), and Wildavsky (1972) have observed that firms react quite differently to variations that would modify the firms’ strategic domains only incrementally than to reorientations that would redefine the strategic domains. Variations exploit firms’ experience, preserve existing distributions of power, and can win approval from partially conflicting political interests. Reorientations take firms outside their familiar domains and alter the bases of power, so reorientation proposals instigate struggles between power holders and power seekers. Based on several studies of firms facing crises, Starbuck (1983) proposed that reorientation usually requires changes in top management.