Research summary: We propose a theory that explains why the relentless pursuit of perfect implementation of strategy may be useful even in a world in which the strategies being implemented are far from optimal. We formulate a computational model in which an organization's strategy adapts based on performance feedback. However, the distinctive feature of our approach is that we abandon the “organization as a unitary actor” assumption, and model a separation of beliefs and actions. The central insight is that, given this separation, precise implementation has benefits beyond the well‐known effect of enabling exploitation of good strategies. It enables the discovery of better strategies by allowing more effective learning from feedback on the value of current strategies. Managerial summary: Given the reality that the strategies coming from the C‐suite are seldom perfect, is it sensible for managers to place such a heavy emphasis on implementing them precisely? In this paper we develop a theory that explains why the answer may be “yes”. In most organizations, the formulators and implementors of strategy are distinct. Imprecise implementation makes it difficult for the formulators to learn the value of their strategies, as neither success nor failure necessarily indicates something about the value of the strategy itself, when implementation is imprecise. Copyright © 2015 John Wiley & Sons, Ltd.