Fast learning and sustained exploration: The role of timely performance feedback

Abstract

How should organizations manage learning dynamics? Strategy theories suggest “more-is-better”—fast, frictionless sharing enhances performance—while organizational learning theory warns that “less-is-more,” as fast learning causes premature convergence. We reconcile this tension by showing that the “less-is-more” prediction depends critically on a key assumption in classic computational models: that information about agents' performance is not continuously updated. When performance information is timely, fast learning enhances exploration. The mechanism is Target Diversity: fast learning allows many individuals to rapidly reach the performance frontier, increasing the set of imitation targets. Organizations thus learn from a diverse, shifting set of targets. The implication is that organizations achieve superior performance not by restricting information or slowing learning, but by making data on choices and performance available more quickly. Innovation relies on recombining diverse knowledge, yet facilitating this is challenging. Organizations often encourage copying stars with established track records or reputations, but this can lead to suboptimal results. We demonstrate a superior approach: provide up-to-date performance data and spotlight emergent top performers, regardless of their history. When feedback is timely, rapid learning allows “underdogs”—employees starting from lower positions—to quickly catch up to the frontier via unique knowledge combinations. Spotlighting these emergent successes creates “Target Diversity,” providing the organization with a continually renewed and diverse set of imitation targets. Such a design enhances the exploitation of diverse knowledge and improves long-run performance.