Academy of Management Proceedings
William Schulze, Matthew Barlow, Ryan Winn Angus
Department of Entrepreneurship & Strategy
Abstract
A large body of quantitative empirical research has shown that organizations can and do learn from failure in contexts characterized by low causal ambiguity and low failure rates. However, it remains unclear under what conditions organizations or more or less likely to learn from failure when performing innovative experiments in contexts characterized by high causal ambiguity and high failure rates. This paper develops a theoretical framework to explain and predict when experimental failures are likely to help?and even potentially harm?an organization?s future chances for innovative success. This framework is tested using a unique dataset obtained from the Google Play app store. The paper finds that, on average, the more failed experiments an organization has performed in the past, the lower its current odds of success. However, the paper also finds that organizations are more likely to learn from failed innovation experiments when their past experiments were low (rather than high) in novelty and when their pace of experimentation was quick (rather than slow).
