Observers claim that public policies fail ‘often’. This paper asks, ‘how often’? It is an important question, because public policies absorb resources to address major social issues. We should know if policies are proving bad social investments; routinely failing to solve focal problems at high costs. Unfortunately, it is not easy to assess this. Many public policy organizations—governments in particular—do not provide accessible views onto overall success or failure. The World Bank does, however, provide such view—and it supports policy interventions one finds in governments across the world. The paper thus examines World Bank failure rates. It finds that there are different answers to the ‘how often’ question, depending on responses to a second question, ‘what is failure anyway?’ In studying both questions, the paper identifies a bias in the World Bank—and probably all organizations adopting rational ‘plan and control’ policy processes—to measuring ‘project and product success’ rather than a broader view of success as ‘problems are solved with development impact’. This means that policy organizations like the Bank judge success based on whether planned products are delivered through an efficient process; not whether policies solve the problems that warranted intervention in the first place, or whether the policies promoted development outcomes. Is this how citizens would want their public policy organizations to conceptualize success?