It is notoriously difficult to provide outside parties with adequate incentives for innovation. Contests—in which solvers compete for a prize offered by the contest holder—have been shown to be an effective incentive mechanism. Despite considerable interest in this concept, we lack a thorough understanding of important aspects of contests; in particular, feedback from the contest holder to the solvers has received only limited attention. This paper discusses how contest holders can improve contest outcomes by devising an optimal information structure for their feedback policies. The authors first identify when, and when not, to give feedback as well as which type of feedback to give: public (which all solvers can observe) or private (which only the focal solver can observe). The authors uncover a nontrivial relationship between contest characteristics and optimal feedback choices. Second, the authors examine whether the contest holder should mandate interim feedback or instead allow solvers to seek feedback at their own discretion. Third, the authors discuss how changing the granularity of feedback information affects its value to solvers.