More powerful managers make more important decisions. Therefore, firm performance is more informative about the abilities of such managers, who, realizing that they are more visible, are more eager to improve performance. If this reputation effect exists, how should firms allocate power? I analyze the optimal allocation of power and derive implications for several issues that often arise in management practice: the choice of departmentation criteria, the importance given to seniority, and the width of job definitions. Finally, I show that the model is consistent with the empirical evidence on managerial succession.