This paper studies sample design for process control in principal-agent settings where deterrence rather than ex post detection is the main issue. We show how the magnitude of gains from additional sampling can be calculated and traded off against sampling costs. It is shown that the optimal...