Abstract The informativeness principle states that a contract should depend on informative signals. This paper studies how it should do so. Signals indicating that the output distribution has shifted to the left (e.g., weak industry performance) reduce the threshold for the manager to be paid; those indicating that output is a precise measure of effort (e.g., low volatility) decrease high thresholds and increase low thresholds. Surprisingly, “good” signals of performance need not reduce the threshold. Applying our model to performance-based vesting, we show that performance measures should affect the strike price, rather than the number of vesting options, contrary to practice.