Effective collaboration within and between organizations requires efficient adaptation to unforeseen change. We study how parties build relational contracts that achieve this goal. We focus on the “clarity problem”—whether parties have a shared understanding of the promises they make to each other. Specifically, (a) a buyer and seller play a trading game in several periods; (b) they know their environment will change but do not know how; and (c) before any trading occurs, they can reach a nonbinding agreement about how to play the entire game. We hypothesize that pairs whose initial agreement defines a broad principle rather than a narrow rule are more successful in solving the clarity problem and in achieving efficient adaptation after unforeseen change. In our baseline condition, we indeed observe that pairs who articulate principles achieve significantly higher performance after change occurred. Underlying this correlation, we also find that pairs with principle-based agreements were more likely to both expect and take actions that were consistent with what their agreement prescribed. To investigate a causal link between principle-based agreements and performance, we implement a “nudge” intervention that induces more pairs to articulate principles. The intervention succeeds in coordinating more pairs on efficient quality immediately after the unforeseen change, but it fails to coordinate expectations on price, ultimately leading to conflicts and preventing an increase in long-run performance after the shock. Our results suggest that (1) principle-based agreements may improve organizational performance but (2) high-performing relational contracts may be difficult to build.