Optimal Search for a Team

We introduce and analyze the team search problem. A firm sequentially searches for workers to assemble a team containing a required combination of types. We characterize the expected search duration as a function of the team's and pool's compositions, and establish three main results. First, the search duration function is a linear combination of its forward marginals. Second, when the firm chooses where to search, the optimal pool is strictly more balanced than its team, generating a compositional mismatch between the firm's positions and the workers' skills. Third, when the firm can automate positions, the optimal policy is often all-or-nothing.