Online, pure-labor service platforms (e.g., Zeel, Amazon Home Services, Freelancer.com ) represent a multibillion-dollar market. An increasing managerial concern in such markets is the opportunistic behavior of service agents who defect with customers off platform for future transactions. Using multiple methods across studies, the authors seek to explain this platform exploitation phenomenon. In Study 1, a theories-in-use approach clarifies why and when platform exploitation occurs and derives some hypotheses. Study 2 empirically tests these hypotheses using data from a healthcare platform that connects nurses and patients. The results indicate that high-quality, long-tenured service agents may enhance platform usage, but customers also are more likely to defect with such agents. Platform exploitation also increases with greater customer–agent interaction frequency (i.e., building stronger relationships). This phenomenon decreases agents’ platform usage, due to capacity constraints caused by serving more customers off platform. These effects are stronger as service price increases (as higher prices equate to more fee savings), as service repetitiveness increases, and as the agent’s on-platform customer pool comprises more repeat and more proximal customers. Finally, we use two scenario-based experiments to establish some managerial strategies to combat platform exploitation.