We design an intervention to foster social ties at work and evaluate its impact on performance and retention. We run a cluster-randomized field experiment in a large microfinance firm, providing small subsidies for geographically spread offices to organize biweekly social activities over three months. The intervention increases perceived collegiality and workplace friendships by about 0.2–0.25 SD. Individual productivity is unchanged, but office-level team performance in the firm’s competition improves in the final intervention month and employee turnover falls by about 4–4.5 pp from a 9–13% baseline in the following months. The pattern is consistent with bonding mitigating free-riding in team tasks and raising job utility as a non-wage amenity; survey evidence suggests an additional role for reciprocity toward the firm.