Spillover experiments often fail due to poor treatment distribution. By using a recentered instrumental variable estimator and optimizing your assignment strategy to match your network structure, you can double your effective sample size and gain clearer insights into indirect policy effects.