Optimal influence under observational learning

We study a problem of optimal influence in a society where agents learn from their neighbors. We consider a firm that seeks to maximize the diffusion of a new product whose quality is ex–ante uncertain, to a market where consumers are able to compare the qualities of two alternative products as soon as they observe both of them. The firm can seed the product to a subset of the population and our goal is to find which is the optimal subset to target. We provide a necessary and sufficient condition that fully characterizes the optimal targeting strategy for any network structure. The key parameter in this condition is the agents’ decay centrality, which is a measure that takes into account how close an agent is to others, but in a way that very distant agents are weighted less than closer ones.