Targeted Information Design

This paper studies information provision by platforms in monopolistic markets. Prior to interacting with the platform consumers obtain information from some other sources- The platform then decides how much additional information to provide to consumers and can condition her information provision on both her knowledge of consumers' preferences and the information they hold at the outset. When this outside information is monotone non-overlapping, solving a relaxed problem featuring stochastic dominance constraints on posterior mean distributions is sufficient to characterize the set of all welfare outcomes. In this setting, efficiency turns out to be highest when information is designed in a buyer optimal way.