Existing theories of electoral targeting focus on voter partisanship and thus are at odds with the significant vote switching and weakly institutionalized parties that characterize young clientelistic democracies. We propose and test a theory of group-level targeting driven by groups' capacity to coordinate votes and parties' differential information about such capacity. Unlike current theories, ours assumes that most groups are nonpartisan and respond to prior targeting as a function of their coordinating capacity, often reflecting broker effectiveness. We empirically exploit the context of Senegal where new incumbent parties are less well-informed than outgoing incumbents about groups' coordinating capacity but, to maximize future support, learn from local vote tallies and refine targeting strategies over time. Using village-level electoral and public goods data, we show our theory can account for new incumbent parties' learning and targeting patterns across groups and over time, thus contributing to understanding electoral targeting in young clientelistic democracies.