Politicians in all democracies have goods to distribute, and they employ different modes of distribution to deliver them. They can offer voters goods in the hope those goods turn into votes. Alternatively, they can try to make the distribution of a good conditional on how someone votes. The latter mode is clientelism. I point out that the literature on clientelism has been preoccupied with the idea that politicians form clientelistic relationships with individuals. This has led to an intense scholarly focus on how politicians can consummate such vote buying deals to their satisfaction, given that the secret ballot prevents them from observing how people vote. I argue that under a certain configuration of political institutions, it makes sense for politicians to form clientelistic relationships with groups of voters. To do so, a politician’s electoral district must be divisible into groups of voters, at which electoral support is observable and to which resources are targetable. I take four longstanding questions of interest in the clientelism literature, concerning brokers, economic development, democratic integrity, and club goods and explain how the theory of group-based clientelism opens up new lines of inquiry in each.