We study third-party data sharing---the sale of consumer data generated as a by-product of ordinary transactions---in a framework with a monopolistic seller, a privately informed consumer, and a third party whose data use imposes privacy costs on the consumer. Our analysis highlights the role of the privacy β, which captures whether consumers with a higher willingness to pay for the seller’s product are more (β > 0) or less (β < 0) averse to greater data use. When β > 0, data sharing is coarse, pooling all consumer types whose data are sold into a single record; in this case,  greater data use lowers consumer rents. When β < 0, data sharing is instead granular, revealing individual consumer types; greater data use raises consumer rents. These differences shape the welfare comparison among three policy regimes: data anarchy, a data ban, and a data market.