Many firms, from retailers to investment management companies, offer their own products alongside products sold by competitors. This form of vertical integration has been the target of regulation in digital markets based on concerns of harm to consumers. We study the effects of this practice on consumer choice in the context of Amazon.com. We run a field experiment using a custom browser extension that allows us to generate random variation in the set of products observable to consumers. Using this variation, we document several patterns about consumer behavior. In the absence of Amazon brands, consumers substitute toward products that are comparable along most observable dimensions, including price, average rating, and shipping speed. One exception is that the substitute products have many fewer reviews---often considered a proxy for quality---compared to Amazon brands. We find no evidence that consumers exhibit additional search effort in the absence of Amazon brands, nor that they shift their shopping behavior to other retail websites. On the supply side, we do not find evidence that Amazon discriminates in favor of its own products in search results. We estimate a structural model and use the model to quantify the channels by which private labels affect consumer welfare on Amazon.