In May 2018, in response to protests, Starbucks changed its policies nationwide to allow anybody to sit in their stores and use the bathroom without making a purchase. Using a large panel of anonymized cellphone location data, we estimate that the policy led to a 7.3% decline in store attendance at Starbucks locations relative to other nearby coffee shops and restaurants. This decline cannot be calculated from Starbucks’ public disclosures, which lack the comparison group of other coffee shops. The decline in visits is around 84% larger for stores located near homeless shelters. The policy also affected the intensive margin of demand: remaining customers spent 4.1% less time in Starbucks relative to nearby coffee shops after the policy enactment. Wealthier customers reduced their visits more, but black and white customers were equally deterred. The policy led to fewer citations for public urination near Starbucks locations, but had no effect on other similar public order crimes. These results show the difficulties of companies attempting to provide public goods, as potential customers are crowded out by non-paying members of the public.