We use a theory of apologies to design a nationwide field experiment involving 1.5 million Uber ridesharing consumers who experienced late rides. Several insights emerge. First, apologies are not a panacea-the efficacy of an apology and whether it may backfire depend on how the apology is made. Second, across treatments, money speaks louder than words-the best form of apology is to include a coupon for a future trip. Third, in some cases sending an apology is worse than sending nothing at all, particularly for repeated apologies and apologies that promise to do better. For firms, caveat venditor should be the rule when considering apologies.
