According to a Blue Yonder survey, nearly a quarter of U.S. consumers have purchased products based on AI recommendations, and of those, 46% returned an item in the past year. Among shoppers who returned AI-recommended purchases, the most common reason was that the product was different than described (36%), followed by incorrect size or fit (23%). This pattern differs from traditional retail returns, where incorrect size or fit is the leading reason (42%).
The root cause appears to be data quality rather than AI algorithm failure. AI assistants struggle to accurately describe products because most product information was originally written for search bars, not for machine-generated recommendations. Retailers are shipping products faster than they are preparing their data infrastructure to support AI agents. This mismatch between shopper expectations and product reality is driving significant return volumes, and consumers are responding by shifting loyalty to retailers with better return policies—over half of surveyed respondents said they choose retailers based on returns policies, and 29% have stopped shopping at retailers that tightened return policies in the past year.
Trust in AI recommendations remains fragile, with 89% of shoppers not fully trusting AI suggestions and a third reporting regrettable purchases. Younger consumers, particularly Gen Z (43% adoption rate), are more likely to follow AI recommendations than older cohorts, creating both opportunity and risk for retailers who fail to invest in product data standardization and accuracy.