Who Gets to Name the Assistant? Consumer‐Generated Naming As a Buffer Against AI Recommendation Failure

ABSTRACT Consumers do not always respond to service failures in the same way. While some quickly disengage after a mistake, others are willing to forgive. This study examines whether an upstream design choice—AI‐agent naming source—shapes consumers' affective reactions to recommendation failures through psychological ownership. Across five experiments, we show that consumer‐generated (vs. firm‐assigned) naming increases forgiveness after recommendation failures (Study 1) and that psychological ownership explains this effect (Study 2a). Study 2b further suggests that this ownership‐based pathway explains the effect beyond anthropomorphism. However, these effects depend on individual differences and shopping orientation. They are stronger among consumers with higher mine‐me sensitivity (Study 3) and differ by shopping orientation. Psychological ownership promotes forgiveness under hedonic orientation but has the opposite effect under utilitarian orientation (Study 4). Overall, this study highlights how pre‐failure design choices shape consumers' post‐failure responses to AI assistants.