Abstract

In e-commerce platform service encounters, anthropomorphism—the attribution of humanlike qualities to AI customer-service agents (e.g., chatbots)—has become a dominant design strategy. However, its role in online service failures remains debated: attribution-centered accounts argue that anthropomorphism heightens blame and undermines trust recovery, whereas perception-centered accounts suggest that perceived warmth and competence drive forgiveness and trust. This study advances the debate within the online retail context by integrating stereotype-content theory and mental accounting into service-failure research. Using scenario-based experimental survey data (N = 400) that reflects post-purchase failures on e-commerce platforms, we demonstrate that anthropomorphism significantly enhances perceived warmth and competence, both of which directly contribute to trust recovery. In contrast, internal attribution is neither influenced by anthropomorphism nor predictive of trust recovery, clarifying the limits of attribution-based explanations in platform-mediated recovery. Moreover, consumers’ psychological investment (mental accounting) directly strengthens perceptions of warmth and competence, extending its scope from economic decision making to social cognition in online shopping. These findings reframe anthropomorphism as context-dependent rather than universally effective, position warmth and competence as primary drivers of trust beyond attribution in chatbot-mediated e-commerce recovery, and establish consumer investment as a novel boundary condition in human-AI trust dynamics.

This work is licensed under a Creative Commons Attribution 4.0 License.