From Failure to Satisfaction: How AI Chatbot Language Affects Service Recovery in Chinese E-Commerce

Abstract

In e-commerce service recovery, AI chatbots are increasingly replacing human agents, and their language style plays a key role in shaping consumer perceptions. This study investigates how anthropomorphic language style used by AI chatbots influences service recovery satisfaction among iGeneration (iGen) consumers in China. Drawing on service recovery theory and social cognitive theory, a moderated mediation model was tested using data from 312 participants with recent service failure experiences. The findings show that ALS has a positive effect on RS. This effect operates primarily through perceived warmth, whereas perceived competence does not play a mediating role. In addition, the mediator influence of perceived warmth weakens when the service failure is perceived as more severe, suggesting that the impact of anthropomorphic communication is contingent on failure severity. These finds highlight the significance of employing warm, human-like language when designing chatbot interactions, which helps improve service recovery outcomes and rebuild customer trust, especially among younger, digitally native consumers. However, companies should apply such strategies with caution in severe failure contexts, where their impact may diminish. This study advances theoretical understanding of emotional mechanisms involved in AI-mediated service recovery, offering practitioners seeking to make AI communication more human and optimize post-failure customer experience in e-commerce settings.