Is an apology enough? How AI chatbots’ composite service recovery influences customers’ continuous usage intention

Service failures of artificial intelligence (AI) chatbots are unavoidable; thus, understanding how AI chatbots’ service recovery effectively retains customers is a matter of considerable significance. Drawing upon the stereotype content model, this study investigates how AI chatbots’ composite service recovery (apology and promise vs. apology-only) affects customers’ continuous usage intention in e-commerce customer service. The examination was conducted using data collected from a between-subjects online scenario experiment. The results show that, compared to an apology-only strategy, AI chatbots using a composite apology and promise strategy significantly enhance customers’ continuous usage intention, with this effect mediated by customer trust. Furthermore, service provider reputation plays a moderating role: when reputation is high, the composite strategy (vs. apology-only) more effectively boosts continuous usage intention by building trust; when reputation is low, this effect weakens. These findings enrich research on AI service recovery in online shopping contexts and offer new insights for e-commerce platforms addressing AI service failures.