The beauty premium in online services: Driving trial adoption but not repeat purchases? Evidence from AI-based big data analysis

This study investigates how tutors' facial attractiveness influences sequential consumption decisions in online education and examines the moderating effects of different attribute cues. Based on a panel data analysis of 11,450 online language tutors over 10 consecutive months, the results show that a one-standard-deviation increase in tutors' facial attractiveness is associated with a 3.26% rise in first-time purchases (number of students), but has no significant impact on repeated purchases (average lessons per student). This indicates that the beauty premium effect of physical appearance only exists in the initial transaction stage, verifying its boundary as a heuristic cue in the context of information asymmetry. Further heterogeneity analysis reveals that gender and age, which are also visual cues, significantly moderate the beauty premium in first-time purchases, whereas professional qualification labels as text-based certification cues exert no significant moderating effect. This finding suggests that in online initial decision-making, visual cues are processed with higher priority and integration than text label cues, highlighting the non-equivalence of multimodal information in the digital consumption environment. As a pioneering study exploring the impact of the beauty premium on two key performance indicators of service providers—initial customer acquisition and customer retention—using AI facial analysis technology, this study systematically analyzes the dynamic role and boundary conditions of the beauty premium in the “customer acquisition-retention” process based on halo effect theory, providing theoretical basis and managerial implications for platforms to optimize the collaborative presentation of visual and textual information in online profiles.