Less human, more effective? Consumer responses to human-like and cartoon-like virtual influencers

This study examines how influencer type shapes brand attitude, user engagement, and purchase intention across physical and digital products. Adopting a contingency-based logic, the study integrates linear anthropomorphism assumptions with uncanny valley arguments that anticipate non-linear responses to intermediate levels of visual humanness, and theorises that the effectiveness of each influencer type depends on the type of product endorsed. Using a 2 × 3 between-subjects experimental design, the study investigates source credibility as a key explanatory mechanism. Results show that influencer type does not significantly affect brand attitude; however, cartoon-like virtual influencers outperform human-like virtual influencers and perform comparably to real human influencers in driving engagement and purchase intention, particularly for digital products. Consistent with the anticipated uncanny valley dynamics, increased visual humanness does not enhance effectiveness. Instead, human-like virtual influencers appear disadvantaged due to perceptual ambiguity. Source credibility exerts strong direct effects on all outcome variables but does not mediate the relationship between influencer type and consumer responses. These findings extend and refine linear anthropomorphism assumptions and advance a contingency-based understanding of virtual influence in AI-mediated human – computer interaction contexts.