Virtual influencer marketing: A meta-analytic review

Abstract Virtual influencers have become an established part of influencer marketing, yet their effectiveness, the mechanisms underlying their persuasive impact, and the boundary conditions shaping these effects remain unclear. We address these questions through a meta-analysis of 210 independent experimental studies and 643 effect sizes. Virtual influencers generate greater novelty but lower credibility than human influencers, revealing a systematic trade-off between novelty- and credibility-based persuasion, with the relative importance of these mechanisms shifting over time from novelty toward credibility. Virtual influencer effectiveness further depends on anthropomorphic design and contextual congruence. Highly anthropomorphic virtual influencers benefit from artificial intelligence integration, futuristic positioning, and rational, proximal messaging, whereas low-anthropomorphism virtual influencers perform better with emotional and distal framing. Meta-analytic structural equation modeling supports a dual-pathway process in which credibility-based indirect effects are stronger for highly anthropomorphic virtual influencers, whereas novelty-based indirect effects are stronger for low-anthropomorphism virtual influencers. These results establish anthropomorphism as a structural boundary condition in influencer marketing.