In recent years, virtual influencers (VIs) have been increasingly used in digital commerce. Despite the rise in VI research, past studies have yet to comprehensively examine the effectiveness of VIs, often focusing only on isolated partial models rather than an integrated framework and boundary conditions that drive consumer responses. This meta-analysis fills this gap by synthesizing 186 effect sizes from 76 studies (N = 64,545) to examine the mechanisms and moderators of purchase intention in VI marketing. The results indicate that human-likeness is a central antecedent that directly and indirectly affects purchase intention through source credibility, customer engagement, and attitude. More importantly, this study challenges prior social proof assumptions by showing that follower size has no significant impact on purchase intention in VI marketing. In addition, purchase intention is independent of a nation’s AI readiness, suggesting a borderless potential for commerce regardless of a country’s technological maturity. This study also examined the moderating effects of product type, consumer age, and uncertainty avoidance culture. Although these moderators showed initial significance, none remained significant after the Benjamini–Hochberg false discovery rate (FDR) correction. Therefore, these effects were viewed as exploratory rather than confirmatory, providing directions for future research. These findings offer new insights for e-commerce managers: success in the metaverse era depends on anthropomorphism and targeted alignment rather than metrics such as follower counts or a nation’s AI readiness.
Nguyen et al. (Sat,) studied this question.