A meta-analysis of 76 studies with over 64,000 participants reveals that human-likeness—not follower count—is the key driver of purchase intention in virtual influencer marketing.
They have flawless skin, perfect smiles, and millions of followers. They never age, never cause scandals, and never ask for a raise. They are virtual influencers—AI-generated digital characters like Lil Miquela, Imma, and Shudu Gram—and they are taking over social media marketing.
But do they actually work? Do consumers really buy products endorsed by computer-generated faces? And if so, what makes a virtual influencer effective?
The study, titled “How and When Do Virtual Influencers Work? A Meta-Analysis of Mechanisms and Moderators in Digital Commerce,” synthesized 186 effect sizes from 76 studies with a total sample of 64,545 participants.
The findings are clear: human-likeness is the central driver of purchase intention. Virtual influencers who look, act, and express emotions like real people generate higher source credibility, deeper customer engagement, and more positive attitudes—all of which lead to increased purchase intention.
But here is the surprise: follower size has no significant impact on purchase intention. Unlike human influencers, where mega-influencers with millions of followers drive sales, virtual influencers succeed based on their believability, not their popularity.
The Explosive Growth of Virtual Influencers
Since 2018, the number of virtual influencers has exploded. These digital characters are created using CGI (computer-generated imagery) and powered by artificial intelligence. They post photos, share stories, interact with followers, and partner with brands—just like human influencers.
The advantages are obvious. Virtual influencers are:
- Easier to control: No scandals, no bad behavior, no unpredictable human emotions.
- Less risky: They cannot say something offensive or get caught in a controversy.
- Cheaper: No travel expenses, no wardrobe costs, no management fees.
- Always available: They can post 24/7 without breaks.
But until now, researchers have struggled to understand exactly how and when virtual influencers influence consumer behavior. Some studies focused on anthropomorphism (human-likeness). Others emphasized source credibility or customer engagement. No single study had integrated all these factors into a comprehensive framework—until now.
The meta-analysis reveals that research on virtual influencers has grown dramatically. In 2022, only a handful of papers existed. By 2025, the number of publications had increased by 7.5 times. The most prolific journals included the Journal of Retailing and Consumer Services, Computers in Human Behavior, and the Journal of Theoretical and Applied Electronic Commerce Research.
Top Journals Publishing Virtual Influencer Research
| Journal | Number of Studies |
|---|---|
| Journal of Retailing and Consumer Services | 6 |
| Computers in Human Behavior | 4 |
| Journal of Consumer Behaviour | 3 |
| Journal of Theoretical and Applied Electronic Commerce Research | 3 |
The Key Finding: Human-Likeness Drives Purchase Intention
The meta-analysis used a framework called Stimulus-Organism-Response (S-O-R) theory. In simple terms:
- Stimulus: Human-likeness (the visual, behavioral, and emotional realism of the virtual influencer)
- Organism: Source credibility, customer engagement, and attitude (internal psychological states)
- Response: Purchase intention (the consumer’s willingness to buy)
The results show that human-likeness directly affects purchase intention (β = 0.25, p < 0.001). But more importantly, it also works indirectly through three key mediators:
- Source Credibility (β = 0.43 from human-likeness): When a virtual influencer looks and acts like a real person, consumers perceive them as more attractive, trustworthy, and expert. This credibility then drives purchase intention (β = 0.10, p < 0.001).
- Customer Engagement (β = 0.48 from source credibility): Credible virtual influencers generate emotional interaction—likes, comments, shares, and feelings of connection. This engagement strongly predicts purchase intention (β = 0.32, p < 0.001).
- Attitude (β = 0.58 from customer engagement): Engaged consumers develop positive attitudes toward the influencer, the brand, and the advertisement. Positive attitude is the strongest predictor of purchase intention (β = 0.43, p < 0.001).
The mediation analysis revealed that the indirect effects are actually larger than the direct effect. The total effect of human-likeness on purchase intention is β = 0.45, with 44% of that effect explained by the mediating mechanisms.
Key Path Coefficients in Virtual Influencer Effectiveness
| Path | Beta Coefficient | Significance |
|---|---|---|
| Human-likeness → Source Credibility | 0.43 | p < 0.001 |
| Source Credibility → Customer Engagement | 0.48 | p < 0.001 |
| Customer Engagement → Attitude | 0.58 | p < 0.001 |
| Attitude → Purchase Intention | 0.43 | p < 0.001 |
| Human-likeness → Purchase Intention (direct) | 0.25 | p < 0.001 |
| Total effect (Human-likeness → Purchase Intention) | 0.45 | p < 0.001 |
In plain English: A virtual influencer who looks, moves, and expresses emotions like a real person is far more likely to generate sales. The more human-like the digital character, the more consumers trust them, engage with them, and ultimately buy from them.
The Surprise: Follower Count Doesn’t Matter
One of the most striking findings challenges conventional wisdom about social media marketing. For human influencers, follower count is everything. Mega-influencers with millions of followers command higher fees and generate more sales.
But for virtual influencers, the meta-analysis found no significant moderating effect of follower size. Whether a virtual influencer has 10,000 followers or 10 million, the impact on purchase intention is the same.
Why? The researchers explain that consumers perceive virtual influencers differently than human influencers. For human endorsers, follower count is a cue of “social proof”—if millions follow, the person must be credible. But virtual influencers are understood as digital entities. Consumers do not apply the same social validation rules. Instead, they focus on the virtual influencer’s believability, psychological appeal, and human-likeness.
This finding has enormous practical implications. Brands do not need to spend millions on virtual influencers with huge followings. A smaller, more human-like virtual influencer may be just as effective—or even more so.
Product Type Matters (Exploratory Finding)
The meta-analysis also examined whether product type moderates the effectiveness of virtual influencers. The results, while not surviving statistical correction, suggest interesting patterns:
- Experiential products (food, cosmetics, travel) show stronger effects from human-likeness than search products (electronics, appliances). Unadjusted p = 0.007.
- Hedonic products (pleasure-oriented, like fashion and entertainment) show stronger effects from human-likeness than utilitarian products (practical, like insurance and tools). Unadjusted p < 0.05.
These exploratory findings make intuitive sense. Virtual influencers excel at creating emotional engagement and sensory imagination. For products that consumers buy for pleasure or experience, the emotional appeal of a human-like virtual influencer is highly effective. For purely functional products, consumers may focus more on technical specifications and rational features.
Consumer Age: Another Exploratory Finding
The meta-analysis also explored whether consumer age moderates the effectiveness of virtual influencers. The unadjusted results suggest that older consumers (over 40) may rely more heavily on source credibility than younger consumers (under 30). Unadjusted p = 0.05.
This makes sense. Older consumers tend to be more analytical, skeptical, and concerned about trust and information control. They are less swayed by emotional engagement and more persuaded by credible, trustworthy sources. For brands targeting older demographics, virtual influencers should emphasize expertise, trustworthiness, and attractiveness.
Younger consumers, in contrast, are more responsive to customer engagement—likes, comments, shares, and emotional connection. They have grown up with digital characters and are more comfortable with mediated interactions.
Cultural Factors: Uncertainty Avoidance
The meta-analysis also examined cultural dimensions from Hofstede’s framework. The only exploratory finding that approached significance was uncertainty avoidance—the degree to which people feel threatened by ambiguous situations.
In cultures with high uncertainty avoidance (e.g., Greece, Japan, Korea), the relationship between customer engagement and purchase intention may be stronger. Unadjusted p = 0.03.
This suggests that in uncertain environments, engaging with a virtual influencer may reduce perceived risk. The familiarity and predictability of the interaction provide psychological comfort, making consumers more willing to purchase.
What Doesn’t Matter: Government AI Readiness
Surprisingly, a country’s government AI readiness—its technological infrastructure, data capabilities, and AI policies—did not moderate any relationship. Consumers in countries with low AI readiness were just as likely to purchase from virtual influencers as consumers in technologically advanced nations.
This finding challenges assumptions about technology adoption. Purchase intention is a micro-level psychological process. Consumers care about whether the virtual influencer looks real and trustworthy, not about their country’s AI policies.
Practical Implications for Marketers
1. Prioritize Human-Likeness Over Follower Count: Do not waste money on virtual influencers with millions of followers. Invest in creating digital characters that look, move, and express emotions realistically. Focus on visual quality, behavioral realism, and emotional expression.
2. Build Source Credibility: Consumers need to perceive virtual influencers as attractive, trustworthy, and expert. Use cognitive cues that showcase expertise. Incorporate humor and curiosity to build trust.
3. Foster Customer Engagement: Encourage likes, comments, shares, and emotional interaction. Use storytelling, first-person pronouns, and content that aligns with audience interests. Make the experience entertaining and fun.
4. Match Product Type to Strategy: For hedonic and experiential products (cosmetics, fashion, travel), emphasize human-likeness and emotional appeal. For utilitarian and search products (electronics, insurance), focus on technical quality and informative clarity.
5. Tailor to Age Groups: For older consumers, emphasize source credibility. For younger consumers, focus on customer engagement.
6. Adapt to Culture: In high-uncertainty-avoidance cultures, use customer engagement to reduce perceived risk. Provide clear, consistent, and predictable interactions.
Limitations and Future Research
The meta-analysis acknowledges several limitations. First, the operationalization of key constructs varied across studies. Source credibility was sometimes measured as expertise, sometimes as trustworthiness, sometimes as attractiveness. Future research should develop standardized measurement frameworks.
Second, the meta-analysis could not examine other potentially important constructs like perceived warmth, perceived competence, parasocial interaction, uncanny valley effects, or authenticity due to data limitations.
Third, most studies focused on purchase intention rather than actual purchase behavior. Future research should examine whether virtual influencers drive repeat purchases and brand loyalty.
Finally, some of the moderating effects (product type, age, uncertainty avoidance) did not survive rigorous statistical correction. These should be treated as exploratory findings requiring replication in future research.
A Final Word of Encouragement
Virtual influencers are here to stay. As AI technology continues to advance, these digital characters will become even more realistic, engaging, and persuasive. The meta-analysis provides a roadmap for success: prioritize human-likeness, build credibility, foster engagement, and match strategy to product type and audience.
The future of digital commerce is not human vs. AI. It is human-like AI that works with humans to create better experiences, better engagement, and better outcomes. This study shows that when done right, virtual influencers can be powerful partners in the marketing mix.
Reference: here
Other Articles: