Virtual influencers are more effective when paired with a companion, which makes them seem more human and trustworthy, according to research from Wharton’s Jonah Berger, USC Marshall’s Giovanni Luca Cascio Rizzo, and LUISS Guido Carli University’s Francisco Villarroel Ordenes.
Transcript
Transcript
What Is a Virtual Influencer?
Angie Basiouny: Brands have long used celebrities, athletes, and social media influencers to pitch their products. Now, they’re getting endorsements and driving sales in a different way, no humans required — virtual influencers. These are computer-created personalities, and they are gaining immense popularity around the world. I’m Angie Basiouny, and I’m here today with Wharton marketing professor Jonah Berger. He’s bringing us some of his research on virtual influencers and what he discovered companies are doing to actually boost the effectiveness of these avatars. Jonah, welcome back.
Jonah Berger: Thanks for having me.
Basiouny: Let’s just jump right into this. What is a virtual influencer? Why are they so popular? And why would a company choose to use one, as opposed to a real person? What’s the advantage?
Berger: Virtual influencers really started out as almost online avatars in some ways, right? Think about an image of someone that’s not actually a real person. It’s a fabricated image of someone who looks real. Initially, they might have taken a real person’s body but superimposed a face on top of it that would have been consistent across images. This isn’t an actual human being; it’s an online avatar that is run by a brand or media company.
Influencers are interesting. Obviously, companies want to get people to share word-of-mouth about their products, services, and ideas, and so many companies and organizations started paying traditional human influencers to do this. But there are some challenges. You may not be able to find someone who has the exact mix of characteristics that you’re looking for. Maybe you’re looking for someone who is from a particular country and has a certain set of hobbies or interests. Finding that match between geography and demographics and psychographics and interests might be difficult to do.
Second, working with virtual influencers gives companies and organizations a lot more control. When you work with a human influencer, you get some say in what they say about the brand, but you don’t always get full, complete sign-off on everything that they do. They may have some leeway in what they get to say, or if they get in trouble it can hurt the brand.
The benefit of working with these virtual influencers is that media companies or the individuals who run them have much more control over what these virtual influencers say and do. Obviously, they’re not going to get in trouble with the law because they’re not real people. They can diffuse word of mouth and influence in a different, novel way.
Basiouny: There was something I saw in your paper regarding that, and it said that consumers don’t necessarily believe that a real social media influencer has tried the product. But you’re sort of disabused of that notion when it’s a virtual influencer. You already know that’s not possible, so you get past that worry.
Berger: Yes. What’s also changed so much in the past few years, even the time we’ve been working on this paper, is the way the virtual influencer space has changed. Originally, it might have been an image superimposed on a real person’s face. But now, with everything going on with LLMs and related approaches, they can create images and create videos and do things that were not possible two, three, four, or five years ago. There is going to be a lot more content that’s created by people other than humans. Already, some of it is going on, and there’s only going to be more in the coming years.
Why Choose Virtual Influencers Over Human Influencers?
Basiouny: You had some great stats in your paper that I want to share with our audience. You had that 58% of American consumers follow at least one virtual influencer; that 40% of American consumers have purchased something that was promoted by a virtual influencer; and that the VI market is going to reach 11 billion dollars this year, which is huge. So is the message that if you’re a brand or a company, just make an avatar, stick it online, and watch the money roll in?
Berger: Well, I think there’s not just one choice; there are a couple of choices brands have. One is, should we create our own virtual character? I think the challenge there is they can’t only post about you. Most people probably wouldn’t want to follow a virtual influencer if they only talk about Coca-Cola, Delta Airlines, or Nike all the time. One of the reasons people want to follow these virtual influencers is because they seem like regular people, right? Yes, sure, they do some collaborations with brands, but they also post some unbranded content. They also might talk about someplace they’re going this weekend or show a picture of them doing something with friends that has nothing to do with brands that makes them seem more like a real person, as we’ll talk about in a couple of minutes.
Many brands are considering, “Might we want to work with one?” Lots of big brands have started collaborating with virtual influencers, just like they might with a regular human influencer. I think the question, though, really becomes: Is it going to work? Just as you work with a regular influencer, is working with this person, is paying them — this is a form of paid media, after all. Is paying them going to have that return on investment? Are we going to get more from working with this person than doing something else with our marketing dollars?
Basiouny: There is a main takeaway in your paper that I would like for us to talk about, and that is the presence of virtual influencers online and what boosts their effectiveness. Let’s talk about that.
Berger: I think this nicely ties to the question you just talked about. But the question is not just “Did I work with a virtual influencer” but “What makes virtual influencers effective?” If you look at their posts, some of them get a lot of attention, get a lot of likes, get a lot of comments, get a lot of shares. Others, not so much. Some get a lot of engagement, some don’t. Why? And you could say the same thing about regular influencers. Certain posts get more engagement, certain don’t. Why?
As we talked about already, what’s really important in this space of virtual influencers is trust. It’s already important in the space of influencers more generally, but it’s particularly true in this virtual influencer space because a virtual influencer can’t have tried the product that they’re talking about because they’re not real. They can’t have been to that restaurant, gone to that soccer game, or worn the shoes. It’s unlikely that they actually engaged with that product, and so why would you want to follow them? Why would you want to engage with them?
We started looking at this across a number of different projects. We looked at some of the language that influencers and virtual influencers use, but we also started looking at the images, and there’s a lot of interesting work going on in computer vision and other things in this space. Content analytics. Not just from language — but to think about how to analyze images or video or audio features. Other dimensions of content .
What we did here is we looked at a whole bunch of different influencers. We looked at a whole bunch of posts they made, and we analyzed aspects of the photos they posted to see what about those photos might increase engagement.
How Virtual Influencers Build Trust
Basiouny: What did you find?
Berger: We found something quite interesting. It turns out that appearing with someone else in an image — so if you’re a virtual influencer, appearing in an image not just by yourself, but with someone else, particularly someone who seems human — it is much more likely to lead to positive engagement on that post. So, posts where I’m in a photo [and] it’s not just me — I’m appearing in a photo with other people, particularly those that look human — get more engagement. Whether it’s being at a restaurant with some folks that look like your friends, being at the gym with somebody who looks like someone you know, and so on.
We dug deeper, and we found something quite interesting, which is that these posts essentially humanized the virtual influencer. As consumers, when we look at posts online, we’re trying to figure out, “Should I trust this thing?” Whether it’s a virtual influencer or a regular one, should I trust this post? And with virtual influencers, trust is low to begin with. But one thing I look for in trying to figure out if I can put my trust in this is: Do they seem human? The more they seem human, the more likely I am to say, “Okay, this is more like a regular person. I can trust what they’re saying. I’m going to be more likely to engage with this post.” So, appearing with someone else, appearing like you have a friend, a social tie, or a relationship with a real person — seeming like you do, even though you’re a virtual influencer — makes the individual viewing that post go, “Wow, you seem more like a human. You seem more anthropomorphic. You have human characteristics. Now I’m going to trust you more.” And as a result, that post is going to get more engagement.
Basiouny: It’s this idea that this avatar is not really in isolation, that it has a social network, a community, a tie. Does it matter if the avatar is appearing with other characters that are also avatars, or is it better that they appear with real people, real humans?
Berger: Great question. I think the question is we don’t know, as viewers of these posts, exactly who they’re appearing with, right? But the more human that person seems, the more human that image seems — if I’m appearing in a photo, my face is there, and there’s another face, and it looks like that other face is an actual, real human being, the bigger the impact is. Why? Because if I’m showing up with someone who looks like a cartoon, or doesn’t really look like a real person, then the consumer viewing this will go, “Ah, this isn’t really a human. And if it’s not really a human, I’m not going to trust it. And if I’m not going to trust it, I’m not going to engage with it.” Engagement is higher in posts when it looks like the person you’re appearing with is actually a real human, rather than some other avatar that you might be connected with.
Basiouny: One of the important threads I read in your paper is that you really emphasize that consumers are smart. They will not be duped into purchasing something just because there is a cute avatar or a puppy or what have you. I think that gets back to this issue of trust that you’re talking about with VI influencers. As we’re moving deeper into AI-generated content, as we’ve discussed before, how do you see brands dealing with that trust factor?
Berger: It’s been interesting. I’ve been working in the word-of-mouth field now for probably almost 20 years, thinking about issues related to word-of-mouth. Initially, people said, “Wow, we trust word-of-mouth because we know it’s diagnostic. We know that the person is objective, compared to, let’s say, a piece of paid or owned media.” A piece of earned media — some friend telling me they liked something — that’s really useful information. They’re not getting paid. They’re only going to tell me they like it if it’s actually pretty good. So I can trust them, and that information is going to be more targeted, so it’s going to have more impact.”
As we migrated online, that was still the case, at least to some degree, but today, things have gotten more complicated. Not only is there AI content out there, but there are influencers who may be getting paid to post things. Well, should I trust that influencer or not? When should I trust them, and when should I not trust them? What’s going to make me be more likely to trust them?
I think this is a really interesting space, both for consumers, as individuals, thinking about, “How should I consume content online, and what should I trust, and what should I not trust?” And also for brands thinking about, “How do we get the word out about our products and services and ideas? Should we work with these online influencers? Should we work with regular Joes and Janes and have them sample the product?” That way [seems] much more organic and probably more [trustworthy]. Maybe they don’t have as big of an audience. If I post about something, compared to an influencer, my post may not have as big a reach as their post does — but hopefully my friends and colleagues trust me because they know I didn’t get paid, and maybe they’re going to be more likely to listen to that message. It’s certainly an interesting space, but with lots of interesting changes going on.
Basiouny: Yes, it’s definitely evolving because social media platforms themselves are evolving, so who knows what they’re going to look like in 5 to 10 years.
Berger: Definitely. It’s an interesting space to watch.