Latest / Influencer Marketing with Fexingo: Creators, Sponsorships, and Modern Endorsement Deals / How Influencer Agencies Are Using Data to Predict Viral Potential
Transcript
- Lucas: There's a quiet war happening inside influencer agencies right now, and it's not about who has the biggest follower count. It's about who can predict the next breakout creator before the algorithm makes it obvious. Luna: So we're talking about data-driven scouting, like sports analytics but for Instagram and TikTok. Lucas: Exactly. A few years ago, agencies would discover talent almost by gut feel or by monitoring who was getting traction on hashtags. Now they're building predictive models that score creators on their likelihood to go viral within six months. Luna: What kind of data goes into that? Is it just engagement rate and follower growth? Lucas: Those are part of it, but the interesting stuff is deeper. Some platforms like CreatorIQ and HypeAuditor are now incorporating factors like comment sentiment polarity, the velocity of brand mentions before a paid partnership, even the network structure of a creator's audience—how many of their followers are themselves creators with high engagement. Luna: That's almost like mapping a social graph to find nodes that are about to pop. Lucas: Right. And the results are starting to speak for themselves. I looked at a case from early this year where a mid-size skincare brand used a predictive scoring tool to identify a creator who had only 12,000 followers but scored in the 95th percentile on their 'breakout potential' metric. They signed her to a six-month retainer. She hit 400,000 followers within four months and their return on ad spend from that one partnership was over 300 percent. Luna: So the brand basically got in on the ground floor before her rates skyrocketed. That's the dream. Lucas: It is. But it also raises a tricky question: if every agency is using similar models, are we going to see a homogenisation of what 'viral potential' looks like? Luna: Meaning the algorithms might all identify the same type of creator—maybe someone with a certain aesthetic, posting frequency, niche—and overlook the genuinely weird or original voices that don't fit the pattern. Lucas: That's exactly the concern. There's already evidence from a study out of USC that predictive models trained on historical breakout creators tend to favour creators with lighter skin tones and Western-centric content styles. So the tool can encode bias if you're not careful. Luna: And yet, if you're an agency and you have a tool that's proven to pick winners, it's hard not to use it. The pressure to deliver ROI for brands is intense. Lucas: It's a classic tension between efficiency and serendipity. The agencies that are doing this well are using the predictive scores as a filter, not a final decision. They'll have a human scout review the top candidates and make the final call. Luna: So the machine narrows the pool, but the human still picks the talent. That feels like the right balance. Lucas: It is, for now. But I think the bigger shift is that this kind of data is making influencer marketing more accountable. Brands can now say, 'We're not just throwing money at followers; we're investing in predicted outcomes.' That changes the conversation with CFOs. Luna: I can see that. And it might push more brands to sign long-term deals with creators, because if you have a model that says this creator is likely to grow, you want to lock in the rate before they blow up. Lucas: Exactly. We're already seeing more multi-year contracts in the influencer space, especially for micro-creators. It's almost like equity investing in human talent. Luna: There's a parallel to how record labels used to scout bands. They'd have A&R reps go to clubs, but now labels use streaming data to find artists with high 'skip rate' or 'playlist addition velocity.' Lucas: That's a great comparison. And the music industry learned the hard way that leaning too heavily on data can kill artistry. So there's a cautionary tale here for influencer marketing. Luna: What about the creators themselves? Are they aware they're being scored? Lucas: Most aren't, and that's another ethical layer. Some agencies argue it's just market research, but others say creators deserve transparency if their data is being used to predict their future earning potential. Luna: I could see a future where creators get access to their own scores, like a credit report for influence. That might even help them negotiate better deals. Lucas: It could. And I think we'll see startups emerge that offer exactly that—a dashboard where creators can see how agencies and brands are likely to perceive them, based on the same predictive signals. Luna: That would level the playing field a bit. Right now, the information asymmetry is heavily in favour of the agencies. Lucas: Yeah, and that's the next frontier. But for brands listening, the takeaway is that data-driven scouting is real, it's becoming standard, and if you're not at least testing it, you might be paying a premium for talent that a predictive model could have surfaced months earlier. Luna: If that kind of insight is useful to you, if it helps you think about your own marketing or investment decisions, that's exactly why we do this show ad-free. And the way it stays that way is listener support—buy me a coffee dot com slash fexingo. Lucas: Yeah, a small gesture goes a long way. No pressure, just if you get value out of these conversations, that's the link. Luna: So back to the data question—do you think the predictive models will ever replace human scouts entirely? Lucas: I don't think so, at least not in the next five years. Human intuition is still critical for assessing things like on-camera charisma, storytelling ability, and the kind of authenticity that resonates with an audience. Data can tell you who's growing, but not who's genuinely compelling. Luna: And that human element is what makes influencer marketing different from programmatic ad buying. It's still a relationship business. Lucas: Exactly. The best agencies will be the ones that combine the predictive power of data with the judgment of experienced talent scouts. That hybrid model is where the real competitive advantage lies. Luna: I want to watch this space closely. If creators start getting access to their own predictive scores, it could change the whole negotiation dynamic. Lucas: That would be a fascinating development. And it's probably coming sooner than we think. For now, the takeaway is that the tools exist, they're getting smarter, and they're reshaping how talent gets discovered and valued. Luna: Thanks for digging into this with me, Lucas. I feel like I'll never look at a trending creator the same way again. Lucas: That's the idea. See you next time.