30% data gathering, 70% thinking: What AI changed at BeInfluence

Midway through a client call, a BeInfluence campaign manager needs a number. Two years ago, they would have opened Excel, run exports and cobbled together a formula. Today, they ask Claude or ChatGPT, connected to Kolsquare through MCP, and get answers on the spot. At BeInfluence, an agency based in Brussels, Paris, Geneva and London, data collection used to eat up 90% of the team's time. Now it takes 30%. Still, Bruno Durbecq, Data & Insights at BeInfluence, does not act with blind faith in the machine. When the agency pitted AI against its own team to select influencers, the humans won. And for agencies in a hurry to connect everything, he has contrarian advice: wait six months and clean up your databases first.

Posted on
18/11/2026

TABLE OF CONTENTS

What has AI changed most in influencer marketing?

Access to data. Two years ago, we were still doing Excel exports with our own formulas. Today, it's like talking to a colleague who has access to everything. I ask, "How's my campaign doing?" and in under a minute I get far broader results than I could have hoped for. With the MCP and API, our campaign managers cross-reference almost unlimited data across campaigns to benchmark themselves.

Before, 90% of our time went into pulling the data and, if there was any time left, we asked the right questions. Now it's 30% extraction and 70% thinking: how to interpret it, how to present it, how to do better next time. You have to know how to filter, so we keep answers simple by default.

Do you also use AI to find creators?

It's not off-limits, but our internal policy is clear: people have a much sharper eye for discovering profiles, and creativity has to stay at the heart of the job. AI should make us more efficient, not do our thinking for us. Our best campaigns came from four people standing around a whiteboard. That human nuance is what spots the profile that seems to have nothing to do with the campaign but carries a real creative idea.

We do use it, for sure, for much more niche profiles. Some of our clients work in Romania, for example, a market we don't know very well, and they're looking for rural creators. Those are harder to find.

We also ran a test, an AI selection against a human selection. The humans win. AI is obsessed with efficiency. Out of 10 influencers, it will tell you that two big accounts with 1 million followers are enough to hit your reach target. Our 10 profiles hit the same reach, but with completely different audiences and much higher engagement. What matters is who you reach and how people interact.

Virtual creators and AI-generated content: How much space should they take?

Not much. Virtual is an add-on, not a replacement. Animation never took over from live-action film, after all.

AI content is like an influencer who only does ads: It works for a while. You pay an influencer to watch them be human and make mistakes, so people can relate to them. Using AI now and then for editing, with a real purpose, why not? You need a happy medium.

With almost unlimited data cross-referenced from one campaign to the next, where will influencer marketing be next year?

Forecasts will be much more accurate. Today, they're still purely mathematical: An influencer hit 1 million reach three times, so they'll hit 1 million again. We built our own prediction model and, in 2027, I think we'll factor in external elements. Is the video ahead of a trend or behind it? Will the music and the hooks carry it?

That will change how we pay and negotiate with influencers, maybe with incentives rather than cash alone. Everyone is going to be challenged, influencers first, and us too.

Will influencer marketing go self-serve, like other channels?

That will put pressure on agencies. Ten years ago, there were maybe 10 of us in Belgium doing social media advertising. Today, clients hire in-house and run their campaigns themselves, the way they do on Google or Amazon. For influencer marketing, I don't have the answer. YouTube launched its program in the U.S., but will Big Tech open up self-serve access or keep control?

What I see coming is real attribution. More and more creators will have to connect their accounts to the platforms, so we'll have actual figures instead of estimates based on scraping. We still don't know whether a poster at a bus shelter makes someone buy a Coke. Online, the data exists, tied to each user, but it sits untouched in the black box of Meta, TikTok or Snapchat. Eventually, we'll get anonymized access to it and be able to model the real path to purchase.

Is it time to say goodbye to earned media value?

No, but it will be challenged. Advertisers will demand more complete reporting, and earned media value (EMV) relies on formulas that vary, with ranges anywhere from €50,000 to €120,000 (about $56,000 to $134,000). We're working on new metrics around visibility and hooks to put a more realistic value on what advertisers are buying.

We'll have new KPIs and a new way of reading the old ones. Look at Instagram. You see impressions less and less, and its head even announced he would remove some metrics because, in his view, they serve no purpose. What matters is interaction. I see content all day long, but when I comment or share something with a friend, there's real intent behind it. Within six months, Instagram added watch time and the drop-off curve: Do people leave after three seconds, after seven? That's what will measure a piece of content's long-term impact.

Will AI cut jobs in influencer marketing?

I don't think so. Twelve years ago, I was working on Google Ads and people were already telling me, "In five years, you won't have a job." It's still there, and the people who were around before automation keep an edge: They know how the machine optimizes and how to influence it. A job like ours will always require relationships with clients and creators, the ability to interpret a brief and a creative idea. We'll just be more efficient. Influencer marketing is still a young medium, and so many advertisers have yet to invest in it. A manager might handle 10 clients instead of seven. That's a long way from getting laid off.

In practical terms, what do Kolsquare's API and MCP do for you?

I use the API to build dashboards and benchmark our campaigns by influencer size, niche and market. We cross-reference that data with our own history: which influencers, for which clients, what worked and why. That lets us challenge casting choices and give a client a range of expected results based on their budget. People talk about predictive models, but it's often a simple ratio calculation.

Because Kolsquare's modules don't all talk to each other, I got them talking in our own database, alongside our PR, paid and events data. That's where I see a campaign's real impact. And I keep the data in-house.

MCP is about responsiveness. On a call, a campaign manager asks Claude or ChatGPT for the past seven days' performance and answers the client on the spot. Clients don't expect it yet, but some are starting to ask for it. It's a bet on the future.

Any advice for those who haven't taken the plunge yet?

Do it thoughtfully. Everywhere I've seen an MCP integration work, the foundations were solid: good naming conventions, data that connects, a clean structure. And you have to ask yourself: If I lose Claude or ChatGPT tomorrow, what do I do?

A philosopher worried that today's juniors will stay juniors their whole lives. At Deloitte or KPMG, the junior asks Claude for their slides and everyone's happy. In 20 years, they won't know what a P&L is or how to question what AI tells them. Seniors can see what's off. So wait six months, clean up your databases and understand what you're doing before you connect anything. Training tomorrow's campaign managers won't be enough. We'll also face the challenge of teaching tomorrow's advertisers to approach these new tools with caution.

About Kolsquare

Kolsquare is Europe’s leading Influencer Marketing platform, offering a data-driven solution that empowers brands to scale their KOL (Key Opinion Leader) marketing strategies through authentic partnerships with top creators.

Kolsquare’s advanced technology helps marketing professionals seamlessly identify the best content creators by filtering their content and audience, while also enabling them to build, manage, and optimize campaigns from start to finish. This includes measuring results and benchmarking performance against competitors.

With a thriving global community of influencer marketing experts, Kolsquare serves hundreds of customers—including Coca-Cola, Netflix, Sony Music, Publicis, Sézane, Sephora, Lush, and Hermès—by leveraging the latest Big Data, AI, and Machine Learning technologies. Our platform taps into an extensive network of KOLs with more than 5,000 followers across 180 countries on Instagram, TikTok, Twitch, Facebook, YouTube, and Snapchat.

As a Certified B Corporation, Kolsquare leads the way in promoting Responsible Influence, championing transparency, ethical practices, and meaningful collaborations that inspire positive change.

Since October 2024, Kolsquare has become part of the Team.Blue group, one of the largest private tech companies in Europe, and a leading digital enabler for businesses and entrepreneurs across Europe. Team.Blue brings together over 60 successful brands in web hosting, domains, e-commerce, online compliance, lead generation, application solutions, and social media.

FAQ

What happens if I disagree with the AI's creator ranking?

You override it. Rankings are a starting point, not a verdict: refine the prompt, adjust the criteria, or ask for more options. AI accelerates the shortlist; your team owns it.

How is Kolsquare's AI different from a basic keyword search tool?

Keyword tools match tags and captions. Kolsquare's AI understands content: what creators talk about, what their content looks like, and what they say out loud in videos.

Who in my company can see our data through the MCP?

Access works through a company-level key today: anyone in your company who holds it can query your company's data. Per-user access is on the roadmap. Share the key with care.

Can I see the AI matching logic, or is it a black box?

Search results come back ranked so your team can see why a creator surfaced and defend the shortlist. Your team always makes the final call.

Is Kolsquare's AI trained on my data?

No. Your data is never used to train models.

Is Kolsquare an AI company, or a platform that uses AI?

A platform that uses AI where it makes a real difference, and says clearly where it doesn't. Search, sentiment, and the MCP are AI. Scoring, compliance, and EMV are precision algorithms, and we name them as such.

What does responsible AI mean in influencer marketing?

At Kolsquare: AI runs on verified, GDPR-compliant data, client data is never used to train models, every model has human oversight, and AI-assisted access to your data is read-only. We're a B Corp, and we apply that standard to AI too.

Can AI detect fake followers or bot-driven engagement?

Kolsquare catches inflated audiences with credibility scoring: a precision algorithm over verified audience data. We deliberately don't badge it as AI, because it doesn't need to be. It flags risk before the budget is committed.

How does AI help with influencer discovery and creator matching?

AI-based search understands meaning rather than keywords. You describe the creator you need in plain language, search by a reference image, or find spoken brand mentions in video, and get a ranked shortlist back.

What AI features does Kolsquare offer for influencer marketing?

AI prompt search and image search (both in beta), speech-to-text search across YouTube and TikTok, sentiment analysis on campaigns and brand Listening, and the Kolsquare MCP, which connects your campaign data to AI assistants like Claude, ChatGPT, and Gemini.