Antoine Khaitrine

Let’s start with you. Who are you, and what lens do you bring to understanding people online?

I run Licter, a Social Intelligence agency. We help brands leverage web and social media data. My lens is simple : most insight failures don’t come from lack of data, but from a bad interpretation. I don’t believe the Internet is a mysterious and dark place. It’s a massive, noisy, incentive-drivent system. People online behave less like individuals in the real world and more like participants in platforms designed to reward visibility, conformity, and emotional shortcuts.

At Licter, our work starts with one rule : before asking why people say something, we make sure we understand what structural forces made that speech likely. At Licter, we prioritize small signals over clever narratives. This lens helps us avoid the classic trap of mistaking loud minorities, algorithmis artifacts, or ironic speech for genuine shifts in opinion.

Social data doesn’t reveal truth ; it reveals constraints. My job is to read those constraints clearly.

‍

What’s a working theory you have right now about how people behave online?

People online optimize for belonging under uncertainty, not for truth, identity, or even persuasion. Most digital behavior is reactive. Users don’t wake up wanting to express opinions ; they respond to signals already circulating : trending formats, dominant framings, emotional rewards. The result is what looks like polarization but is often just coordination under algorithmic pressure.

This is why brands and institutions keep misreading “sentiment shifts”. What they’re tracking isn’t conviction, it’s temporary alignment.

The theory sounds cynical, but it’s actually useful. If you accept that behavior is shaped by incentives and visibility, you stop over-moralizing audiences and start analyzing systems.

‍

What’s an insight you surfaced that you still think about? What one stuck with you?

We once analyzed a topic with millions of mentions and discovered that fewer than 5% of active accounts were responsible for the dominant narrative, yet that narrative was being cited as “public opinion” by decision-makers. What stuck with me wasn’t the number. It was how easily everyone (analysts included) accepted the illusion of consensus because the data looked big.

Since then, I’ve been obsessed with separating volume from distribution, and noise from structure. Big data doesn’t protect you from bias. It just hides it better, in a certain way. And our job is to see clearly through this bias.

‍

What’s the weirdest rabbit hole your work has ever sent you down? And what did it teach you?

I’d say it’s irony : how much online speech is strategically unserious. People post things they don’t believe, don’t fully endorse, or wouldn’t defend online : simply because irony travels well.

This breaks a lot of traditional analysis models. Sentiment, stance, even topic attribution become unstable when speech is performative.

I learnt that not all data can be taken literally. Sometimes, the insight isn’t what is said, but how safe it is to say it.

‍

What skills or mindsets do you think the next generation of analysts will need?

The next generation of analysts will need to unlearn a hierarchy that has dominated corporate insight work for decades.

Historically, the mot “legitimate” analyses were those that aggregated upward : surveys, panels, studies designed to reflect executive-level decision-making. Social listening sat lower in the hierarchy, often treated as qualitative texture rather than core evidence.

That hierarchy no longer holds.

Today, analysts must be able to blend synthetic, AI-generated data with live, organic social data - not as opposites, but as complementary systems. Synthetic data helps define scope, test hypotheses, and simulate scenarios. Social data shows where those models break when exposed to real human behavior, cultural noise, and platform dynamics.

The skill is not choosing one over the other, but knowing when each becomes misleading. This requires a mindset shift : less obsession with producing “clean” narratives, more comfort with probabilistic thinking and uncertainty. Analysts will need stronger statistical intuition, but also a deep understanding of how platforms shape expression - incentives, formats, visibility mechanics.

Finally, the best analysts won’t just interpret data. They’ll build systems : AI agents trained on social signals, capable of continuously adjusting the perimeter of what counts as a “real” insight versus synthetic projection.

The future analyst isn’t a storyteller. They’re an architect of insight constraints.

‍

What’s a niche community, account, or corner of the internet you’re watching right now? And why?

In a counterintuitive way, I’d say mainstream silence : the most interesting signal today isn’t where people talk loudly, it’s where expected conversations fail to appear. Absence is becoming more informative than presence.

When topics don’t break through mainstream feeds despite high relevance, it usually signals fatigue, risk, or misaligned incentives. That’s where the real strategic insight lives.

‍

Last non-work thing you read that shaped your thinking?

The last non-work readings that truly shaped my thinking came from very different places, yet all influenced how I see entrepreneurship, technology, and long-term journeys.

“A Strong Urge to Shake Things Up” by Xavier Niel left a strong impression on me. He is one of the entrepreneurs I find most inspiring, both for his personal story and the scale of what he has built. Beyond success, what resonated most was his freedom of tone and his refusal to accept established rules as immutable. It reinforced my belief that entrepreneurship is not just about execution, but also about questioning norms and having the courage to disrupt systems that no longer make sense.

“Toxic Data” by David Chavalarias offered a very different, yet equally important perspective. The book does an excellent job of reframing the role of algorithms and social intelligence in modern democracies. It deepened my understanding of how technological systems shape collective behavior and public debate, and reminded me that innovation comes with a real responsibility toward society.

Finally, the most unexpected and personal “reading” was “My Telegram Conversation.” It’s a book my co-founder gifted me, compiling seven years of our Telegram exchanges — from our early days as colleagues at the Élysée to today as leaders of Licter. Reading it was a powerful reminder of where we started, the doubts, the energy, and the small decisions that led us here. It grounded me, emotionally and intellectually, in the value of long-term commitment and shared vision.

‍