Avi Chai Outmezguine is CEO of Becausal, powering next-generation audience intelligence through causal AI-driven data innovation.
The data exposures I worry about most don’t begin with an attacker. They begin with someone doing their job well.
It’s no secret that over the past 20 years, adtech has become incredibly sophisticated. We’ve built an extensive data ecosystem, AI tools and audience models that help us predict consumer behavior at an unprecedented scale. Marketers can make billions of decisions in seconds—even in real time.
But the same question still plagues marketers—and if it doesn’t, it should: Can I trust this data? Is this audience accurate? Can I trust the measurement?
I’d argue the next competitive advantage in our industry won’t belong to whoever has the most data or the smartest AI. It’ll go to whoever can earn marketers’ trust.
The Wrong Question
For years, our industry has focused on tackling one question: How do I build a bigger audience? That era is slipping away. Tomorrow’s question is: Can I trust the audience I’m buying?
If nothing else, it’s definitely a more interesting discussion, and I’m betting that it’s the one that will define the next decade of marketing.
Why? Today, everybody says they have AI. Everybody has machine learning, predictive models and better lookalikes. AI is no longer a differentiator. It’s table stakes. What’s important is whether the AI can explain and prove itself.
Marketing’s Black Box Problem
Too much of marketing today requires marketers to accept conclusions without seeing the evidence behind them. The platform tells you this is the right audience. The model assigns a probability. The measurement provider reports an outcome.
But nobody tells you why.
Why is this person in the audience? Why do you think they’re going to buy? Why should I spend media dollars on them? Marketers shouldn’t have to accept “because the algorithm said so” as an answer, but for years that’s exactly what we’ve been asked to do.
That question of why is the entire reason my company is named Becausal. We built it around one obsession: can AI explain itself, not after the fact, but while it’s making decisions? In practice, that means every audience segment carries its own evidence trail back to the underlying purchase signal—not just a score—so a marketer can see why before they spend against it.
More Data Isn’t The Answer. Evidence Is
We don’t lack data. We have an enormous amount of it.
The problem is that it’s fragmented, inconsistent and difficult to connect.
Purchase data is messy. Retailers describe products differently. Categories and brands don’t always line up. Identity is imperfect. Everyone talks about AI. Almost nobody talks about the quality of the information that the AI is working off of. And that’s where the real value gets created (or lost).
Real explainability means that if you ask why somebody belongs in an audience, I should be able to answer, not with a probability score, but with data you can actually understand. And since every audience member is different, you shouldn’t treat every audience member the same. There’s a scale. Some we know with certainty. Some we’re highly confident about. Some are just educated guesses.
Marketers need to know the difference, and they shouldn’t use the word “deterministic” loosely to blur it. If someone actually bought the product, that’s deterministic. If we think they look like a buyer, that’s inference. Both are valuable. But they aren’t the same thing.
And when deterministic evidence is available, it should win. Modeling should extend what we know, not blur the distinction between what was observed and what was inferred.
Fragmentation Is The Enemy Of Trust
Retail media has a similar issue. Every retailer can tell you what happened inside their own ecosystem. Is this useful? Yes. But brands don’t compete inside one retailer. They compete across an entire market, so having a broader view is essential.
Measurement has the same fragmentation problem. Today, so many operations take place in silos: one company builds the audience, another activates it, another measures it and another analyzes it. Each can introduce its own data, methodology and assumptions.
Eventually, marketers are left reconciling multiple versions of the truth.
Privacy has to be embedded in the underlying architecture from day one. If you bolt privacy on afterward, you’re already too late.
Where This Is Headed
Five years from now, brands won’t be asking how many consumers are in an audience. They’ll want to confidently understand how much evidence supports that audience. That’s a fundamentally different buying decision, and it’s the one our industry needs to prepare for
Imagine evaluating an audience not simply by its size, but by understanding how much of it is grounded in observed behavior, how much is inferred and how confident we should be in each.
We’re entering what I call the trust economy. The scarce resource in marketing isn’t data or AI anymore—it’s trust. The last decade of marketing was about finding consumers. The next decade will be about trusting the AI that found them. Data informs. AI predicts. Trust enables the decision. That’s the company we’re building, and it should be the standard our entire industry is held to.
Forbes Technology Council is an invitation-only community for world-class CIOs, CTOs and technology executives. Do I qualify?