Anastasia Braitsik is a name synonymous with the evolution of data-driven marketing. As we navigate the complex landscape where AI agents now handle the heavy lifting of audience research, her insights have become a North Star for brands trying to find their footing. With a background rooted in the granular world of SEO and deep-funnel analytics, Anastasia understands that the current obsession with “AI visibility” might be leading marketers into a sophisticated trap. In this conversation, we delve into the mechanics of agentic commerce, the persistent myths of generative engine optimization, and why the scale of modern data can often hide the very customers a brand is trying to reach. We explore the transition from human-led hypotheses to machine-speed pattern recognition and why the historical lessons of data management platforms are more relevant today than ever.
AI agents can analyze thousands of behavioral and purchase signals simultaneously, far surpassing human capabilities. How does this shift the dynamic of audience research, and what are the hidden risks of letting the machine take the lead?
It fundamentally changes the scale of our operations, but it is a double-edged sword that many are failing to handle with care. While a human researcher might look at several sources, identify a few key patterns, and form a thoughtful hypothesis, an agent works across thousands of behavioral, purchase, interest, and intent signals in real-time. I often look at the work being done at companies like Skydeo, which draws on a staggering 1.4 trillion data points across more than 320 million people, to see the sheer volume of raw material we are dealing with. The risk isn’t that the agent will fail to find an audience; it’s that it will find the wrong one with terrifying efficiency. If the underlying data is flawed, the agent doesn’t fix it—it amplifies it, making those bad assumptions louder and more expensive than they ever were when humans were at the helm.
In this landscape of Generative Engine Optimization, many marketers celebrate when their brand shows up in a ChatGPT or Gemini response. Why is being mentioned fundamentally different from being chosen by a customer, and where are brands missing the mark?
This is the most common mistake I’ve seen throughout the year, where marketers treat a mention as the finish line rather than the starting block. It’s the difference between a name-drop in a crowded room and an actual handshake that leads to a transaction. We’ve seen brands race to get cited by generative engines, focusing on clean structure and clear answers, which does improve visibility but often fails to move the needle on conversions. The “checkout underneath” is frequently neglected; getting into the results is the easy half, but ensuring the infrastructure can handle a machine-speed transaction is where the strategy often collapses. If you are showing up for the wrong reasons or for the wrong audience, you are effectively automating your own strategic blind spots and wasting resources on “AI visibility” that doesn’t result in qualified traffic.
You’ve pointed out a specific warning sign where output keeps climbing while engagement or conversion quietly flattens. What does this “volume trap” feel like for a marketing team, and how can they diagnose it before it’s too late?
It feels like running a marathon on a treadmill—there is a tremendous amount of exertion, but the scenery never actually changes. This is a trap that a huge portion of SEO teams have already tripped over, where they produce more content, more variations, and more campaigns because automation makes it cheap and easy to do so. Volume looks like progress on a spreadsheet, but it often masks a decline in the quality of the connection with the consumer. The gut-punch moment comes when no one on the staff can actually explain why a specific audience was targeted or why a particular message was sent out. Once the honest answer becomes “the AI chose it,” the feedback loop that used to catch bad assumptions is broken, and a strategy can run itself into the ground for months before the leadership realizes the numbers that actually matter are sliding.
Looking back at the history of data-driven marketing, you’ve drawn parallels to the DMP era of the early 2010s. What lessons from that period are most relevant to the current AI agent craze?
The DMP era promised that if we just stitched enough third-party data together at scale, we could out-target anyone relying on direct, first-party relationships. It largely failed because that third-party data was frequently inaccurate or outdated, and the scale simply meant that the “wrongness” of the data compounded at a faster rate. We saw the industry forced back toward first-party and declared signals after cookie deprecation in Safari and Firefox, proving that a direct relationship with the customer is the only thing that lasts. Today’s AI agents are essentially a new, more powerful engine running on that same old lesson from the past. Access to a powerful model is becoming a commodity, but what you feed into that model is where the true competitive advantage sits, and brands that forget this are doomed to repeat the same expensive mistakes.
For a brand ready to scale their AI-driven strategy, what are the three key checks they should perform immediately to ensure their targeting data is actually working?
First, you must conduct a brutal audit of your audience signals to separate what customers actually declared, what you observed them do, and what a model merely inferred about them. You have to be honest about how much of your current targeting rests on the weakest of those three categories. Second, stop measuring AI visibility as a simple citation count and start tracking whether the prompts you are appearing for actually match what your customers are asking. Finally, you need to build in a standing check where a human team member has to explain, in plain language, the logic behind a message or audience choice. If that person can’t explain the “why” without pointing to the algorithm, that is your signal to slow down, not speed up, because your strategy has lost its human anchor.
What is your forecast for the evolution of AI visibility and audience targeting?
I believe we are approaching a period of “quality consolidation” where the novelty of being mentioned by an AI agent will wear off and the focus will shift entirely to conversion integrity. Brands will realize that the agent itself isn’t the differentiator; the quality of the data they fed it before it ever started working was the real hero. We will see a massive reinvestment in “declared signals” because, in an age of machine-speed transactions, you cannot afford to guess what a customer wants. The winners will be those who treat AI agents as a way to scale human intent rather than as a replacement for it, ensuring that every automated interaction is grounded in a verifiable truth about their audience.
