Anastasia Braitsik stands as a towering figure in the digital marketing landscape, renowned for her razor-sharp expertise in SEO, content strategy, and the increasingly complex world of data analytics. As a global leader who has navigated the industry’s most significant shifts, she brings a wealth of practical knowledge to the table, helping brands turn abstract data into actionable growth. In this conversation, we explore the seismic shifts occurring within Google Search Console following the June 3, 2026, announcement of dedicated generative AI performance reports. Anastasia breaks down the nuances of these new reports, moving beyond the surface-level excitement of impression counts to uncover how businesses can use this diagnostic lens to understand their visibility in AI Overviews and AI Mode. We discuss the critical differences between traditional and generative search metrics, the psychological traps of executive dashboards, and why the most important tool in an SEO’s arsenal is still the ability to actually look at a website’s HTML and content structure.
Generative search reports currently provide visibility into impressions but exclude clicks and specific queries. How should analysts navigate this lack of granular data to determine if their AI visibility is actually driving business value?
It is a bit of a bittersweet milestone because while we finally have a dedicated report for AI visibility, it feels like receiving a gift box that is beautifully wrapped but currently empty of the most important things: clicks and queries. On June 3, 2026, Google gave us a diagnostic lens rather than a final scoreboard, which means we have to be much more clever about how we interpret these numbers. Since we cannot see exactly what people are typing to find us in AI Mode or AI Overviews yet, we have to look for the signal in the noise by comparing these AI impressions against our traditional organic data for the same URLs. You shouldn’t just celebrate a rising graph; instead, you need to use this data to identify which pages Google is “retrieving” for its synthesized answers and then check your separate analytics tools for referral traffic spikes that correlate with those periods. It is about using the impression report to build a list of “AI-active” pages and then performing a deep dive into their traditional performance to guess at the intent behind the visibility.
The counting rules for generative impressions are significantly different from what we are used to in standard organic search. Could you explain how the aggregation levels at the property and page level might confuse someone who is just glancing at their reports?
This is where many analysts are going to run into trouble when they try to make their Excel spreadsheets balance, because the math of generative search is not a simple 1:1 addition. At the property level, Google aggregates data so that if two different URLs from your site—say, a product page and a research study—both appear in a single AI Overview, they only count as one single impression for your entire website. However, when you switch your view to the page table, both of those URLs might be credited with an individual impression, which leads to a total that looks much larger than the property-level summary. It is not that the data is broken; it is simply that the dimensions of the report are measuring different levels of exposure. You have to be incredibly careful when reporting these numbers to stakeholders so you don’t accidentally double-count your visibility or present a “total search visibility” metric that mixes these two very different accounting methods into one confusing bucket.
Google has introduced a specific control that allows site owners to opt out of being included in generative features without disappearing from traditional search. Under what circumstances would you actually recommend a brand use that “off switch”?
Right now, the default setting is inclusion, and for the vast majority of sites, that is exactly where they should stay because opting out means you are walking away from potential traffic and any baseline visibility in the future of search. However, I can see a scenario where a site owner might use that control if they find that Google’s AI is consistently misrepresenting their content or using it in a way that cannibalizes high-value conversions without providing any citation value. Before anyone reaches for that dramatic off-switch, they really need to use the new reports to establish a solid baseline of what they would be losing. You are essentially forfeiting your “grounding” in Google’s generative experiences, so you need to be 100% sure that the “impressions-only” relationship you currently have is actively hurting your brand more than the potential visibility is helping it. It is a powerful lever, but it’s one that should only be pulled after a rigorous analysis of the current generative footprint.
We often see cases where a page has massive organic visibility but almost zero presence in AI Overviews. What are the common technical or content-related reasons for this “visibility gap”?
It is a fascinating phenomenon to see a page that ranks perfectly in traditional results but is essentially invisible to the AI, and usually, it comes down to how “digestible” the content is for a machine looking for a quick answer. Even if a page is a ranking powerhouse, it might be a “terrible source for synthesis” if the most important information is buried behind JavaScript, decorative elements, or complex images that the AI systems don’t prioritize during the retrieval phase. I often find that pages with “high organic, low AI” visibility are the ones that require a reader to excavate the main point from six paragraphs of introductory fluff or positioning language. To fix this, you have to look at the HTML structure to ensure the answers are clear, the headings are descriptive, and individual passages can actually stand on their own without the surrounding context. Technical SEO still matters immensely here; if your content is hidden in tabs or requires specific user interactions to load, Google might rank the URL based on its overall authority but skip it when it needs to lift a clean, useful answer for a generative response.
On the flip side, some “underdog” pages with modest organic rankings seem to punch way above their weight in generative search. What can we learn from these outliers?
These are my favorite pages to study because they represent the “clues” to what Google’s generative systems actually find useful, even if they aren’t the top-ranked pages in a traditional sense. Often, a page with modest visibility becomes an AI star because it contains a very specific, well-structured definition, a unique statistic, or a clear comparison table that provides a direct solution to a user’s query. I’ve seen cases where a site has published 300 articles on a topic, but only three of them ever show up in AI Mode because those three have a focused topical scope and language that matches exactly how users ask questions. You should group these overperforming pages by author, template, or intent to see if there is a recurring pattern, such as having the direct answer right at the beginning of the section. It’s not about a secret AI ranking factor; it’s about having a “clean” page that makes it incredibly easy for Google to cite you as a primary source of truth.
When it comes to executive reporting, there is always a temptation to put the biggest “AI” number front and center. How do you advise marketers to build a dashboard that is actually responsible and not just “celebrating numbers that went up”?
The moment an executive sees a large new number labeled “AI,” they are going to want it highlighted in green at the top of the deck, but we have to resist that urge because a generative impression does not have the same business weight as a traditional click. You should never, under any circumstances, report a “blended CTR” or a “total search visibility” metric that combines these two worlds, because they are not interchangeable units of value. A responsible dashboard should instead focus on the number of pages receiving these impressions and the relationship between those AI-visible pages and your actual business outcomes. I suggest including things like the topics and page types that are winning in AI search, alongside any identifiable referral traffic from these features, reported entirely separately from your organic search totals. We already have enough dashboards that celebrate meaningless growth; the goal here is to provide a diagnostic view that shows where Google is willing to surface your links and what that teaches us about our content’s utility.
You mentioned that “someone actually has to look at the website” as a part of the analysis process. In an era of automated tools and 12-tool AI visibility stacks, why is this manual investigation still so critical?
It sounds shocking to say in 2026, but the most sophisticated data export in the world is no substitute for actually visiting a URL and seeing what a user sees. The report can tell you that a page is an outlier, but it won’t tell you that the reason it’s failing is because your “topical expert” author is actually writing in a way that is too clinical, or that your “decorative bull-jive” in the sidebar is confusing the scraper. You need to look at the visible HTML, the placement of the answers, and the internal link structure to understand the “why” behind the numbers. A list of URLs and impression counts is just an inventory; the actual analysis happens when you compare the successful templates against the unsuccessful ones and realize that one exposes content cleanly while the other hides it behind scripts. We have to be analysts, not just data collectors, which means using our eyes and our human understanding of how a question is answered.
Do you have any advice for our readers?
My biggest piece of advice is to maintain a healthy level of restraint and avoid the “daily drama” of checking your AI visibility every single morning. Google has explicitly warned that this Search Console data is preliminary and subject to change, so if you see a drop on a Tuesday and send out an emergency alert to your team, you are likely reacting to noise rather than a meaningful trend. Instead, focus on multi-week or monthly comparisons and look for sustained signals that follow substantial content revisions, like adding a clear summary or consolidating overlapping pages. Don’t try to reverse-engineer AI Mode based on one heading change; the systems are too complex for that kind of simplistic correlation. Treat this new data as a diagnostic tool to understand how Google perceives your site’s authority and helpfulness, and use those insights to build a content library that is genuinely useful to the 75,000-plus marketers and users who are looking for clear, original answers every day.
