Google Put AI Overviews Data in Search Console. Here Is What It Actually Tells You

For about two years, anyone selling you "AI search optimization" had one unbeatable pitch: nobody could see inside the black box, so any story about how to win was as good as any other. That pitch just got a lot harder to make. In the space of a few weeks in June 2026, Google did two things that pull the curtain back. It started showing AI Overviews and AI Mode impressions inside Search Console, and it published an official guide that flatly debunks most of the GEO tricks people have been paying for.

Treat these as one story, not two. Google handed SEOs a scoreboard for AI search and a reality check in the same breath. If you run a business in Cyprus, the USA, or anywhere else watching AI eat into your search traffic, both matter. Here is what each one actually says, and what we would do with it.

What the new Search Console report shows

Google rolled out a Search Generative AI performance report on June 3, 2026. It gives you a dedicated view of how often your links appear inside generative AI features on Search: AI Overviews, AI Mode, and the AI features inside Discover.

The data is broken down by page, country, device, and date, with granularity down to the hour. For the first time you can open a Google-owned tool and see, in black and white, that your site is showing up inside the AI answers, not just under them. That is genuinely new. Until now you were inferring AI visibility from third-party trackers and guesswork.

It is rolling out to a subset of sites first while Google tests it, so if you do not see the report yet, that is expected. It is coming.

So far, so good. Then you read the limitations, and the report gets a lot more interesting.

The gap that matters more than the report

Here is the headline most coverage buries: the report shows impressions and nothing else. No clicks, no click-through rate, no query data.

Sit with that for a second. You can see that you appeared inside an AI Overview. You cannot see whether anyone clicked, what they searched to trigger it, or whether that appearance sent you a single visitor. As Search Engine Land put it in their write-up, you can see that you showed up, but not whether it was worth anything.

It gets thinner. Every link inside a single AI Overview shares one position value in Search Console, so you cannot tell a prominent citation from a buried one. And the newest data is preliminary, marked with a dotted line, because it is still settling.

We have written before about how AI features break the old link between being seen and being clicked. The decoupling of impressions and clicks is the defining SEO problem of this era, and we covered it in detail in the great decoupling. This new report does not solve that problem. If anything it sharpens it, because it gives you a clean count of a number (AI impressions) that has the loosest possible relationship to revenue.

Our take: the AI Overviews report is a vanity metric with real uses, as long as you never confuse the two. It is excellent for answering "is Google citing my content in AI answers at all, and on which pages." It is useless for answering "is that making me money." Watch it for presence and trend. Do not put it in a board deck as a success metric, and do not let an agency wave rising AI impressions at you as proof anything is working. Impressions you cannot tie to clicks are a direction, not a result.

The reality check most people skipped

While everyone debated the report, Google quietly published the more important document: an official guide on optimizing for generative AI features. It is worth reading in full, because it takes a flamethrower to the GEO playbook that vendors have been selling since 2024.

Google explicitly says you do not need these things to show up in AI search:

• llms.txt or other "special" files and markup. Google may index many file types, but they get no special treatment in AI features. We tested this ourselves and reached the same conclusion in our piece on whether llms.txt is a real standard or SEO theater.
• Chunking your content into tiny pieces. Google's systems can read a page covering several topics and surface the relevant part. You do not need to atomise everything.
• Rewriting your copy in some special "AI-friendly" way. The systems understand synonyms and meaning. You do not need a separate voice for the robots.
• Chasing inauthentic mentions across the web. Manufacturing brand mentions to game the models is, in Google's words, not as helpful as it might seem.

The through-line is blunt: AEO and GEO are not new disciplines. Search Engine Journal summed up Google's position as "AEO and GEO are still SEO," and that matches what we see across client sites every week. The work that earns AI citations is the same work that earned rankings: genuine expertise, content with a point of view, a site that is crawlable and indexable, and trust built over time.

We have argued this for a while. If you want the longer version, we laid out the relationship in GEO vs SEO and what answer engine optimization actually means in our AEO explainer. Google has now said the quiet part out loud.

What this means if you run a business, not an SEO team

You do not need to learn a new acronym this quarter. That is the practical upshot, and it should save you money.

If a vendor is pitching you a "GEO package" built around llms.txt files, content chunking, AI-specific rewrites, or buying mentions, you can now point at Google's own documentation and ask why they are charging you for things Google says do not work. That conversation alone is worth having this month.

What actually moves the needle has not changed:

• Be the non-commodity source. Google's guide keeps returning to content with a unique point of view. Generic, recycled information does not get cited because the model already has a hundred versions of it. First-hand experience, real numbers from your own work, and an actual opinion are what stand out. This is exactly why we publish insight, not data dumps.
• Get the fundamentals right. A page has to be indexed and eligible for a normal snippet before it can appear in an AI feature. If your technical SEO is shaky, no amount of AI tactics rescues it.
• Use the new report for what it is good at. Once you have access, check which of your pages Google is citing in AI answers, and treat that as a signal of which topics you have earned authority on. Then double down on those topics with better content, not with markup tricks.

If you want a clearer picture of where AI Overviews are eating your clicks and what is worth doing about it, we went deep on that in getting into Google AI Overviews in 2026.

The honest bottom line

June 2026 was a good month for anyone who hates snake oil. Google gave you a way to confirm you are showing up in AI answers, and in the same breath told you that most of the special AI optimization tactics are unnecessary. The measurement is still half-built (impressions without clicks is a frustrating place to start), but the direction is clear and welcome.

The businesses that win AI search are not the ones with the cleverest llms.txt file. They are the ones publishing content a real expert would put their name to, on a site Google can actually crawl. That was true before this report shipped, and the report is the closest thing to proof Google has ever handed us.

If you would rather spend your budget on what works instead of what is fashionable, a free SEO review is a sensible place to start, and if you are based locally we work with businesses across Cyprus on exactly this.