Español
English
Google promised transparency about AI Overviews and half-delivered. In June it turned on a specific report inside Search Console so we can see how much impact AI answers have on our sites. After eight weeks of data already loaded, the SEO community's conclusion is uncomfortable: the report shows numbers, but those numbers mislead if you read them like the rest of Search Console.
It's not a Google bug. It's that AI Overviews broke the measurement model we'd been using for 20 years.
What Search Console reports today about AI Overviews
Since June 3, 2026, Search Console has dedicated reports for AI Overviews and AI Mode — Google's conversational mode. Both were filled with data starting around May 18, so there's more than two months of history to read.
What the report returns:
- Impressions of your page inside an AI Overview or an AI Mode answer.
- Breakdown by page, country, device and date (with granularity down to the hour).
What it does not return:
- Clicks. There's no click data from an AI Overview to your site.
- CTR. With no click, there's no ratio either.
- Query. You don't know which question Google answered using you as a source.
With that limited data, the first articles analyzing 8 weeks (Search Engine Journal, July 30, 2026) flagged the problem: the metrics look good and reality doesn't.
The first trap: impressions without clicks
In Search Console, an AI Overview impression counts the same as a classic blue-listing impression. In practice they aren't the same.
When your page is shown inside a generated answer, the user reads your content — or the summary Google built from your content — without entering your site. It logs an impression, doesn't count a click, and to you it looks like "we're appearing but it doesn't bring traffic." The correct reading is different: you appeared, you served a function, and you saw no traffic return because the traffic model changed, not because your content failed.
Translated to the report: if you watch the AI Overviews impressions line rising month over month, you're not seeing business growth; you're seeing how much your content's role as an AI source is growing. It's an authority signal, not a traffic one.
The second trap: average position inflates
This is the most dangerous because it doesn't jump out at you. When Google includes your URL inside an AI Overview, the reported position is usually 1. The AIO occupies the top block, your URL appears cited there, and Search Console interprets it as "you're first."
In the aggregate report, this raises your average position without you having improved anything in the classic ranking. You can have a site that in the blue listing fell from position 6 to 12 and still see "average position: 3.4" in Search Console because the AIO appearances are biasing the number upward.
If that metric is reported to the business without breaking it down, the conclusion will be "SEO is better than ever" while organic sales fall. It's the classic decoupling — impressions up, clicks down — but disguised inside a metric that until now was reliable.
The third trap: daily averages hide the volatility
AI Overviews don't always appear on the same query. Google decides on each search whether to generate an AI answer, and that decision swings hard by country, device and day. The same term can have an AIO on Monday and not on Thursday.
When the Search Console report averages those swings into a daily or weekly figure, the result gets smoothed. You see a calm line, but underneath there are days where you lose 60% of the CTR to AIO and days where none appears. If you trust the average to justify decisions, you'll make decisions late: when the drop is already structural, not a bad day.
Main idea: in the pre-AIO world, a smoothed average was a reasonable reading; today, a smoothed average hides the part where the site loses money.
How to read the report without falling for the trap
Four practical criteria to avoid the most common mistakes:
- Always segment "with AI Overview" vs "without AI Overview." Don't look at the aggregate. The comparison between queries that triggered an AIO and those that didn't is what shows the real impact. If the CTR of the former is 0.5% and the latter 3%, there's your business figure.
- Cross AIO impressions with site clicks, not with average position. The question isn't "how much do I appear?" but "how much traffic converted and how much didn't?".
- Look by page, not by site. AIOs hit a handful of pages very hard — the informational "what is," "how to," "difference between X and Y" type — and barely touch commercial ones. A site average hides that bias.
- Use full 7-day windows to compare weeks. Daily swings are too noisy for decisions; the signal appears at the weekly level, not the daily one.
This last point has an interesting twist: as we saw in detail when the first wave of data came out, the decoupling between rising impressions and falling clicks is the classic fingerprint of an active AI Overview. Now that the report exists, that fingerprint can be isolated; before it could only be inferred. We covered it in the article on losing clicks to AI Overviews.
What to do with pages losing traffic to AIO?
The answer hasn't changed from what we saw in the article on why AI cites your competitor, but now with the report you can prioritize better:
- Identify the pages where AIO impressions grow and site clicks fall. They're the first to work on.
- Reinforce extractability and proof: direct answers up top, figures with sources, headings as questions. It's what decides whether Google cites you as a source or mentions you without giving you a place.
- Cover the full intent: AIOs combine sub-questions; content that resolves them all wins more cited surface.
- Measure in 4-6 week windows. The impact of a change on an informational page takes one to two months to show in the report.
None of these steps are SEO news. What's new is that you can now aim them with Google's official data in hand, instead of inferring from CTR drops.
The bottom line
The AI Overviews report in Search Console is valuable information, but it demands new reading criteria. Impression isn't traffic. Average position is no longer a reliable thermometer. Daily averages smooth the real impact and delay decisions. If you carry classic SEO measurement practices into the AIO report, you'll see numbers that look good while the business loses traffic. With two months of data already loaded, it's time to rebuild the reporting dashboards and separate what shows AI visibility from what shows real clicks to the site.
Frequently asked questions
Does Search Console report clicks from an AI Overview? No. As of today, the report only shows impressions from AI Overviews and AI Mode. Google anticipated it will add more metrics "over time," but the click is still missing.
Can I know which query triggered the AI Overview where I appear? Not in the AI Overviews report. The query dimension isn't included yet. It can be inferred by crossing pages with the classic Performance report, but it's approximate.
Did my site's average position rise because of AI Overviews? Probably. Appearances inside an AIO count as position 1 in the aggregate report. It's worth breaking it down by "AIO appearance / no AIO appearance" to read the blue-listing position separately.
How much historical data do I have? It depends on when the property was activated. Google backfilled from around May 18, 2026, so most accounts already have more than two months to read trends.
Want your AI reporting to show what actually matters? Book a meeting with Seotronix.
