Can Search Console Show AI Overview Impressions?

Search Console cannot show you AI Overview impressions, and no setting changes that. Google is explicit: sites appearing in AI features such as AI Overviews and AI Mode are included in the overall search traffic in Search Console, inside the Web search type (Google, AI Features and Your Website). There is no filter, no toggle, no hidden report — the data is blended, and as of August 2026 Google has not announced a breakdown.

That leaves inference. Below is what GSC genuinely records, the approximation method our GSC miner runs across our builds, and the honest limits of that method — because an approximation presented as a measurement is how bad AI-visibility data gets made.

Can you filter AI Overview traffic in Search Console?

No. The performance report groups AI Overview appearances with ordinary results under the Web search type, and no dimension separates them. Google's documentation on AI features says the traffic is "included in the overall search traffic in Search Console" — which means an impression earned inside an AI Overview and an impression earned as a blue link are indistinguishable rows in the same dataset.

This is worth stating plainly because several vendors imply otherwise. Any tool claiming to report your "AI Overview impressions" is scraping search results pages and inferring, not reading a private Google dataset. Inference can be useful — we do it ourselves — but it should be labeled as inference.

What Google does document is the input side: AI Overviews draw on pages that are indexed and eligible to appear in Search with a snippet, with no additional technical requirements (Google). So your normal GSC coverage and indexing reports remain fully relevant; it is only the AI-specific exposure that goes unreported.

What signature does an AI Overview leave in GSC data?

The signature is divergence: impressions hold steady or rise while click-through rate falls, concentrated on informational queries. The page keeps being shown — often cited inside or below the AI Overview — but the summary answers the question before the click happens.

Public studies calibrate how strong that effect is. In March 2025 browsing data from 900 U.S. adults, Pew found users clicked a traditional result on 8% of visits when an AI summary was present, versus 15% without one — and 18% of all Google searches in the study produced an AI summary (Pew Research Center, July 2025). Ahrefs measured the same phenomenon from the site side: a −34.5% CTR drop for top-ranking pages when an AI Overview was present, in March 2025 (Ahrefs).

So the arithmetic of the signature is grounded: an AI Overview roughly halves the click rate on affected queries while leaving impressions intact. That is a pattern query-level GSC data can surface.

How does the GSC miner approximation work?

Our method is a filtered comparison run on query-level exports, and every step of it is reproducible without paid tools [our data]. The miner automates the first three steps across our builds; by hand, the same pass takes an afternoon.

Period discipline matters: compare like with like (28 days against the prior 28, or year over year for seasonal niches), because a seasonal dip in CTR mimics the AIO signature. Position stability matters for the same reason — a page that slid from position 2 to 6 will lose CTR with no AI Overview involved.

Step 4 is not optional. On our own builds, a meaningful share of flagged queries turn out to be ordinary SERP changes when we actually look [our data] — which is exactly why we treat the miner's output as a watchlist, not a report. The manual check is also how our one documented AI Overview win was observed in the first place: dated looks at the live SERP, because GSC has no AI Overview dimension to consult — the full narrative is in our first AI Overview citation win.

What are the limits of the approximation?

The method infers presence; it cannot measure it. Four limits keep it honest:

• It cannot separate citation from coincidence. An AI Overview can appear on your query without citing you. Only the manual SERP check tells you which.
• It lags. GSC data arrives days behind, and comparison windows smear the timing of any change.
• It misses AI Mode. Google's docs group AI Mode with the same blended reporting, and its query fan-out behavior leaves an even fainter query-level trace.
• It inherits SERP volatility. AI Overviews appear and disappear on the same query across days — Google describes them as shown when they add value beyond what the page results alone provide (Google, AI Overviews help) — so a single snapshot proves little. We re-check monthly.

Against interest: if this approximation sounds like too much work, the honest alternative is to not track AI Overviews at all for now — a small site loses little by waiting for Google to ship real reporting, and nothing in this method will itself improve your visibility. What moves visibility is the content work, which is the subject of the operator's guide to generative engine optimization.

What should you do with the output?

Treat confirmed AIO queries as a maintenance list, not a victory lap. A query where you are cited gets the same freshness and accuracy treatment as any money page, because citations follow the current, extractable version of an answer. A query where an Overview shows but cites competitors tells you which passage to sharpen — usually the answer block: a 40–75-word, self-contained passage that answers the query in plain declarative prose.

And measure the other side of the funnel separately: GSC approximates Google's AI surfaces, but says nothing about ChatGPT, Perplexity, or Claude. For those, referral tracking in GA4 catches the clicks, and a monthly prompt battery catches the citations.