Impressions Up, Clicks Down: What's Really Happening in Search Console (2026)

Our own Search Console tells this story: impressions up 49% in four weeks, clicks down 12% over the same period, average position improving from 11.3 to 8.4. Every number moved the right way except the one that pays. Here is what is actually happening, what Google documents, and how to tell — in your own data — whether you are winning or quietly losing.

What Google actually documents

Start with the counting rules, because most explanations of this phenomenon skip them. Google's Search Console documentation states that an impression is counted "whenever an item appears in the current page of results, whether or not the item is scrolled into view, as long as the user need not click to see more results." So an impression has never meant a human looked at your listing — it means your link was on the page that loaded. That was true in 2019 and it is true now.

Two things about AI Overviews follow directly from Google's own wording. First: "An AI Overview occupies a single position in search results, and all links in the AI Overview are assigned that same position." One position, shared by every cited source. Second, on impressions inside AI Overviews, the documentation is brief to the point of anticlimax: standard impression rules apply. There is no special AIO impression rule to learn — which also means there is no documented way to separate AI-driven impressions from ordinary ones in the standard performance report.

That single detail — one position, many citations — is the mechanical reason impression counts inflate. A traditional results page gave a handful of sites a top-of-page impression. An AI Overview can cite five, eight, ten sources, and each one banks an impression at the same shared position. More publishers get counted per query. Nobody gets more clicks.

The evidence — including the part that complicates the story

Ahrefs re-ran its AI Overview study on December 2025 data across 300,000 keywords — 150,000 with an AI Overview present and 150,000 informational keywords without — and found the presence of an AI Overview correlates with a 58% lower average click-through rate for the top-ranking page. That is the headline most people quote, and it is well-sourced.

Seer Interactive's tracking adds something more uncomfortable, and it is the reason to be careful before blaming AI Overviews for everything. On queries with an AI Overview, organic CTR fell from 1.76% (June 2024) to a low of 0.57% (July 2025), recovering slightly to 0.61% by September 2025. But on queries with no AI Overview at all, CTR went from 2.72% in June 2024 to a peak of 3.14% in February 2025 and then down to 1.62% by September 2025 — roughly halving from its peak.

The queries no human typed

Here is the part that surprised us most in our own data, and it is the least-documented mechanism in this whole area. When we investigated our worst-converting position band, we pulled every query where we ranked between positions 4 and 10, took the top twenty by impressions, and inspected each live results page. Two findings, both awkward:

Every single inspectable results page carried an AI Overview. Fourteen out of fourteen. Featured snippets: zero. The snippet slot we spent years learning to win no longer exists on those queries — it has been absorbed.

Six of the twenty queries returned no results page at all. They were machine-phrased questions — full sentences like "how do i choose an ai marketing workflow tool that actually improves business results?" — one of which had logged 7,134 impressions in 28 days. Near-identical duplicate variants of the same question sat beside them in the data. These are the sub-queries AI assistants generate while assembling an answer, not things people type into a search box. In total, those twenty queries produced 38,685 impressions and zero clicks.

To be straight about the evidential status: Google documents that follow-up questions asked inside AI Mode are treated as new queries, but we found no Google documentation confirming or denying whether the internal sub-queries an AI system issues during retrieval are logged as publisher impressions. Our data is consistent with it. That is an inference from pattern, not a documented mechanism, and anyone telling you otherwise is guessing with more confidence than the evidence supports.

A 20-minute diagnostic for your own data

Averages hide all of this. Site-wide CTR blends branded and unbranded, human and machine, AIO and not. Run these four checks instead:

Then change what you report. Raw clicks and site-wide CTR are now poor proxies for whether search is working. The measures that survive contact with AI results are citation share (does your brand get named in AI answers to the questions you care about — check it by asking them, on a schedule, and logging the results), branded search volume, conversions from organic rather than sessions, and impression share on the queries that actually matter to your pipeline. We track citation share by running the same set of buyer questions through the major assistants roughly every two weeks and recording which sources they name. It is unglamorous and it is the only number in this article that has moved in the direction we wanted all quarter.

Frequently asked questions

Why are my Search Console impressions going up while clicks go down?

Three things compound. AI Overviews cite multiple sources at a single shared position, so more pages earn an impression per query than a traditional results page produced. AI assistants generate machine-phrased sub-queries during retrieval that log impressions no human will ever click. And when an AI Overview is present, click-through on the top organic result is around 58% lower. Impressions measure eligibility to be seen; they always did, but the gap between eligibility and attention has widened sharply.

Does a citation inside an AI Overview count as an impression?

Yes. Google's documentation says standard impression rules apply to AI Overviews, and that the AI Overview occupies one position with every link inside it assigned that same position. Note the consequence: your reported "average position" can improve because you are cited in an AI Overview sitting at position 1, while your actual clicks do not move at all.

Can I separate AI Overview traffic from normal organic in Search Console?

Not in the standard performance report — there is no filter that isolates it, which is why so many teams cannot prove what is happening to them. Google has been adding generative-AI reporting, so check your own Search Console for what is currently available on your property rather than relying on any article, including this one, for the state of a fast-moving product.

Is falling CTR always caused by AI Overviews?

No, and this is the most-missed point. Seer Interactive's tracking shows CTR on queries with no AI Overview fell from a 3.14% peak in February 2025 to 1.62% by September 2025. Clicks are getting scarcer across search generally. AI Overviews accelerate it; they are not the whole of it.

Should I stop reporting impressions?

Keep them, but stop treating growth in them as a result. Impressions remain useful for spotting new query territory and for tracking whether you are entering retrieval sets at all. Pair them with citation share, branded search and organic conversions, and band your CTR by position rather than reporting one site-wide average.

Sources checked against the original documents in September 2026: Google Search Console performance report documentation (impression counting; AI Overview position rule) · Ahrefs, AI Overviews and clicks update (300,000 keywords, December 2025 data, published February 2026) · Seer Interactive, AIO impact on Google CTR, September 2025 update. First-hand figures are from Marketing Mary's own Google Search Console property and live SERP inspection, August–September 2026. Search products change quickly; verify current behaviour in your own account.