Do AI Overviews Cut Clicks? How to Measure It in Search Console

Pew Research Center did not ask people what they think they do. It recorded what they did. The panel was 900 U.S. adults running an app that logged their browsing in March 2025, for a total of 68,879 unique Google searches, of which 12,593 returned an AI summary¹. This is U.S. data, but it is observed behavior, not self-reported behavior.

The headline findings, all from the same dataset:

• With an AI summary on the page, users clicked a traditional search result on 8% of visits; without one, on 15%¹.
• Clicks on a link cited inside the summary itself happened on just 1% of visits¹.
• Users ended their browsing session on 26% of pages with a summary, versus 16% of pages with only traditional results¹.
• Over the period, 18% of Google searches returned an AI summary¹.

The limits: one country, one month, and Pew scraped the results pages weeks after the searches, so the summary a user saw may have differed. This is everyday consumer behavior, not a B2B buyer researching vendors.

Ahrefs measured the effect on click-through rate (CTR) for the page ranking first. In April 2025, using 300,000 keywords and aggregated Search Console data, the presence of an AI Overview correlated with a 34.5% lower CTR at position one³. When Ahrefs re-ran the study with December 2025 data, the estimated drop rose to 58%².

The method is the part worth copying. Ahrefs split keywords that started showing an AI Overview from a control group of informational keywords that did not. The control group's decline shows how much CTR would have fallen anyway; anything beyond that is attributed to the AI Overview. In the April study, position-one CTR for AI Overview keywords was 0.073 before; applying the control group's decline, the forecast was 0.040, and the actual figure was 0.026³.

The 2026 re-run also measured positions below the top spot: the loss persists all the way down to position ten, shrinking as you move down the page.

These are desktop CTR figures for informational keywords. Ahrefs did not measure the effect on transactional terms, which rarely trigger an AI Overview.

Seer Interactive worked from client data rather than tool samples. The study covered 3,119 informational search terms across 42 organizations and 25.1 million organic impressions between June 2024 and September 2025⁴. For queries with an AI Overview, organic CTR went from 1.76% to 0.61% over the period, a 61% decline⁴.

Two Seer findings change how you should read your own Search Console. First, queries without an AI Overview also lost clicks, with organic CTR down 41% over the same window⁴. Part of the loss has nothing to do with AI Overviews, which is why a control group is mandatory in any measurement.

Second, when a brand is cited inside the AI Overview, it earns 35% more organic clicks than when it is not cited⁴. Seer itself cautions that this does not prove causation: brands with more authority may get cited precisely because they already have higher CTR.

Outside the U.S. and global panels, the evidence is thinner. The closest data point we found comes from the Brazilian agency Conversion, which applied the Ahrefs method to two of the country's largest informational content portals, comparing March 2024 with March 2025. The result tracked the global trend: position-one CTR fell on both portals, by amounts close to the Ahrefs figure⁵.

• Globo.com: a 31.47% drop for keywords ranking first, in a sample of 26,486 keywords⁵.
• UOL: a 35.42% drop, in a sample of 30,000 keywords⁵.

These are two media sites, with a search profile very different from a B2B company. Read them as a directional signal, not a forecast for your industry.

Side by side, the five measurements cited so far (all global or U.S., except Conversion's):

Search Console counts clicks and impressions from AI Overviews inside the "Web" search type of the Performance report, blended with regular results, according to Google's documentation on AI features. There is no "AI Overview" filter. So the CTR impact on your site has to be inferred by comparison, the same way Ahrefs and Seer did it.

Three signals show up together when an AI Overview is the cause: average position holds steady, impressions hold or rise, and CTR falls. If position dropped too, you have a ranking problem, not an AI Overview problem. If impressions fell, demand changed. Separating the three cases keeps you from blaming AI for a lost ranking.

The method we recommend reuses the Ahrefs control-group logic on your own Search Console data. You compare the same period across two years, separate queries that currently show an AI Overview from those that do not, and compare CTR within the same ranking band. The gap between the two declines is your estimate of the AI Overview effect.

1. 01
   In Search Console, open Performance, choose the "Web" search type, filter to your main market's country, and compare the same range of months this year against last year to cancel out seasonality.
2. 02
   On the Queries tab, export clicks, impressions, CTR and average position for both periods.
3. 03
   Isolate informational queries with a custom regex query filter, for example ^(how|what|why|which|when|who)\b, and keep everything else as a second group.
4. 04
   For the informational queries with the most impressions, check in an incognito window whether Google shows an AI Overview and whether your site is cited in it. Tag each query as "AI Overview, cited", "AI Overview, not cited" or "no AI Overview".
5. 05
   Drop queries whose average position moved to a different band between periods (for example, from the top of page one to page two). Otherwise you are measuring lost rankings, not the AI Overview effect.
6. 06
   Calculate aggregate CTR for each group (total clicks divided by total impressions) in both periods. Forecast the expected CTR of the AI Overview group by multiplying its earlier CTR by the change in the no-AI-Overview group.
7. 07
   Compare actual CTR with the forecast. The percentage gap is your estimate of the AI Overview effect, and the "cited" versus "not cited" comparison shows how much a citation offsets the loss.

Two caveats. If the no-AI-Overview group has few queries, the control is weak and the result is a hypothesis. And the clicks that remain may be worth more: Google says people who click from results pages with AI Overviews spend more time on the site, so check the numbers against Google Analytics conversions before concluding you lost revenue.

The first deliverable fits in a spreadsheet: the informational queries that bring your site the most impressions, with AI Overview presence and citation tagged by hand. Once that exists, steps 5 through 7 become formulas, and you can rerun the same math every month as new Search Console data comes in.

Start by exporting your queries from Search Console and tagging the top of the list, sorted by impressions. If your site is not cited in the AI Overviews for the queries that matter most, that is the content work to prioritize: indexable pages, copy that answers the question directly, and sourced data, which is what AI Overviews cite.