Guide · AI search

Are AI Overviews taking your clicks? How to check on your own site

Published . Updated . First published.

Search Console cannot show directly whether AI Overviews are taking your clicks, because Google reports clicks from AI features inside the Performance report's Web search type. You can compare click-through rate, impressions and position for the same pages across equal periods, and read the result as timing evidence, not proof.

  • Google says sites appearing in AI features, such as AI Overviews and AI Mode, are reported in the Performance report's Web search type (Google, Search Central documentation, read 30 October 2026).
  • Google counts a click on an external link in an AI Overview as a click and gives every link in it the same position (Google, Search Console Help, read 30 October 2026).
  • Google's 3 June 2026 launch post for the Generative AI performance report lists impressions, pages, countries, devices and dates but no click metric (Google, Search Central Blog, read 30 October 2026).
  • Search Console keeps data for the last 16 months (Google, Analytics Help, read 30 October 2026).
  • Google said on 6 August 2025 that overall organic clicks from Google Search to websites were relatively stable year over year, a claim with no figures (Google, The Keyword, read 30 October 2026).

Can Search Console show how many clicks AI Overviews take?

No, not as a separate number: Google reports clicks and impressions from AI Overviews and AI Mode inside the regular Performance report, mixed in with other web search traffic. Google's AI features page, last updated 10 December 2025, says sites appearing in these features are reported on in the Performance report, within the "Web" search type. The pages we read describe no filter that isolates them, so the check is indirect: look for a pattern in clicks, impressions, position and click-through rate that fits AI Overviews, and rule out the alternatives.

Google counts a click on an external link in an AI Overview as a click, and it gives the whole AI Overview a single position that every link inside it shares. Our reasoning from the position rule, which Google does not state: your average position can move when more of your impressions come from links inside AI answers rather than ordinary results, even though no ordinary result moved. Read position beside clicks and impressions, never alone.

What does the Generative AI performance report add?

The Generative AI performance report adds impressions: how often a link to your site was shown inside a Google generative AI feature, which Google lists as AI Overviews and AI Mode. It adds no clicks, though Google's launch post of 3 June 2026 says it is working with website owners on which data would help most, such as additional metrics over time, so check the report itself (how to read the report).

What each source can show about clicks from AI Overviews, from Google's pages
SourceWhat it showsWhat it does not show
Performance report, Web search typeClicks, impressions, click-through rate and position, with AI features includedWhich clicks came from an AI Overview; the pages we read describe no filter
Generative AI performance reportImpressions of links to your site in AI features, by page, country, date and deviceClicks, click-through rate, position or queries

What does Google say about AI Overviews and clicks?

Google says overall organic clicks from Google Search to websites have been relatively stable year over year, and that clicks from results pages with AI Overviews are higher quality. Both are claims about all sites together, with no figures, and say nothing about yours. The first is from a Google blog post of 6 August 2025. The second is from Google's AI features page, which says higher quality means users are more likely to spend more time on the site; the blog post tests quality differently, as clicks where users do not quickly click back. In our reasoning, neither an aggregate claim nor a third-party report about averages can be checked against one site: only your own data speaks to your site.

How do you check your own site, step by step?

Compare the same pages across two equal periods, find which of clicks, impressions, position and click-through rate moved, then add the Generative AI report to see where your links appear in AI features. These steps are our method, not a Google procedure.

  1. Choose the periods. Use two 28-day periods that start on the same weekday and end a few days before today, because our methodology notes that Search Console data usually arrives 2 to 3 days late. The same dates a year earlier avoid many seasonal swings (seasonality in SEO data), but 16 months of history leaves only 16 − 12 = 4 months with a partner a year back.
  2. Export by page. In the Performance report, choose the Web search type, set the Pages dimension and export clicks, impressions, click-through rate and position for each period. Pages are where the two reports line up: the Generative AI report has no query dimension.
  3. Find what moved. For each page and for the whole site, compute click-through rate as clicks ÷ impressions in each period and note which of the four numbers changed (see impressions up, clicks down). Rule out the usual causes first (what to check first when traffic drops).
  4. Add AI impressions. Export the Generative AI report for the latest period with the same dates and Web search type (text-based or multimodal, the same in both) and drop it into our free AI search report tool beside the regular export. It runs in your browser and divides each page's AI impressions by its web impressions, which we read as a share of the same impressions: our inference from Google's wording, not something Google states.
  5. Compare groups, not only the site. Sort pages by AI share and compare how click-through rate moved on the highest-share pages against the lowest. Skip pages with very few impressions; as a rough floor, our content decay check answers "can't tell" for a page under 3,000 impressions in the trailing 90 days, rather than a guess.
  6. Look at the results. Search a few pages' main queries and note whether an AI Overview appears and links to you. Our methodology says one observation of a model's answer is not a measurement, so treat it as a lead.
  7. Save your baseline. Download each month's exports and append them. The tool offers a baseline CSV for the AI report, and our guide to the 16-month limit covers the regular report.

How to read what moved

The table is our reasoning, not a Google rule, and each row names candidates, not findings. Click-through rate also falls as position worsens, so compare it at similar positions (your own click-through curve by position).

How to read common patterns in the Performance report (our reasoning, not Google's)
ClicksImpressionsPositionConsistent with
DownFlat or upAbout the sameA click-through fall: AI Overviews are one candidate, alongside changes to your titles and snippets or to the rest of the results page.
DownDownWorseA ranking or visibility loss: check rankings, indexing and updates first.
DownDownAbout the sameFewer searches or a data problem: check demand, season and Google's known issues (Search Console data anomalies).

Example (illustrative numbers)

These numbers are invented and are not data from any site. A site compares its latest 28 days of Web search type data with the 28 days starting 364 days earlier (52 × 7 = 364, the same weekday). Its Generative AI export joins to the regular export on 14 pages. Page /a has 380 AI and 2,375 web impressions, so its share is 380 ÷ 2,375 = 16.0%. The tool flags 2 pages whose AI impressions exceed their web impressions and gives them no share; 2 ÷ 14 = 14.3% is under its 20% line, so the other 12 keep theirs. The site splits those 12 into the 6 highest and the 6 lowest.

Click-through rate (clicks ÷ impressions) in the example
GroupEarlier periodLatest periodChange
Highest AI share (6 pages)4,185 ÷ 93,000 = 4.50%3,360 ÷ 96,000 = 3.50%4.50 − 3.50 = 1.00 point down
Lowest AI share (6 pages)2,640 ÷ 66,000 = 4.00%2,405 ÷ 65,000 = 3.70%4.00 − 3.70 = 0.30 point down

Average position stayed within 0.2 of its earlier value in both groups, so ranking is an unlikely explanation.

Counting noise. Treating impressions as independent trials, which they are not, a rate p on n impressions has a standard error of √(p × (1 − p) ÷ n), which after multiplying by 100 is 0.068, 0.059, 0.076 and 0.074 points for the four cells. The gap between the groups' changes then has a standard error of √(0.068² + 0.059² + 0.076² + 0.074²) = √0.019357 = 0.139 points and a 95% range of 1.96 × 0.139 = ±0.27 points. The highest-share group fell 1.00 − 0.30 = 0.70 points more, and 0.70 ÷ 0.27 = 2.6 times that range. Counting noise does not explain the gap, but comparability might: pages that Google links in AI answers can differ in topic and demand, so this is timing evidence, not proof that AI Overviews took the clicks.

What this cannot tell you

The check cannot show that AI Overviews are the reason your clicks changed: a comparison of periods or page groups is timing evidence, and changes to your titles, competitors, demand, ranking or the rest of the results page can produce the same pattern.

  • It cannot see searches where an AI Overview appeared without a link to you: the Generative AI report counts only links to your site shown in AI features (our reasoning from Google's definition).
  • It cannot separate AI Overviews from AI Mode. Google reports both in one report and inside Web, and the pages we read name no filter that splits them.
  • It cannot tell which way AI exposure pushes a page: clicks on links inside AI Overviews count as clicks, so a high AI share may add clicks as well as remove them (our reasoning).
  • It cannot judge click quality. Google's higher-quality claim has no figures, and clicks, impressions, click-through rate and position say nothing about what visitors did after arriving; your own analytics can.
  • It cannot go back past your data. Search Console keeps 16 months, Google's help page gives no start date for the Generative AI report's data, and a Generative AI export saved before 21 August 2026 may under-count impressions for 13 to 17 August 2026, a logging error that Google's data anomalies page says an update dated 21 August restored.
  • It can compare unlike data. Google's data anomalies page says impressions in the Performance report were not reported accurately from 13 May 2025 until 27 April 2026, which also affected click-through rate and average position but not clicks, so a period inside that window and one outside it may differ for reasons unrelated to AI Overviews.

Questions people also ask

Do AI Overviews reduce website traffic?

Google's published statements describe all sites together and give no figures, so they cannot tell you what happened on yours, and the Google pages we read show no AI Overview click count in Search Console. A fall in clicks while impressions and position hold steady fits AI Overviews and several other explanations. Compare your own pages across equal periods, add the Generative AI report's impressions, and keep the conclusion at timing evidence.

What is the impact of AI on website traffic?

For Google's AI features, the dedicated report gives impressions of your links in AI Overviews and AI Mode with no clicks, and the clicks sit inside the regular report's Web figures, so the traffic impact has to be estimated from your own before-and-after data. Other products are separate questions: Bing's AI Performance report counts citations in Microsoft Copilot, AI summaries in Bing and select partner integrations, and Bing says a citation does not represent traffic, clicks or user engagement. For answers from ChatGPT, Perplexity and Claude, our guide to measuring AI visibility explains why one answer is not a measurement and how repeated sampling works.

How do AI Overviews work?

Google says its generative AI features on Search are rooted in its core ranking and quality systems, and that both AI Overviews and AI Mode may use query fan-out: the model generates several related queries at once and fetches additional results for them. The AI Overview then links to supporting pages. The pages we read give no figure for how often AI Overviews appear or how many clicks they take.

How do you get your site into an AI Overview?

Google's AI features page says a page must be indexed and eligible to be shown in Google Search with a snippet to be a supporting link in AI Overviews or AI Mode, with no additional technical requirements. Google's optimization guide, last updated 10 July 2026, adds that a site must be included in Search generative AI features in Search Console to be eligible for display. Eligible is not the same as linked, and Google says SEO best practices remain relevant (how SEO, AEO and GEO differ).

Sources

  1. AI features and your website, Google Search Central documentation, Google, read 30 October 2026. reported on in the Performance report, within the "Web" search type
  2. What are impressions, position, and clicks?, Search Console Help, Google, read 30 October 2026.
  3. Generative AI performance report (Search), Search Console Help, Google, read 30 October 2026.
  4. Introducing Search Generative AI performance reports in Search Console, Google Search Central Blog, Google, read 30 October 2026.
  5. AI in Search is driving more queries and higher quality clicks, The Keyword (blog.google), Google, read 30 October 2026.
  6. [GA4] Queries report, Analytics Help, Google, read 30 October 2026.
  7. Data anomalies in Search Console, Search Console Help, Google, read 30 October 2026.
  8. Optimizing your website for generative AI features on Google Search, Google Search Central documentation, Google, read 30 October 2026.
  9. AI Performance, Bing Webmaster Tools Help, Microsoft, read 30 October 2026.
  10. AI search report analyser, Penlid, read 30 October 2026.
  11. Methodology: how we measure, and what we can't, Penlid, read 30 October 2026.

About this guide

Who: Penlid builds the tools these guides use.

How: Drafted with Claude, Anthropic's AI model. A second pass re-read every source and checked each number, quote and policy statement against it on 30 October 2026.

Why: To help you measure content and search changes honestly.