> blog // 18 October 2026
Google AI Overviews and clicks: what the studies say
Seven studies on what Google's AI Overviews do to website clicks, where they agree and differ, and what Australian small businesses can do about it.
The short answer
When Google shows an AI summary, fewer people click through to websites. Six of the seven studies we checked point that way. The size of the drop varies widely and the studies measure different things, so the figures can’t be compared directly. They run from a few per cent to about 60% fewer clicks.
The one randomised experiment we could read, a US university preprint, forced people onto Google’s AI Mode for a week. Their clicks per Google search fell by 18.8 percentage points. That is points, not per cent: the paper gives no starting rate, so it can’t be turned into a percentage drop.
None of these studies sampled Australian searches or Australian small business websites. AI Overviews have been rolling out across Australia since 29 October 2024 and AI Mode since 8 October 2025. We found no Australian study that measures the effect with a stated method. Agencies have made claims about Australian traffic in news coverage, with no sample or method published.
The studies side by side
“Vendor” means the company sells search or SEO tools. “Our sum” means we did the arithmetic from the study’s own figures.
Click-through is the share of searches that end in a click on a result. An AI Overview is the AI summary above the results. AI Mode is Google’s separate chat-style search. Informational means someone is looking for an answer, not a supplier.
| Who | Published | Sample | Measured | Result | Source type |
|---|---|---|---|---|---|
| Pew Research Center | 22 Jul 2025 | 900 US adults, 68,879 Google searches | Visits with a click on a result | 8% with a summary, 15% without (our sum: 47% fewer) | Independent |
| Ahrefs | 17 Apr 2025 | 300,000 keywords | Desktop, top result click-through vs Ahrefs’ forecast of the rate without AI Overviews | 34.5% lower | Vendor |
| Ahrefs re-run | 4 Feb 2026 | 300,000 keywords | Same, desktop, top result, on a two-year gap | 58% lower | Vendor |
| Seer Interactive | 4 Nov 2025 | 3,119 search terms, 42 clients | Organic click-through on informational queries with an AI Overview, June 2024 to Sept 2025 | 1.76% to 0.61% (our sum: 65% fewer) | Vendor (agency) |
| Seer Interactive | 24 Apr 2026 | 53 brands, 5.47M queries | Organic click-through, AI Overview shown vs not, Feb 2026 | 2.36% vs 3.82% (our sum: 38% fewer; the groups were the other way round in Jan 2025, so not all of this gap is the AI Overview) | Vendor (agency) |
| Amsive | 16 Apr 2025 | 10,000 keywords, 10 websites | Click-through after AI Overviews appeared | 15.49% lower | Vendor (agency), no data source or dates given |
| Semrush with Datos | 15 Dec 2025 | Over 200,000 keywords | Zero-click rate on the same keywords, before and after | Searches ending in no click fell from 33.75% to 31.53% (so slightly more clicks) | Vendor |
| Khosravi and Yoganarasimhan | 2 Sep 2026 (v6) | About 500,000 Wikipedia article pairs per language | Search referrals, English vs German and French | 5.45% and 4.82% lower | Independent preprint |
| Wang and colleagues | 18 Aug 2026 | 1,100 US participants, 7 days | Clicks per Google search, random assignment | Forced AI Mode: down 18.8 percentage points | Independent preprint |
What they agree on
Direction. Six of the seven studies (nine rows, counting the Ahrefs and Seer updates separately) show fewer clicks when an AI answer is on the page.
Links inside the summary barely get clicked. Pew Research Center found it happened in 1% of visits to pages with a summary.
AI Overviews mostly appear on informational searches. Ahrefs found 99.2% of keywords that trigger an AI Overview are informational. Seer’s 2026 data shows AI Overviews on 36% of informational queries and 5% of transactional ones.
Cited brands did slightly better. In Seer’s Q3 2025 data, brands cited in the AI Overview got 0.70% organic click-through and uncited brands 0.52%. Queries with no AI Overview got 1.45%. Seer says it “cannot definitively prove that citation causes higher CTRs”.
Mostly correlation. Ahrefs says an AI Overview “correlates with” lower click-through. Seer says “We cannot claim causation.” Only academic work has a designed comparison group.
Where they differ, and why
They count different things. Visits, keywords, client query sets, clicks per person, monthly referrals. A 35% drop on one measure is not a 35% drop on another.
They use different baselines. Pew compares searches that happened to show a summary with searches that did not, and those are different kinds of search. Ahrefs compares against its forecast of click-through without AI Overviews, Seer year on year, the Wikipedia paper across languages. In Seer’s 2025 data, queries with no AI Overview also fell 46% year on year.
The numbers move. Ahrefs ran the same method twice: 34.5% on a one-year gap (March 2024 to March 2025), 58% on a two-year gap (December 2023 to December 2025). Seer reports organic click-through on AI Overview queries “climbed from a floor of 1.3% in December 2025 to 2.4% in February 2026”. Every figure is a snapshot with a date.
The samples are not small businesses. Wikipedia is a global encyclopaedia. Seer used its agency clients’ data. Amsive does not say whose websites it studied. The Wikipedia figure may look small because many searches that reach an article never show an AI Overview. That is our reading; the paper does not test it.
One study dissents. Semrush tracked keywords before and after an AI Overview appeared and found the zero-click rate went slightly down. It concludes AI Overviews “don’t automatically increase zero-click behavior”.
Google disagrees. Google’s head of Search wrote in August 2025 that “total organic click volume from Google Search to websites has been relatively stable year-over-year”, and that clicks are higher quality. Google published no figures. Both can be true if total searches grow while the share that clicks falls. We cannot check it from outside.
Seer’s 2025 page describes the fall from 1.76% to 0.61% as 61%. We get 65%, so we quote the raw figures.
What none of them measure
No study we found samples Australian searches or websites. Pew and the experiment are US only. Ahrefs, Seer, Amsive and Semrush do not say which country.
Local service searches are missing. We found no measurement that reports separately on searches like “electrician Maitland”, the kind of search many local businesses depend on. One Sydney agency told ABC News in October 2025 that a search like “best plumber in my suburb” showed no AI summary, but it gave no data.
None measure calls, bookings or sales.
What to do about it
Check your own numbers. Since June 2026, Search Console has a Search Generative AI performance report that shows how often your pages appear in AI Overviews and AI Mode. Google says it reached all websites on 31 August 2026. It counts impressions only, not clicks: Google says AI feature clicks are counted inside the “Web” search type of the main Performance report, mixed in with other clicks. So compare the pages that show up in the AI report against the main Performance report: impressions holding or rising while clicks fall on an informational page is consistent with the studies, though seasonality or ranking changes can look the same.
Make the page answer the question. Answer it plainly, high on the page. Our framework post covers the writing side.
Give people a reason to click. A summary can say roughly what a job costs. It cannot give your price, take a booking or show the job you did two streets away. Quote calculators, booking forms, current prices and local detail all need a visit.
Watch enquiries as well as traffic. We don’t know whether the lost clicks were ones that turn into work, so check calls and forms.
Jezweb builds and looks after websites for Australian small businesses. We can help you read your Search Console numbers.
How we worked this out
We read each study on its own page and re-checked every figure on 13 October 2026. “Our sum” is new value divided by old, minus one; we did not pool the studies into one average. The two arXiv papers are preprints, not yet peer reviewed; the Wang paper is funded by the US National Science Foundation. A second randomised experiment, by Agarwal and Sen (April 2026), is on SSRN, which blocked us; we have not used its figures.
- Pew Research Center, Google users are less likely to click on links when an AI summary appears in the results, 22 July 2025.
- Ahrefs, AI Overviews Reduce Clicks by 34.5%, 17 April 2025, and update, 4 February 2026.
- Seer Interactive, September 2025 update, 4 November 2025, and 2026 update, 24 April 2026.
- Amsive, AI Overviews study, 16 April 2025.
- Semrush, AI Overviews study, 15 December 2025.
- Khosravi and Yoganarasimhan, Impact of AI Search Summaries on Website Traffic, arXiv v6, 2 September 2026.
- Wang, Gleason, Bart, Wilson and Metaxa, AI in Search Reduces Publisher Referrals Without Improving User Experience, arXiv, 18 August 2026.
- Google, AI in Search is driving more queries and higher quality clicks, 6 August 2025.
- Google Australia, AI Overviews in Australia, 29 October 2024, and AI Mode in Australia, 8 October 2025.
- Google Search Central, Introducing Search Generative AI performance reports in Search Console, 3 June 2026, note of 31 August 2026.
- Google Search Central, AI features documentation, updated 10 December 2025.