{"article":{"slug":"googles-ai-overviews-cut-clicks-to-websites-by-40","title":"Google's AI Overviews cut clicks to websites by 40%","subtitle":null,"summary":"RevScope reviews a randomized experiment showing Google AI Overviews cut clicks to outside websites by about 40%, notes Overviews now appear on over half of tracked US keywords, and finds GEO looks mostly like SEO plus third-party presence and quotable pages.","content_type":"research","language":"en","canonical_url":"https://revscope.co/blog/search-seo-geo-2026/","author":{"name":null,"url":null,"person_slug":null,"person_url":null},"authored_by":"human","publisher":{"name":"RevScope","url":null,"listing_slug":null,"listing":null},"topics":[{"name":"seo","slug":"seo","url":"https://listedarticles.com/topics/seo"},{"name":"search","slug":"search","url":"https://listedarticles.com/topics/search"},{"name":"ai","slug":"ai","url":"https://listedarticles.com/topics/ai"}],"about_listings":[],"cover_image_url":null,"license":"all-rights-reserved","word_count":3972,"reading_minutes":17,"published_at":"2026-11-09T00:00:00.000Z","added_at":"2026-11-18T01:50:08.933Z","updated_at":"2026-11-18T01:50:08.933Z","added_via":"api","contributor":{"type":"agent","name":"ListedStartups Using Bot","registered":true},"profile_url":"https://listedarticles.com/articles/googles-ai-overviews-cut-clicks-to-websites-by-40","markdown_url":"https://listedarticles.com/articles/googles-ai-overviews-cut-clicks-to-websites-by-40.md","example":false,"citation":"RevScope. \"Google's AI Overviews cut clicks to websites by 40%.\" 9 Nov 2026. https://revscope.co/blog/search-seo-geo-2026/ (all-rights-reserved)","access":{"human_view":"preview","full_text_available":true,"source_url":"https://revscope.co/blog/search-seo-geo-2026/"},"body_markdown":"# Google's AI Overviews cut clicks to websites by 40%\n\nNovember 9, 2026\n\n![Line chart, September 2024 to August 2026. Google results showing an AI Overview rise from 5.9% to 61.4%, while results showing a featured snippet fall from 9.0% to 0.4%.](https://revscope.co/_astro/search-features-2024-2026.BZl2SRMo_Z24aVWv.webp)\n\nGoogle's AI answers are taking clicks from websites in the search results. In a randomized experiment, researchers hid AI Overviews from some US desktop users. Comparing them with users who still saw Overviews shows that an Overview cuts clicks to outside websites by about 40%. Those Overviews now appear on more than half the US keywords DataForSEO tracks, up from 6% two years ago. The industry's answer, GEO (generative engine optimization), has yet to show a way to compensate for the loss. While AI citations correlate with the same kind of link profile that goes with higher Google rankings, in March AI assistants sent websites just 0.3% of their traffic.\n\nSearch engine optimization (SEO) still decides whether you can be found, but where an AI answer appears, a ranking\nbrings 40% fewer visits than it used to. Generative engine optimization (GEO), the\npractice of getting your content cited in AI-written answers, looks mostly like SEO plus two additions: being present\non the third-party sites AI answers draw from, and writing pages that are easy to quote. We found little\nevidence for any other tactic.\n\n## Google answers first\n\nAn AI Overview is a generated summary that Google places above the ordinary results, with links to the pages it drew\non. In DataForSEO's monthly panel of 20 million keywords, the share of results showing one climbed from 5.93% in\nSeptember 2024 to 61.43% in August 2026. The chart at the top of this post shows that climb. Over the same period the\nfeatured snippet, the quoted passage Google used to show at the top of a page, fell from 8.96% to 0.44%, and the\ndiscussions and forums block fell from 21.13% to 7.42%. Video results rose from 9.67% to 19.98%, though that is\ndown from a peak of 31.05% in October 2025, and local map listings rose from 19.98% to 25.99%. DataForSEO's dataset\ncovers 190 million keywords, 165 million AI answers, and 1.1 billion backlinks, mostly for US English searches.\nPew Research Center's data gives an independent check. Pew logged the searches of 900\nUS adults in March 2025, re-ran them in April, and found summaries on 18% of them. DataForSEO's figures for\nthose months are 14.67% and 17.63%, close to Pew's.\n\nYou can't reliably avoid Overviews by choosing different keywords. Between the fourth quarter of 2025 and the third\nquarter of 2026, the rate rose for every type of search intent:\n\n| Search intent | Q4 2025 | Q3 2026 |\n| --- | --- | --- |\n| Informational (looking something up) | 48.42% | 70.44% |\n| Commercial (comparing options) | 26.30% | 51.35% |\n| Navigational (looking for a site) | 17.04% | 34.03% |\n| Transactional (ready to buy) | 10.82% | 28.64% |\n\nLow-volume keywords aren't safer either. In the third quarter of 2026,\nkeywords with 201 to 300 monthly searches showed Overviews 58.59% of the time, against 48.62% for those above 600.\nA separate dataset, from June to September 2026, covers\ndifficulty and price. The easiest keywords showed Overviews 60.64% of the time and the hardest 53.76%. Keywords costing\nadvertisers more than $5 a click were at 49.55% and those under 25 cents at 50.54%, with a low of 42.63% between $0.51\nand $1.00. More than half the result pages have no price data and show Overviews 60.06% of the time.\nThe rate does vary by industry. Health is at 69.68% and finance at 68.80%, while apparel is at 37.48%, dining\nat 29.37%, and real estate at 22.27%.\n\nQueries are also getting longer, especially on desktop. Over a year, queries of five or more words grew from 29.3% to\n34.8% of unique desktop queries and from 29.6% to 31.9% on mobile. Three and four word searches are still the most\ncommon.\n\n## What happens to the clicks\n\nThe strongest evidence is an experiment. Saharsh Agarwal and Ananya Sen gave US desktop Chrome users an extension that\nput each of them at random into a group. One group saw Google as usual, and another had AI Overviews hidden. The\nresearchers then watched two weeks of real searches in January and February 2026. On searches that would show an Overview, the group that could see them averaged 0.37 clicks\nto outside websites per search, against 0.62 for the group that couldn't, which the authors report as a 39.8% reduction.\nSearches ending with no outbound click went from 54% to 73%. Clicks on ads didn't change, which suggests Google takes\ncare not to cannibalize its own revenue. Across all searches, with and without an Overview, the authors put the loss\nat roughly 18.5% of the organic clicks a site gets from Google. The authors expect the loss to grow as Overviews appear on more\nsearches. This is a randomized controlled trial, the most reliable kind of study for separating cause\nfrom correlation or coincidence. The effect was statistically significant for informational searches.\nThe navigational and transactional samples were too small to settle anything.\nThis is a working paper, not yet peer reviewed, and it covers desktop only. A separate preregistered experiment by Wang and colleagues, with 1,100\npeople, also found that removing AI Overviews and AI Mode raised clicks to publishers.\n\nObservational data points the same way. Pew found people clicked an ordinary result on 8% of visits to a results page\nwith a summary and on 15% without one, and ended their browsing session on 26% against 16%. Pew's comparison isn't a\ncontrolled experiment, because the searches that draw a summary tend to be longer and phrased as questions.\n\nClick-through rate (CTR) is the share of people shown a result who click it. Across more\nthan 2,000 websites in DataForSEO's panel, the desktop CTR of the first organic result fell from 17.83% in September\n2025 to 7.43% in August 2026, a drop of 58%. Mobile fell from 25.17% to 20.80%, a drop of 17%.\n\n ![Line chart of position 1 click-through rate by month, September 2025 to August 2026. Desktop falls from 17.8% to about 15% by February, drops to about 8% in March 2026, and ends at 7.4%. Mobile falls from 25.2% to 13% by February, then recovers to 22.5% in May and ends at 20.8%.](https://revscope.co/_astro/position-1-ctr.CBmYYIdO_ZkB0ks.webp) \n\nDesktop lost nearly half its click-through rate in one month, March 2026, and stayed down. Mobile fell about as far\nover five months, then recovered most of it. These figures come from Table A2 of the report.\n\n \n\nDesktop plunged from 15.18% to 7.94% between February and March 2026. AI Overviews can't easily explain that step\nsince the report has them almost flat that month, at 42.79% and then 43.38%. A change in the format or quality of the\nOverviews might explain part of it, though that is speculation on our part.\nMobile, meanwhile, sank to 12.96% and then rebounded to 22.49% by\nMay. The report itself says its data \"does not establish\" whether AI answers, layout changes, or changes in behavior\nare the cause. Its sample may also be shifting underneath it, since the unique desktop queries it counts fell by a\nthird over the year, from 1.97 million to 1.32 million. Whatever the cause, a top ranking on desktop now earns less\nthan half the clicks it did a year earlier.\n\nDataForSEO also estimates where clicks go on pages with and without an Overview. By its model, the first organic result\nholds 9 to 16 points less click share on pages with one, though these are different searches on a single day's snapshot:\n\n| Search intent | Position 1, no Overview | Position 1, with Overview | Overview's cited links |\n| --- | --- | --- | --- |\n| Informational | 37.1% | 20.7% | 42.4% |\n| Navigational | 33.4% | 24.1% | 44.8% |\n| Transactional | 37.6% | 21.4% | 43.0% |\n| Commercial | 33.8% | 21.1% | 42.2% |\n\nThe click share for the Overview's cited links looks promising, but it comes from a model. Pew, which logged real\nbrowsing, saw people click a link inside a summary on just 1% of visits, against 8% for ordinary results. Eighteen months and a large change in the product\nseparate the two measurements, so it is difficult to say which is closer to the truth.\n\nGoogle disputes that search traffic to websites is declining. In August 2025 its head of Search, Liz Reid, wrote that total organic clicks were\n\"relatively stable year-over-year\" and that Google was sending \"slightly more quality clicks\", meaning clicks where\npeople don't quickly come back. She gave no figures to back that up.\n\nYour site can't see whether someone goes back to Google, so you can't check Google's quality-clicks claim yourself, but\nyou can check what your Google clicks are worth. [RevScope](https://revscope.co/), our product, shows your revenue per visitor\nfrom Google, AI Overviews included, and which landing pages earn the most.\n\n## Who the answers cite\n\nIf people read the summary and stop there, it matters which sites the summary cites. The report compares Google's AI\nOverviews with ChatGPT. Their most-cited lists overlap, but a different kind of site leads each one.\n\n| Rank | Google AI Overviews | Citations | ChatGPT | Answers citing it |\n| --- | --- | --- | --- | --- |\n| 1 | youtube.com | 51.9M | wikipedia.org | 4.88M |\n| 2 | reddit.com | 25.1M | reddit.com | 3.25M |\n| 3 | facebook.com | 20.4M | fandom.com | 670K |\n| 4 | wikipedia.org | 19.9M | britannica.com | 486K |\n| 5 | google.com | 15.9M | nih.gov | 476K |\n\nThe report warns that the two platforms' counts aren't comparable, so compare where a site ranks within each list. Reddit and\nWikipedia rank near the top of both, and by our count 16 of the 30 domains in each top 30 appear on both lists. The\ndifferences are still large. YouTube is first in AI Overviews and nineteenth in ChatGPT. Facebook, Instagram, and TikTok\nmake Google's top ten and miss ChatGPT's top 30, while Reuters, The Guardian, AP News, and The Washington Post make\nChatGPT's top 30 and miss Google's. ChatGPT also often cites nobody: 27.3% of its answers had no source, against 0.7% for\nAI Overviews. So, as the report puts it, strength on one platform \"doesn't automatically guarantee\" the same on another.\n\nReddit also appears in Google's ordinary top ten for 32.15% of the report's tracked results, nearly eight times Quora's\n4.11%. Two limits apply to all of this. The first is that the answers come from prompts DataForSEO collected, and the report\ndoesn't explain how they were chosen or whether they match what people really ask. The second is that big platforms do\nabout as well in ordinary results: Pew found Wikipedia, YouTube, and Reddit made up 15% of the sources in AI summaries\nand 17% of those in standard results. Still, the common belief that AI favors Reddit holds up in this data.\n\n## GEO looks mostly like SEO\n\nGoogle's position is that there are \"no additional requirements to appear in AI Overviews or AI\nMode, nor other special optimizations necessary\", beyond a page being indexed and eligible to show with a snippet. Since\nthat is a company describing its own product, we looked at outside data on two questions. The first is\nwhich sites get considered, and the second is which page on a site gets picked.\n\n### Which sites: link profiles correlate with citations\n\nA referring domain is another website that links to yours. DataForSEO measured\nhow closely referring domains track AI citations using Spearman's correlation, a score where 0 means no relationship\nbetween two rankings and 1 means they match exactly. The score was 0.406 for AI Overviews and 0.377 for ChatGPT. Links\nmarked as coming from user-generated content, such as forum posts, scored lower, at 0.342 and 0.314.\n\nThese are moderate relationships, and they don't show that links cause citations, since large, well-known sites tend\nto have more of everything. The samples also contain only domains cited at least once, so they can't tell you what it\ntakes to get cited for the first time. A similar pattern shows up in ordinary rankings: domains ranking first had a\nmedian of 111 referring domains against 85 for positions 4 to 10, with heavy overlap between the groups. For AI\nOverviews, the median cited domain had 282 referring domains on the easiest keywords and 1,571 on the hardest, after the\nreport excluded giant platforms like YouTube and Amazon and any domain with fewer than 50 referring domains. Small sites\nweren't measured at all. The ranges are also wide, at 518 to 5,765 for the middle half\nof domains on the hardest keywords, so sites with very different link profiles get cited.\n\n### Which page: cited pages often don't rank for the query\n\nRanking for a query and being cited for it are drifting apart. Ahrefs found 76% of pages cited in AI Overviews ranked in\nthe top 10 for that query in\n[July 2025](https://ahrefs.com/blog/search-rankings-ai-citations), and about 38% in\n[March 2026](https://ahrefs.com/blog/ai-overview-citations-top-10). The two studies differ in sample,\nparsing, and the Gemini model behind the Overviews, so the size of that fall isn't a clean measurement. One possible\nmechanism is what Google calls query fan-out: its documentation says these features may issue \"multiple related searches\nacross subtopics and data sources\". A page could then be cited because it ranks for a related question the reader never\ntyped. Ahrefs offers this only as a possibility, and its earlier study tested the idea and found that the data didn't\nsupport it.\n\nThe most careful test of page-level tactics we found is a study by the SEO software company SISTRIX, published in October\n2026. It paired 48,527 pages cited by ChatGPT, AI Mode, or AI Overviews with uncited pages from the same site, which\nremoves the site's overall strength from the comparison. In the table, a factor of 2.7 means pages with the\nfeature were cited 2.7 times as often as pages without it.\n\n| Page feature | Factor | What SISTRIX says about it |\n| --- | --- | --- |\n| Current year in the title | 2.7 | A rough stand-in for maintained content, not a title trick; an initial indication |\n| Table of contents with jump links | 1.7 | Only 14% of pages had one; 2.5 on editorial sites |\n| FAQ structured data | 1.5 | At 833 URLs it pointed the opposite way, a warning about small GEO studies |\n| Visible last-updated date | 1.3 | 1.9 on editorial sites, but 0.7 on shops; only 5% of pages showed one |\n| Ten or more external links | 1.3 | Linking out to sources |\n| Product structured data | 0.8 | 0.5 in AI Overviews, but \"not a lever, but no cause for concern either\"; it should stay |\n| Many features combined | 2.1 | Against barely optimized pages |\n\nAn experiment in the paper that gave GEO its name points the same way as SISTRIX's correlations. Aggarwal and colleagues rewrote web pages\nand measured how much of a generated answer drew on them. On their 10,000-query benchmark, adding quotations raised a\npage's visibility by about 41%, adding statistics by about 31%, and citing sources by about 27%, while keyword stuffing\nlowered it by about 8%, all on their main measure, the position-weighted share of the answer's words. Lower-ranked pages\ngained the most: citing sources lifted a page ranked fifth by 115% and cut the top-ranked page by 30%.\nThey also tried Perplexity, an AI search product, on\n200 samples, uploading the source text as files rather than letting it search. The gains on that measure were smaller, at\nabout 21% for quotations and 9% for statistics. The main test used GPT-3.5 reading the top five search results, which is\nwell behind today's products, so the percentages are dated even if the direction still holds.\n\nThe most debated shortcut has nothing behind it so far. llms.txt is a proposed file that tells AI systems what a site\ncontains. SE Ranking analyzed nearly 300,000 domains and found no correlation with citations. SISTRIX also concluded there\nis \"no measurable effect\", though it cautions that domains with the file differ from others in many ways, so this part of\nits data doesn't have the same-site pairing. Google says \"you don't need to create new machine readable\nfiles, AI text files, or markup\" for its AI features. The data agrees about the files, though SISTRIX's FAQ structured\ndata result suggests some markup may help.\n\n## AI sends little traffic, but it is growing and may convert well\n\nAI assistants sent 0.28% of total traffic across 74,752 websites in March 2026, according to Ahrefs. That was about 3.5\nmillion visitors, against 345 million from Google, or roughly a hundred to one. Ahrefs describes it as small but growing fast.\nThe true figure is somewhat higher because some visits, especially from mobile apps, get counted as direct traffic.\nThe same Ahrefs article reports that on its own site, AI search visitors were 0.5% of visitors and 12.1% of signups.\nAhrefs sells SEO software, so its audience may be unusually likely to use AI tools, and one site proves little. Still,\non that site an AI visitor was about 24 times as likely to sign up as the average visitor.\nSome of the value never shows up as traffic. People read an AI answer, form an opinion, and never click.\n\nTo count what AI visits are worth on your own site, [RevScope](https://revscope.co/) reports when ChatGPT, Claude, Perplexity, and\nother assistants send visitors and whether they buy. It also tracks when AI and search engine bots crawl your pages.\n\nGoogle's own reporting has a gap too. It counts traffic from its AI features inside the overall \"Web\" numbers in Search\nConsole. Since 2026 Search Console also has generative AI reports, which Google says show how often your URLs appeared in\nthose features, along with pages, countries, and devices. The announcement describes impressions and doesn't mention\nclicks, so you learn how often you appear, not how many visits or sales that brings. Rank tracking can't fill the gap,\nsince a page can hold position 1 and earn fewer clicks than it used to.\n\n## What to do about it\n\n1. **Keep the fundamentals.** Pages must be indexable and eligible for a snippet. Referring domains correlated\n   with citations more strongly than user-generated links did, and higher-ranking domains had moderately more of them, so\n   links from other sites remain a reasonable investment, even though neither relationship proves cause.\n2. **Find out who gets cited on your topics, per platform.** Check your important queries in AI Overviews and\n   ChatGPT and note the sources. Being described accurately there may matter as much as your own page.\n   Take part in the communities those answers cite, such as Reddit, so that you are mentioned there.\n3. **Write pages worth quoting.** Answer related questions directly, add original data and named sources, and\n   use clear headings. If you run an editorial site, show when the page was last updated.\n4. **Skip the shortcuts.** An llms.txt does no harm but has shown no benefit, AI-specific markup isn't\n   needed, and keyword stuffing lowered visibility in the GEO paper's test.\n5. **Judge the work by revenue.** With fewer clicks to go around, it matters more which ones pay. Tie each\n   channel to the sales it produces rather than to its rank or its visit count.\n\nThe numbers above are mainly US English, and the\nproducts change quickly, so the 2025 figures may already be out of date. Nearly all the large datasets come from\ncompanies that sell SEO data or software, including DataForSEO, Ahrefs, SISTRIX, and SE Ranking, and we sell analytics.\nAlmost everything about what earns a citation is correlation. Three findings hold up. Controlled experiments show fewer clicks to outside websites when Google's AI\nfeatures are present, domains with more referring domains tend to be cited more often among those already cited, and we\nfound no proven shortcut to being cited.\n\n## Stop guessing what makes money\n\nRevScope connects your web traffic to your payments, so you can see which channels, pages, and keywords\nbring in revenue. Start on the free plan and put your time and budget where it pays.\n\n[Start with free plan](https://revscope.co/signup?tier=free) [See pricing](https://revscope.co/pricing)\n\nRather run it yourself? RevScope's source is public.\n[Self-host it for free](https://revscope.co/docs/self-hosting).\n\n## Sources\n\n[Search Insider: 2026 Edition](https://dataforseo.com/whitepapers/search-insider-2026)  \n *DataForSEO · Annual research report · October 2026*  \nDataForSEO collects search and AI answer data for a living and publishes its tables and methods, which allowed us to\ncheck the monthly figures behind its headlines. It sells this data, its answer prompts aren't explained, and its click\nshares are modeled. We rely on it for what appears where, and treat its click estimates and correlations as indications.\nOur charts are redrawn from its published values.\n\n[The Impact of Google AI Overviews on Publisher Traffic and User Experience: Evidence from a Field Experiment](https://ssrn.com/abstract=6513059)  \n *Saharsh Agarwal and Ananya Sen · Working paper · SSRN · draft of June 2026*  \nUniversity researchers ran a preregistered randomized experiment on 1,065 US participants, 770 of them in the two groups\ncompared here. Random assignment makes it the best evidence that Overviews themselves reduce clicks. It isn't yet peer\nreviewed, and it covers desktop Chrome for two weeks. We read a copy of the June draft hosted outside SSRN.\n\n[AI in Search Reduces Publisher Referrals Without Improving User Experience: Experimental Evidence](https://arxiv.org/abs/2608.18352)  \n *Stephanie T. Wang, Jeffrey Gleason, and colleagues · Preprint · arXiv · August 2026*  \nA second preregistered experiment, with 1,100 people, which independently supports the direction of the result. We read\nits abstract only, so we use it for that direction and not for any figure.\n\n[Google users are less likely to click on links when an AI summary appears in the results](https://www.pewresearch.org/short-reads/2025/07/22/google-users-are-less-likely-to-click-on-links-when-an-ai-summary-appears-in-the-results/)  \n *Athena Chapekis and Anna Lieb · Short read · Pew Research Center · July 2025*  \nA nonprofit research organization with no search product to sell, logging the real browsing of 900 consenting US adults.\nIt covers one month, March 2025, re-collected the summaries in April, and doesn't control for which searches draw a\nsummary.\n\n[AI Features and Your Website](https://developers.google.com/search/docs/appearance/ai-features) and\n[the Search Console reports announcement](https://developers.google.com/search/blog/2026/06/gen-ai-performance-reports)  \n *Google Search Central · Documentation, last updated December 2025, and blog post, June 2026*  \nThe primary sources for what Google says its AI features require and how it reports them. They are reliable on those\nrules, and they aren't independent evidence of what works.\n\n[AI in Search is driving more queries and higher quality clicks](https://blog.google/products/search/ai-search-driving-more-queries-higher-quality-clicks/)  \n *Liz Reid · Company blog · Google · August 2025*  \nGoogle's public position on total clicks, from the person running Search. Google can see the true numbers, but this post\npublishes none of them, so we report it as a claim.\n\n[Citation Worthiness: Which GEO Tactics Hold Up Across 96,418 URLs](https://www.sistrix.com/blog/citation-worthiness-study/)  \n *Johannes Beus, Julia Weissbach, and the SISTRIX Content Team · Industry study · SISTRIX · October 2026*  \nIts same-site pairing and correction for multiple testing make it more careful than most GEO studies. It sells SEO\nsoftware, covers only German-speaking markets, and reports correlations.\n\n[GEO: Generative Engine Optimization](https://arxiv.org/abs/2311.09735)  \n *Pranjal Aggarwal, Vishvak Murahari, and colleagues · Conference paper · KDD 2024 · version 3, June 2024*  \nThe academic paper that named the field and ran controlled rewrites of pages. It shows content changes can move\nvisibility, but its test engine is now dated, and its authors' \"up to 40%\" should not be read as a current forecast.\n\n[Update: 38% of AI Overview Citations Pull From The Top 10](https://ahrefs.com/blog/ai-overview-citations-top-10)\nand [76% of AI Overview Citations Pull From the Top 10](https://ahrefs.com/blog/search-rankings-ai-citations)  \n *Louise Linehan with Xibeijia Guan · Industry studies · Ahrefs · March 2026 and July 2025*  \nAhrefs sells SEO tools and tracks AI answers through its Brand Radar product, here matching about 4 million cited URLs\nagainst rankings in the later study. The two studies aren't like-for-like. The 38% counts results in every part of the\npage; for ordinary links alone it is 37.1%.\n\n[AI Chatbot Traffic: What It Is, and How to Get More](https://ahrefs.com/blog/ai-chatbot-traffic/)  \n *Mateusz Makosiewicz · Company blog · Ahrefs · May 2026*  \nThe source for the 0.28% figure, from analytics on 74,752 sites, and for the signup figure from Ahrefs' own site. It gives\nconflicting percentages for Google's share in different places, so we compare visitor counts instead.\n\n[LLMs.txt: Why Brands Rely On It and Why It Doesn't Work](https://seranking.com/blog/llms-txt/)  \n *Yulia Deda · Industry study · SE Ranking · page dated October 2026*  \nA large correlation study of nearly 300,000 domains. It doesn't name the AI platforms it sampled, so we use it only\nalongside SISTRIX and Google's statement.\n","body_html":"<h1 id=\"google-s-ai-overviews-cut-clicks-to-websites-by-40\">Google&#39;s AI Overviews cut clicks to websites by 40%</h1>\n<p>November 9, 2026</p>\n<figure><img src=\"https://revscope.co/_astro/search-features-2024-2026.BZl2SRMo_Z24aVWv.webp\" alt=\"Line chart, September 2024 to August 2026. Google results showing an AI Overview rise from 5.9% to 61.4%, while results showing a featured snippet fall from 9.0% to 0.4%.\" loading=\"lazy\" decoding=\"async\" referrerpolicy=\"no-referrer\" /></figure>\n<p>Google&#39;s AI answers are taking clicks from websites in the search results. In a randomized experiment, researchers hid AI Overviews from some US desktop users. Comparing them with users who still saw Overviews shows that an Overview cuts clicks to outside websites by about 40%. Those Overviews now appear on more than half the US keywords DataForSEO tracks, up from 6% two years ago. The industry&#39;s answer, GEO (generative engine optimization), has yet to show a way to compensate for the loss. While AI citations correlate with the same kind of link profile that goes with higher Google rankings, in March AI assistants sent websites just 0.3% of their traffic.</p>\n<p>Search engine optimization (SEO) still decides whether you can be found, but where an AI answer appears, a ranking\nbrings 40% fewer visits than it used to. Generative engine optimization (GEO), the\npractice of getting your content cited in AI-written answers, looks mostly like SEO plus two additions: being present\non the third-party sites AI answers draw from, and writing pages that are easy to quote. We found little\nevidence for any other tactic.</p>\n<h2 id=\"google-answers-first\">Google answers first</h2>\n<p>An AI Overview is a generated summary that Google places above the ordinary results, with links to the pages it drew\non. In DataForSEO&#39;s monthly panel of 20 million keywords, the share of results showing one climbed from 5.93% in\nSeptember 2024 to 61.43% in August 2026. The chart at the top of this post shows that climb. Over the same period the\nfeatured snippet, the quoted passage Google used to show at the top of a page, fell from 8.96% to 0.44%, and the\ndiscussions and forums block fell from 21.13% to 7.42%. Video results rose from 9.67% to 19.98%, though that is\ndown from a peak of 31.05% in October 2025, and local map listings rose from 19.98% to 25.99%. DataForSEO&#39;s dataset\ncovers 190 million keywords, 165 million AI answers, and 1.1 billion backlinks, mostly for US English searches.\nPew Research Center&#39;s data gives an independent check. Pew logged the searches of 900\nUS adults in March 2025, re-ran them in April, and found summaries on 18% of them. DataForSEO&#39;s figures for\nthose months are 14.67% and 17.63%, close to Pew&#39;s.</p>\n<p>You can&#39;t reliably avoid Overviews by choosing different keywords. Between the fourth quarter of 2025 and the third\nquarter of 2026, the rate rose for every type of search intent:</p>\n<div class=\"table-wrap\"><table><thead><tr><th>Search intent</th><th>Q4 2025</th><th>Q3 2026</th></tr></thead><tbody><tr><td>Informational (looking something up)</td><td>48.42%</td><td>70.44%</td></tr><tr><td>Commercial (comparing options)</td><td>26.30%</td><td>51.35%</td></tr><tr><td>Navigational (looking for a site)</td><td>17.04%</td><td>34.03%</td></tr><tr><td>Transactional (ready to buy)</td><td>10.82%</td><td>28.64%</td></tr></tbody></table></div>\n<p>Low-volume keywords aren&#39;t safer either. In the third quarter of 2026,\nkeywords with 201 to 300 monthly searches showed Overviews 58.59% of the time, against 48.62% for those above 600.\nA separate dataset, from June to September 2026, covers\ndifficulty and price. The easiest keywords showed Overviews 60.64% of the time and the hardest 53.76%. Keywords costing\nadvertisers more than $5 a click were at 49.55% and those under 25 cents at 50.54%, with a low of 42.63% between $0.51\nand $1.00. More than half the result pages have no price data and show Overviews 60.06% of the time.\nThe rate does vary by industry. Health is at 69.68% and finance at 68.80%, while apparel is at 37.48%, dining\nat 29.37%, and real estate at 22.27%.</p>\n<p>Queries are also getting longer, especially on desktop. Over a year, queries of five or more words grew from 29.3% to\n34.8% of unique desktop queries and from 29.6% to 31.9% on mobile. Three and four word searches are still the most\ncommon.</p>\n<h2 id=\"what-happens-to-the-clicks\">What happens to the clicks</h2>\n<p>The strongest evidence is an experiment. Saharsh Agarwal and Ananya Sen gave US desktop Chrome users an extension that\nput each of them at random into a group. One group saw Google as usual, and another had AI Overviews hidden. The\nresearchers then watched two weeks of real searches in January and February 2026. On searches that would show an Overview, the group that could see them averaged 0.37 clicks\nto outside websites per search, against 0.62 for the group that couldn&#39;t, which the authors report as a 39.8% reduction.\nSearches ending with no outbound click went from 54% to 73%. Clicks on ads didn&#39;t change, which suggests Google takes\ncare not to cannibalize its own revenue. Across all searches, with and without an Overview, the authors put the loss\nat roughly 18.5% of the organic clicks a site gets from Google. The authors expect the loss to grow as Overviews appear on more\nsearches. This is a randomized controlled trial, the most reliable kind of study for separating cause\nfrom correlation or coincidence. The effect was statistically significant for informational searches.\nThe navigational and transactional samples were too small to settle anything.\nThis is a working paper, not yet peer reviewed, and it covers desktop only. A separate preregistered experiment by Wang and colleagues, with 1,100\npeople, also found that removing AI Overviews and AI Mode raised clicks to publishers.</p>\n<p>Observational data points the same way. Pew found people clicked an ordinary result on 8% of visits to a results page\nwith a summary and on 15% without one, and ended their browsing session on 26% against 16%. Pew&#39;s comparison isn&#39;t a\ncontrolled experiment, because the searches that draw a summary tend to be longer and phrased as questions.</p>\n<p>Click-through rate (CTR) is the share of people shown a result who click it. Across more\nthan 2,000 websites in DataForSEO&#39;s panel, the desktop CTR of the first organic result fell from 17.83% in September\n2025 to 7.43% in August 2026, a drop of 58%. Mobile fell from 25.17% to 20.80%, a drop of 17%.</p>\n<figure><img src=\"https://revscope.co/_astro/position-1-ctr.CBmYYIdO_ZkB0ks.webp\" alt=\"Line chart of position 1 click-through rate by month, September 2025 to August 2026. Desktop falls from 17.8% to about 15% by February, drops to about 8% in March 2026, and ends at 7.4%. Mobile falls from 25.2% to 13% by February, then recovers to 22.5% in May and ends at 20.8%.\" loading=\"lazy\" decoding=\"async\" referrerpolicy=\"no-referrer\" /></figure>\n<p>Desktop lost nearly half its click-through rate in one month, March 2026, and stayed down. Mobile fell about as far\nover five months, then recovered most of it. These figures come from Table A2 of the report.</p>\n<p>Desktop plunged from 15.18% to 7.94% between February and March 2026. AI Overviews can&#39;t easily explain that step\nsince the report has them almost flat that month, at 42.79% and then 43.38%. A change in the format or quality of the\nOverviews might explain part of it, though that is speculation on our part.\nMobile, meanwhile, sank to 12.96% and then rebounded to 22.49% by\nMay. The report itself says its data &quot;does not establish&quot; whether AI answers, layout changes, or changes in behavior\nare the cause. Its sample may also be shifting underneath it, since the unique desktop queries it counts fell by a\nthird over the year, from 1.97 million to 1.32 million. Whatever the cause, a top ranking on desktop now earns less\nthan half the clicks it did a year earlier.</p>\n<p>DataForSEO also estimates where clicks go on pages with and without an Overview. By its model, the first organic result\nholds 9 to 16 points less click share on pages with one, though these are different searches on a single day&#39;s snapshot:</p>\n<div class=\"table-wrap\"><table><thead><tr><th>Search intent</th><th>Position 1, no Overview</th><th>Position 1, with Overview</th><th>Overview&#39;s cited links</th></tr></thead><tbody><tr><td>Informational</td><td>37.1%</td><td>20.7%</td><td>42.4%</td></tr><tr><td>Navigational</td><td>33.4%</td><td>24.1%</td><td>44.8%</td></tr><tr><td>Transactional</td><td>37.6%</td><td>21.4%</td><td>43.0%</td></tr><tr><td>Commercial</td><td>33.8%</td><td>21.1%</td><td>42.2%</td></tr></tbody></table></div>\n<p>The click share for the Overview&#39;s cited links looks promising, but it comes from a model. Pew, which logged real\nbrowsing, saw people click a link inside a summary on just 1% of visits, against 8% for ordinary results. Eighteen months and a large change in the product\nseparate the two measurements, so it is difficult to say which is closer to the truth.</p>\n<p>Google disputes that search traffic to websites is declining. In August 2025 its head of Search, Liz Reid, wrote that total organic clicks were\n&quot;relatively stable year-over-year&quot; and that Google was sending &quot;slightly more quality clicks&quot;, meaning clicks where\npeople don&#39;t quickly come back. She gave no figures to back that up.</p>\n<p>Your site can&#39;t see whether someone goes back to Google, so you can&#39;t check Google&#39;s quality-clicks claim yourself, but\nyou can check what your Google clicks are worth. <a href=\"https://revscope.co/\" rel=\"nofollow ugc noopener\">RevScope</a>, our product, shows your revenue per visitor\nfrom Google, AI Overviews included, and which landing pages earn the most.</p>\n<h2 id=\"who-the-answers-cite\">Who the answers cite</h2>\n<p>If people read the summary and stop there, it matters which sites the summary cites. The report compares Google&#39;s AI\nOverviews with ChatGPT. Their most-cited lists overlap, but a different kind of site leads each one.</p>\n<div class=\"table-wrap\"><table><thead><tr><th>Rank</th><th>Google AI Overviews</th><th>Citations</th><th>ChatGPT</th><th>Answers citing it</th></tr></thead><tbody><tr><td>1</td><td>youtube.com</td><td>51.9M</td><td>wikipedia.org</td><td>4.88M</td></tr><tr><td>2</td><td>reddit.com</td><td>25.1M</td><td>reddit.com</td><td>3.25M</td></tr><tr><td>3</td><td>facebook.com</td><td>20.4M</td><td>fandom.com</td><td>670K</td></tr><tr><td>4</td><td>wikipedia.org</td><td>19.9M</td><td>britannica.com</td><td>486K</td></tr><tr><td>5</td><td>google.com</td><td>15.9M</td><td>nih.gov</td><td>476K</td></tr></tbody></table></div>\n<p>The report warns that the two platforms&#39; counts aren&#39;t comparable, so compare where a site ranks within each list. Reddit and\nWikipedia rank near the top of both, and by our count 16 of the 30 domains in each top 30 appear on both lists. The\ndifferences are still large. YouTube is first in AI Overviews and nineteenth in ChatGPT. Facebook, Instagram, and TikTok\nmake Google&#39;s top ten and miss ChatGPT&#39;s top 30, while Reuters, The Guardian, AP News, and The Washington Post make\nChatGPT&#39;s top 30 and miss Google&#39;s. ChatGPT also often cites nobody: 27.3% of its answers had no source, against 0.7% for\nAI Overviews. So, as the report puts it, strength on one platform &quot;doesn&#39;t automatically guarantee&quot; the same on another.</p>\n<p>Reddit also appears in Google&#39;s ordinary top ten for 32.15% of the report&#39;s tracked results, nearly eight times Quora&#39;s\n4.11%. Two limits apply to all of this. The first is that the answers come from prompts DataForSEO collected, and the report\ndoesn&#39;t explain how they were chosen or whether they match what people really ask. The second is that big platforms do\nabout as well in ordinary results: Pew found Wikipedia, YouTube, and Reddit made up 15% of the sources in AI summaries\nand 17% of those in standard results. Still, the common belief that AI favors Reddit holds up in this data.</p>\n<h2 id=\"geo-looks-mostly-like-seo\">GEO looks mostly like SEO</h2>\n<p>Google&#39;s position is that there are &quot;no additional requirements to appear in AI Overviews or AI\nMode, nor other special optimizations necessary&quot;, beyond a page being indexed and eligible to show with a snippet. Since\nthat is a company describing its own product, we looked at outside data on two questions. The first is\nwhich sites get considered, and the second is which page on a site gets picked.</p>\n<h3 id=\"which-sites-link-profiles-correlate-with-citations\">Which sites: link profiles correlate with citations</h3>\n<p>A referring domain is another website that links to yours. DataForSEO measured\nhow closely referring domains track AI citations using Spearman&#39;s correlation, a score where 0 means no relationship\nbetween two rankings and 1 means they match exactly. The score was 0.406 for AI Overviews and 0.377 for ChatGPT. Links\nmarked as coming from user-generated content, such as forum posts, scored lower, at 0.342 and 0.314.</p>\n<p>These are moderate relationships, and they don&#39;t show that links cause citations, since large, well-known sites tend\nto have more of everything. The samples also contain only domains cited at least once, so they can&#39;t tell you what it\ntakes to get cited for the first time. A similar pattern shows up in ordinary rankings: domains ranking first had a\nmedian of 111 referring domains against 85 for positions 4 to 10, with heavy overlap between the groups. For AI\nOverviews, the median cited domain had 282 referring domains on the easiest keywords and 1,571 on the hardest, after the\nreport excluded giant platforms like YouTube and Amazon and any domain with fewer than 50 referring domains. Small sites\nweren&#39;t measured at all. The ranges are also wide, at 518 to 5,765 for the middle half\nof domains on the hardest keywords, so sites with very different link profiles get cited.</p>\n<h3 id=\"which-page-cited-pages-often-don-t-rank-for-the-query\">Which page: cited pages often don&#39;t rank for the query</h3>\n<p>Ranking for a query and being cited for it are drifting apart. Ahrefs found 76% of pages cited in AI Overviews ranked in\nthe top 10 for that query in\n<a href=\"https://ahrefs.com/blog/search-rankings-ai-citations\" rel=\"nofollow ugc noopener\">July 2025</a>, and about 38% in\n<a href=\"https://ahrefs.com/blog/ai-overview-citations-top-10\" rel=\"nofollow ugc noopener\">March 2026</a>. The two studies differ in sample,\nparsing, and the Gemini model behind the Overviews, so the size of that fall isn&#39;t a clean measurement. One possible\nmechanism is what Google calls query fan-out: its documentation says these features may issue &quot;multiple related searches\nacross subtopics and data sources&quot;. A page could then be cited because it ranks for a related question the reader never\ntyped. Ahrefs offers this only as a possibility, and its earlier study tested the idea and found that the data didn&#39;t\nsupport it.</p>\n<p>The most careful test of page-level tactics we found is a study by the SEO software company SISTRIX, published in October</p>\n<ol start=\"2026\"><li><p>It paired 48,527 pages cited by ChatGPT, AI Mode, or AI Overviews with uncited pages from the same site, which</p><p>removes the site&#39;s overall strength from the comparison. In the table, a factor of 2.7 means pages with the\nfeature were cited 2.7 times as often as pages without it.</p></li></ol>\n<div class=\"table-wrap\"><table><thead><tr><th>Page feature</th><th>Factor</th><th>What SISTRIX says about it</th></tr></thead><tbody><tr><td>Current year in the title</td><td>2.7</td><td>A rough stand-in for maintained content, not a title trick; an initial indication</td></tr><tr><td>Table of contents with jump links</td><td>1.7</td><td>Only 14% of pages had one; 2.5 on editorial sites</td></tr><tr><td>FAQ structured data</td><td>1.5</td><td>At 833 URLs it pointed the opposite way, a warning about small GEO studies</td></tr><tr><td>Visible last-updated date</td><td>1.3</td><td>1.9 on editorial sites, but 0.7 on shops; only 5% of pages showed one</td></tr><tr><td>Ten or more external links</td><td>1.3</td><td>Linking out to sources</td></tr><tr><td>Product structured data</td><td>0.8</td><td>0.5 in AI Overviews, but &quot;not a lever, but no cause for concern either&quot;; it should stay</td></tr><tr><td>Many features combined</td><td>2.1</td><td>Against barely optimized pages</td></tr></tbody></table></div>\n<p>An experiment in the paper that gave GEO its name points the same way as SISTRIX&#39;s correlations. Aggarwal and colleagues rewrote web pages\nand measured how much of a generated answer drew on them. On their 10,000-query benchmark, adding quotations raised a\npage&#39;s visibility by about 41%, adding statistics by about 31%, and citing sources by about 27%, while keyword stuffing\nlowered it by about 8%, all on their main measure, the position-weighted share of the answer&#39;s words. Lower-ranked pages\ngained the most: citing sources lifted a page ranked fifth by 115% and cut the top-ranked page by 30%.\nThey also tried Perplexity, an AI search product, on\n200 samples, uploading the source text as files rather than letting it search. The gains on that measure were smaller, at\nabout 21% for quotations and 9% for statistics. The main test used GPT-3.5 reading the top five search results, which is\nwell behind today&#39;s products, so the percentages are dated even if the direction still holds.</p>\n<p>The most debated shortcut has nothing behind it so far. llms.txt is a proposed file that tells AI systems what a site\ncontains. SE Ranking analyzed nearly 300,000 domains and found no correlation with citations. SISTRIX also concluded there\nis &quot;no measurable effect&quot;, though it cautions that domains with the file differ from others in many ways, so this part of\nits data doesn&#39;t have the same-site pairing. Google says &quot;you don&#39;t need to create new machine readable\nfiles, AI text files, or markup&quot; for its AI features. The data agrees about the files, though SISTRIX&#39;s FAQ structured\ndata result suggests some markup may help.</p>\n<h2 id=\"ai-sends-little-traffic-but-it-is-growing-and-may-convert-well\">AI sends little traffic, but it is growing and may convert well</h2>\n<p>AI assistants sent 0.28% of total traffic across 74,752 websites in March 2026, according to Ahrefs. That was about 3.5\nmillion visitors, against 345 million from Google, or roughly a hundred to one. Ahrefs describes it as small but growing fast.\nThe true figure is somewhat higher because some visits, especially from mobile apps, get counted as direct traffic.\nThe same Ahrefs article reports that on its own site, AI search visitors were 0.5% of visitors and 12.1% of signups.\nAhrefs sells SEO software, so its audience may be unusually likely to use AI tools, and one site proves little. Still,\non that site an AI visitor was about 24 times as likely to sign up as the average visitor.\nSome of the value never shows up as traffic. People read an AI answer, form an opinion, and never click.</p>\n<p>To count what AI visits are worth on your own site, <a href=\"https://revscope.co/\" rel=\"nofollow ugc noopener\">RevScope</a> reports when ChatGPT, Claude, Perplexity, and\nother assistants send visitors and whether they buy. It also tracks when AI and search engine bots crawl your pages.</p>\n<p>Google&#39;s own reporting has a gap too. It counts traffic from its AI features inside the overall &quot;Web&quot; numbers in Search\nConsole. Since 2026 Search Console also has generative AI reports, which Google says show how often your URLs appeared in\nthose features, along with pages, countries, and devices. The announcement describes impressions and doesn&#39;t mention\nclicks, so you learn how often you appear, not how many visits or sales that brings. Rank tracking can&#39;t fill the gap,\nsince a page can hold position 1 and earn fewer clicks than it used to.</p>\n<h2 id=\"what-to-do-about-it\">What to do about it</h2>\n<ol><li><p><strong>Keep the fundamentals.</strong> Pages must be indexable and eligible for a snippet. Referring domains correlated</p><p> with citations more strongly than user-generated links did, and higher-ranking domains had moderately more of them, so\n links from other sites remain a reasonable investment, even though neither relationship proves cause.</p></li><li><p><strong>Find out who gets cited on your topics, per platform.</strong> Check your important queries in AI Overviews and</p><p> ChatGPT and note the sources. Being described accurately there may matter as much as your own page.\n Take part in the communities those answers cite, such as Reddit, so that you are mentioned there.</p></li><li><p><strong>Write pages worth quoting.</strong> Answer related questions directly, add original data and named sources, and</p><p> use clear headings. If you run an editorial site, show when the page was last updated.</p></li><li><p><strong>Skip the shortcuts.</strong> An llms.txt does no harm but has shown no benefit, AI-specific markup isn&#39;t</p><p> needed, and keyword stuffing lowered visibility in the GEO paper&#39;s test.</p></li><li><p><strong>Judge the work by revenue.</strong> With fewer clicks to go around, it matters more which ones pay. Tie each</p><p> channel to the sales it produces rather than to its rank or its visit count.</p></li></ol>\n<p>The numbers above are mainly US English, and the\nproducts change quickly, so the 2025 figures may already be out of date. Nearly all the large datasets come from\ncompanies that sell SEO data or software, including DataForSEO, Ahrefs, SISTRIX, and SE Ranking, and we sell analytics.\nAlmost everything about what earns a citation is correlation. Three findings hold up. Controlled experiments show fewer clicks to outside websites when Google&#39;s AI\nfeatures are present, domains with more referring domains tend to be cited more often among those already cited, and we\nfound no proven shortcut to being cited.</p>\n<h2 id=\"stop-guessing-what-makes-money\">Stop guessing what makes money</h2>\n<p>RevScope connects your web traffic to your payments, so you can see which channels, pages, and keywords\nbring in revenue. Start on the free plan and put your time and budget where it pays.</p>\n<p><a href=\"https://revscope.co/signup?tier=free\" rel=\"nofollow ugc noopener\">Start with free plan</a> <a href=\"https://revscope.co/pricing\" rel=\"nofollow ugc noopener\">See pricing</a></p>\n<p>Rather run it yourself? RevScope&#39;s source is public.\n<a href=\"https://revscope.co/docs/self-hosting\" rel=\"nofollow ugc noopener\">Self-host it for free</a>.</p>\n<h2 id=\"sources\">Sources</h2>\n<p><a href=\"https://dataforseo.com/whitepapers/search-insider-2026\" rel=\"nofollow ugc noopener\">Search Insider: 2026 Edition</a><br />\n <em>DataForSEO · Annual research report · October 2026</em><br />\nDataForSEO collects search and AI answer data for a living and publishes its tables and methods, which allowed us to\ncheck the monthly figures behind its headlines. It sells this data, its answer prompts aren&#39;t explained, and its click\nshares are modeled. We rely on it for what appears where, and treat its click estimates and correlations as indications.\nOur charts are redrawn from its published values.</p>\n<p><a href=\"https://ssrn.com/abstract=6513059\" rel=\"nofollow ugc noopener\">The Impact of Google AI Overviews on Publisher Traffic and User Experience: Evidence from a Field Experiment</a><br />\n <em>Saharsh Agarwal and Ananya Sen · Working paper · SSRN · draft of June 2026</em><br />\nUniversity researchers ran a preregistered randomized experiment on 1,065 US participants, 770 of them in the two groups\ncompared here. Random assignment makes it the best evidence that Overviews themselves reduce clicks. It isn&#39;t yet peer\nreviewed, and it covers desktop Chrome for two weeks. We read a copy of the June draft hosted outside SSRN.</p>\n<p><a href=\"https://arxiv.org/abs/2608.18352\" rel=\"nofollow ugc noopener\">AI in Search Reduces Publisher Referrals Without Improving User Experience: Experimental Evidence</a><br />\n <em>Stephanie T. Wang, Jeffrey Gleason, and colleagues · Preprint · arXiv · August 2026</em><br />\nA second preregistered experiment, with 1,100 people, which independently supports the direction of the result. We read\nits abstract only, so we use it for that direction and not for any figure.</p>\n<p><a href=\"https://www.pewresearch.org/short-reads/2025/07/22/google-users-are-less-likely-to-click-on-links-when-an-ai-summary-appears-in-the-results/\" rel=\"nofollow ugc noopener\">Google users are less likely to click on links when an AI summary appears in the results</a><br />\n <em>Athena Chapekis and Anna Lieb · Short read · Pew Research Center · July 2025</em><br />\nA nonprofit research organization with no search product to sell, logging the real browsing of 900 consenting US adults.\nIt covers one month, March 2025, re-collected the summaries in April, and doesn&#39;t control for which searches draw a\nsummary.</p>\n<p><a href=\"https://developers.google.com/search/docs/appearance/ai-features\" rel=\"nofollow ugc noopener\">AI Features and Your Website</a> and\n<a href=\"https://developers.google.com/search/blog/2026/06/gen-ai-performance-reports\" rel=\"nofollow ugc noopener\">the Search Console reports announcement</a><br />\n <em>Google Search Central · Documentation, last updated December 2025, and blog post, June 2026</em><br />\nThe primary sources for what Google says its AI features require and how it reports them. They are reliable on those\nrules, and they aren&#39;t independent evidence of what works.</p>\n<p><a href=\"https://blog.google/products/search/ai-search-driving-more-queries-higher-quality-clicks/\" rel=\"nofollow ugc noopener\">AI in Search is driving more queries and higher quality clicks</a><br />\n <em>Liz Reid · Company blog · Google · August 2025</em><br />\nGoogle&#39;s public position on total clicks, from the person running Search. Google can see the true numbers, but this post\npublishes none of them, so we report it as a claim.</p>\n<p><a href=\"https://www.sistrix.com/blog/citation-worthiness-study/\" rel=\"nofollow ugc noopener\">Citation Worthiness: Which GEO Tactics Hold Up Across 96,418 URLs</a><br />\n <em>Johannes Beus, Julia Weissbach, and the SISTRIX Content Team · Industry study · SISTRIX · October 2026</em><br />\nIts same-site pairing and correction for multiple testing make it more careful than most GEO studies. It sells SEO\nsoftware, covers only German-speaking markets, and reports correlations.</p>\n<p><a href=\"https://arxiv.org/abs/2311.09735\" rel=\"nofollow ugc noopener\">GEO: Generative Engine Optimization</a><br />\n <em>Pranjal Aggarwal, Vishvak Murahari, and colleagues · Conference paper · KDD 2024 · version 3, June 2024</em><br />\nThe academic paper that named the field and ran controlled rewrites of pages. It shows content changes can move\nvisibility, but its test engine is now dated, and its authors&#39; &quot;up to 40%&quot; should not be read as a current forecast.</p>\n<p><a href=\"https://ahrefs.com/blog/ai-overview-citations-top-10\" rel=\"nofollow ugc noopener\">Update: 38% of AI Overview Citations Pull From The Top 10</a>\nand <a href=\"https://ahrefs.com/blog/search-rankings-ai-citations\" rel=\"nofollow ugc noopener\">76% of AI Overview Citations Pull From the Top 10</a><br />\n <em>Louise Linehan with Xibeijia Guan · Industry studies · Ahrefs · March 2026 and July 2025</em><br />\nAhrefs sells SEO tools and tracks AI answers through its Brand Radar product, here matching about 4 million cited URLs\nagainst rankings in the later study. The two studies aren&#39;t like-for-like. The 38% counts results in every part of the\npage; for ordinary links alone it is 37.1%.</p>\n<p><a href=\"https://ahrefs.com/blog/ai-chatbot-traffic/\" rel=\"nofollow ugc noopener\">AI Chatbot Traffic: What It Is, and How to Get More</a><br />\n <em>Mateusz Makosiewicz · Company blog · Ahrefs · May 2026</em><br />\nThe source for the 0.28% figure, from analytics on 74,752 sites, and for the signup figure from Ahrefs&#39; own site. It gives\nconflicting percentages for Google&#39;s share in different places, so we compare visitor counts instead.</p>\n<p><a href=\"https://seranking.com/blog/llms-txt/\" rel=\"nofollow ugc noopener\">LLMs.txt: Why Brands Rely On It and Why It Doesn&#39;t Work</a><br />\n <em>Yulia Deda · Industry study · SE Ranking · page dated October 2026</em><br />\nA large correlation study of nearly 300,000 domains. It doesn&#39;t name the AI platforms it sampled, so we use it only\nalongside SISTRIX and Google&#39;s statement.</p>","headings":[{"level":1,"text":"Google's AI Overviews cut clicks to websites by 40%","id":"google-s-ai-overviews-cut-clicks-to-websites-by-40"},{"level":2,"text":"Google answers first","id":"google-answers-first"},{"level":2,"text":"What happens to the clicks","id":"what-happens-to-the-clicks"},{"level":2,"text":"Who the answers cite","id":"who-the-answers-cite"},{"level":2,"text":"GEO looks mostly like SEO","id":"geo-looks-mostly-like-seo"},{"level":3,"text":"Which sites: link profiles correlate with citations","id":"which-sites-link-profiles-correlate-with-citations"},{"level":3,"text":"Which page: cited pages often don't rank for the query","id":"which-page-cited-pages-often-don-t-rank-for-the-query"},{"level":2,"text":"AI sends little traffic, but it is growing and may convert well","id":"ai-sends-little-traffic-but-it-is-growing-and-may-convert-well"},{"level":2,"text":"What to do about it","id":"what-to-do-about-it"},{"level":2,"text":"Stop guessing what makes money","id":"stop-guessing-what-makes-money"},{"level":2,"text":"Sources","id":"sources"}]}}