--- title: "Does being cited by ChatGPT boost our Google rankings?" description: "No evidence says so. The one field test found ChatGPT gains left Google clicks flat to falling, and ChatGPT mostly cites pages Google does not rank." canonical: "https://underneath.agency/resources/do-chatgpt-citations-boost-google-rankings" published: 2026-10-07 updated: 2026-10-08 publisher: "Underneath (https://underneath.agency/agent)" entity: "https://underneath.agency/.well-known/entity.json" --- Guide · AI search # Does being cited by ChatGPT boost our Google rankings? There is no evidence that it does. The only published field test found that pages winning more ChatGPT traffic saw Google clicks stay flat or fall, not rise. That is one website over a few months, so it is an absence of evidence rather than proof of no effect, but it means a “ChatGPT lifts Google” assumption does not belong in a forecast. ## The short version 1. In a field test on one website, pages that gained ChatGPT referrals saw Google clicks fall about 25%, against about 20% site-wide; the authors found “no support” for a lift to Google ([Watanabe and Nakayashiki](https://arxiv.org/abs/2606.04362)). 2. ChatGPT and Google largely draw on different pages: only 8.3% of the pages ChatGPT cited ranked in Google’s top 10 for the question ([our AI citations and Google rankings study](https://underneath.agency/research/ai-citations-google-rankings-study)). 3. Where a link exists, it runs from Google to AI: the first organic result was cited in 49.5% of AI Overviews, the ninth in 15.5% ([our AI Overview citations study](https://underneath.agency/research/ai-overview-citations-study)). 4. Being cited by AI does not guarantee visits either: English Wikipedia, among the sources AI summaries cite most, lost about 5.45% of its search traffic after AI Overviews launched ([Khosravi and Yoganarasimhan](https://arxiv.org/abs/2602.18455)). ## Is there evidence that ChatGPT citations lift Google rankings? No: the one direct test found Google traffic flat to declining on pages that gained ChatGPT traffic. [Watanabe and Nakayashiki](https://arxiv.org/abs/2606.04362), who work at Glasp, tracked their own site from mid-2025 to May 2026 using its analytics and Google Search Console data. They optimized one large section for ChatGPT in January 2026 and left the rest alone. The optimized pages gained. Their ChatGPT referrals grew 6.1 times between January and May 2026, and after [allowing for ChatGPT’s general growth](https://underneath.agency/resources/chatgpt-referral-growth-and-geo) the authors estimate a real lift of about 1.8 to 2.3 times. Google is where the flywheel should have shown up next. It did not. Google clicks to the optimized section fell about 25% from the second half of 2025, close to the roughly 20% fall across the whole site. The authors write that they “find no support for the stronger ‘flywheel’ claim,” and that Google traffic to those pages was “flat-to-declining, not rising.” ## How strong is that evidence? It is a single case, so treat it as a caution, not a verdict. The study covers one domain, and ChatGPT supplied nearly all its AI traffic: other AI engines contributed under 1% of AI-referred visits. The window ran about five months after the changes, and a Google effect could take longer to appear. There are other limits the authors flag. The changes were a bundle, and they [protected pages that already earned Google clicks](https://underneath.agency/resources/chatgpt-optimization-google-rankings) from being rewritten. The authors are also reporting on their own company’s site. Still, this is the only published test of the question, and it points the wrong way for the flywheel. Absent better evidence, the safe planning assumption is that ChatGPT and Google results must each be earned on their own terms. ## Do ChatGPT and Google draw on the same pages? Mostly not, which makes a strong link between them less likely. In [our study of 80 US buyer questions](https://underneath.agency/research/ai-citations-google-rankings-study), only 8.3% of the pages ChatGPT cited ranked in Google’s top 10 for the question. 68.5% were not in Google’s top 100 for the question or for any of the searches ChatGPT ran. ChatGPT also does its own searching. It ran a mean of 3.7 searches per answer, none of them a word-for-word copy of the user’s question ([our hidden searches study](https://underneath.agency/research/ai-hidden-searches-study)). It is choosing from a different pool, by different rules. Google’s rankings are themselves a moving target. Between 26 and 28 September, Google’s top 10 for the same questions changed substantially, and ChatGPT’s match fell from 8.3% to 5.2% against the later results. Rankings shift for many reasons that have nothing to do with AI citations. ## Which way does the link run, if there is one? The evidence points from Google rankings to AI citations, at least inside Google’s own AI. In [our study of 4,051 AI Overview citations](https://underneath.agency/research/ai-overview-citations-study), 28.7% were page-one results for the same search. The first organic result was cited in 49.5% of AI Overviews, the ninth in 15.5%. A vendor-authored study takes the same view of cause and effect. In the workflow proposed by [Vishwakarma and colleagues](https://arxiv.org/abs/2605.25517) at Sprinklr, a brand missing from AI citations has a problem with being found, and the stated action is to “improve SEO.” In their framing, search work feeds AI visibility, not the reverse. So the dependable direction is the familiar one. Ranking well on Google helps Google’s AI find you. Nothing published shows ChatGPT returning the favor. ## Does being cited by AI at least bring visitors back? Not reliably, and that weakens the flywheel logic further. [Khosravi and Yoganarasimhan](https://arxiv.org/abs/2602.18455) at the University of Washington studied English Wikipedia after Google made AI Overviews the US default in May 2024. Wikipedia is among the sources AI summaries cite most often. Its monthly search traffic still fell 5.45% relative to the German edition and 4.82% relative to the French one. The authors conclude that any traffic from those citations “did not offset the average decline.” A citation is visibility, but it is not the same as a visit, and visits are what a flywheel would run on. That loss followed a change in [how Google lays out its results](https://underneath.agency/resources/google-search-design-changes-traffic-risk), which publishers cannot control. ## What should you do about it? Plan ChatGPT and Google as separate channels, and do not count on one to lift the other. In practice: 1. Remove any “AI citations will lift SEO” line from forecasts and business cases until there is evidence for it. 2. Keep investing in Google rankings directly. They still drive citations in Google’s own AI Overviews. 3. Set separate targets for ChatGPT visibility, based on the pages and searches it actually uses. 4. If you test the flywheel yourself, keep a comparison group of similar pages you do not optimize, and watch Google clicks for longer than a few months. 5. Report the two channels side by side so a gain in one cannot mask a loss in the other. For help setting up that kind of measurement, see [our generative engine optimization service](https://underneath.agency/services/generative-engine-optimization). ## What does the research not tell us yet? The question has barely been studied, so the honest answer is “not shown,” not “impossible.” - Only one field test addresses it directly, on one website, one AI engine and about five months of data after the changes. - Nobody has measured whether brand searches on Google rise after a brand appears often in ChatGPT answers. - Our studies compare which pages each system cites or ranks; they do not follow pages over time. - Longer-run effects, over a year or more, are untested. ## Frequently asked questions ### Do AI citations help SEO? Not on current evidence. The one field test found Google clicks to pages that gained ChatGPT traffic were flat to declining, about 25% down against 20% site-wide. ### Is there an AEO to SEO flywheel? None has been shown. The authors of the only direct test write that they “find no support” for it in their study window. ### Does ranking on Google help you get cited by AI? For Google’s own AI, yes. In our study, the top organic result was cited in 49.5% of AI Overviews; for ChatGPT the overlap with Google’s top 10 was only 8.3%. ### Why would ChatGPT citations not carry over to Google? Partly because the two systems pick different pages. In our study, 68.5% of the pages ChatGPT cited were not in Google’s top 100 for the question or its own searches. ### Can ChatGPT traffic make up for lost Google traffic? The published evidence cannot tell you. On the one site studied, ChatGPT referrals grew while Google clicks fell, but the authors withheld absolute visit counts, so the two channels cannot be compared in size. Measure both on your own site before assuming one replaces the other. ## Sources - Watanabe and Nakayashiki (2026), [Disentangling Answer Engine Optimization from Platform Growth: A Log-Based Natural Experiment on ChatGPT Referral Traffic](https://arxiv.org/abs/2606.04362), arXiv:2606.04362. - Khosravi and Yoganarasimhan (2026), [Impact of AI Search Summaries on Website Traffic: Evidence from Google AI Overviews and Wikipedia](https://arxiv.org/abs/2602.18455), arXiv:2602.18455. - Vishwakarma, Kumar and Jamidar (2026), [What Gets Cited: Competitive GEO in AI Answer Engines](https://arxiv.org/abs/2605.25517), arXiv:2605.25517. - Underneath (2026), [Do ChatGPT, Gemini, Perplexity and Claude cite pages that rank?](https://underneath.agency/research/ai-citations-google-rankings-study) - Underneath (2026), [The hidden searches AI assistants run before they answer](https://underneath.agency/research/ai-hidden-searches-study) - Underneath (2026), [Do AI Overviews cite the pages that rank? 4,051 citations](https://underneath.agency/research/ai-overview-citations-study) --- This is the Markdown twin of https://underneath.agency/resources/do-chatgpt-citations-boost-google-rankings. The HTML page is canonical. Publisher: Underneath, https://underneath.agency/agent. Site index: https://underneath.agency/llms.txt.