Only 3% of the pages ChatGPT cites rank in Google’s top 10
We typed the same 49 buying searches into Google and ChatGPT and compared the pages. ChatGPT links to brands’ product pages; Google ranks blog posts, review sites and forum threads. Google’s AI Overviews sit in between.
What we found
of the pages ChatGPT cited ranked in Google’s top 10 for the same search
of ChatGPT’s citations were brands’ product, pricing or home pages, against 9% of Google’s top 10
of the pages Google’s AI Overviews cited ranked in the top 10: closer to SEO, still under half
On this page9 sections
The question behind GEO vs SEO
Every SEO team now gets the same question from clients: if we rank on Google, do we show up in ChatGPT? If ChatGPT read the pages Google ranks, the answer would be yes, and GEO would be SEO under a new name.
Ahrefs looked at this across 15,000 long prompts and found that about 8% of the links ChatGPT cited ranked in Google’s top 10 for the prompt [1]. Long prompts are not what most buyers type into Google, though. So we asked the narrower question that matters to a brand: for the same short buying search, typed word for word into both, how many of the pages ChatGPT cites rank on Google?
What we compared
50 searches someone makes before buying: “best payroll software for a US small business”, “Shopify vs WooCommerce for a new store”, “best robot vacuum for pet hair”. Each went to Google and to ChatGPT on November 6, 2026, from the United States, in English. From Google we kept the organic results and the AI Overview; from ChatGPT, the answer and the pages it cited. One answer cited no pages at all, which leaves 49 searches and 320 ChatGPT citations, 6.5 per answer.
How we measuredShowHide
Searches: 50. 30 about software (the 20 questions of our first study plus 10 more), 10 about products and 10 about services such as banking, insurance and subscriptions. 11 of them compare two brands (“X vs Y”).
Collection: one pass through DataForSEO’s standard queue [2], the same provider and queue our product uses: Google’s organic results two pages deep, with the AI Overview, and ChatGPT’s answer from its web interface (model “gpt-6”).
Same page: addresses compared without http or https, “www.”, tracking parameters or a trailing slash. Google’s top 10 is its first 10 different organic pages: the second page of results sometimes repeats the first. Same site: subdomains count as their site (blog.hubspot.com is hubspot.com).
Kinds of site: the lists our product uses for review sites, forums, video, media and reference sites. The 707 sites not on them were tagged by a language model as a brand’s own site or a publisher; in a hand check of 100 of them, 5 were in the wrong group. Brands ChatGPT named count as brand sites. A brand’s page counts as a blog post when its address says so (/blog/, “best-”, “-vs-”).
Ranges: 95% Wilson intervals [3], as in all our reports (how we measure). Citations from one answer are not independent, so the ranges are, if anything, too narrow.
Limits: one day, one country, one pass, short searches. ChatGPT’s answers change from one run to the next (our first study), so a second pass would cite partly different pages; the rates describe this one. The whole study cost $0.18 in data and model calls. Every page behind the figures is in the CSV.
Three pages in a hundred
Of the 320 pages ChatGPT cited, 11 ranked in Google’s top 10 for the same search: 3% (2–6%). Widening to Google’s first two pages adds almost nothing (4%). In 41 of the 49 answers, not one cited page was in the top 10.
“Best email marketing platform for an ecommerce store”: Google’s top 10 and the pages ChatGPT cited
Google top 10
ChatGPT cites
- 1sender.net/blog/email-marketing-services
- 2mailersend.com/blog/email-marketing-tools
- 3sequenzy.com/blog/best-email-marketing-platforms-ecommerce
- 4omnisend.com/blog/ecommerce-email-marketing
- 5digitorm.com/blog/best-ecommerce-email-marketing-platforms
- 6moosend.com/blog/best-email-marketing-software-for-ecommerce
- 7apps.shopify.com/categories/marketing-and-conversion-marketing-email-marketing
- 8flowium.com/blog/email-marketing-software
- 9youtube.com/shorts/hak17ejtcc8
- 10theretailexec.com/tools/ecommerce-email-marketing-software
- klaviyo.com
- omnisend.com
- mailchimp.com
- drip.com
- brevo.com
- technologyadvice.com/marketing/ecommerce-email-marketing-software
- sprout24.com/email-marketing-for-ecommerce
- techradar.com/best/best-email-marketing-software
0 of 8 pages ChatGPT cited also rank in Google's top 10 for the same question.
The example is typical: the two lists share no page. They do share a brand. Google ranks a blog post on omnisend.com, and ChatGPT links to the same site’s home page. Counted by site rather than by page, the overlap grows to 14% (10–19%), still a small minority.
ChatGPT’s citations that rank on Google for the same search
- Page, top 103% (2–6%)
- Page, pages 1–24% (2–6%)
- Site, top 1014% (10–19%)
- Site, pages 1–220% (16–25%)
The rate was low in every group, though the samples are small: 1% of citations for software searches, 6% for products and 14% for services. “X vs Y” searches overlapped a little more (7%) than “best X” ones (3%).
Same brands, different pages
The pages differ because the two engines answer in different ways. Google ranks pages that compare products. ChatGPT names products and, next to each one, links to the maker’s own site. 51% (46–57%) of its citations were brands’ product, pricing or home pages, against 9% of Google’s top 10. On Google, brands rank with their blog posts and guides (32% of the top 10); ChatGPT cited those 10% of the time.
What each one points to, by kind of site
ChatGPT’s citations
- Brand product pages51%
- Publishers & blogs27%
- Brand blog posts10%
- Stores6%
- Review sites5%
- News & media2%
- Reference<1%
- Forums0%
- Video0%
Google’s top 10
- Brand product pages9%
- Publishers & blogs30%
- Brand blog posts32%
- Stores1%
- Review sites3%
- News & media4%
- Reference1%
- Forums17%
- Video4%
Two kinds of site Google ranks never showed up in ChatGPT’s citations: forums (17% of Google’s top 10, mostly Reddit, Quora and Facebook) and video (4%). Publishers, the review blogs and “best X” lists, are the one kind both lean on about as much: 30% of Google’s top 10 and 27% of ChatGPT’s citations. They are rarely the same publishers, though: 5 of the 85 publisher pages ChatGPT cited were in the top 10.
A #1 ranking barely helps
If ranking fed ChatGPT, the top result would be cited most. It was, but not by much. ChatGPT cited the #1 page itself in 2 of 49 searches. Counting any page of the same site, the #1 result’s site was cited in 12% of answers, against 6% for sites ranked 11th to 20th.
ChatGPT cited the same site, by its Google rank for the search
- Ranked #112% (6–24%)
- Ranked #2–38% (4–15%)
- Ranked #4–108% (5–11%)
- Ranked #11–206% (4–9%)
ChatGPT searches a different web
ChatGPT doesn’t borrow Google’s results. It runs its own searches, and a site only appears in them if it lets OpenAI’s search crawler, OAI-SearchBot, in [4]. The data shows those searches: for these 49 questions they returned 642 pages, and 3% of them were in Google’s top 10 for the same words.
That splits ChatGPT’s citations in two. About half (50%) were pages its searches returned: lists, reviews and guides it read. The other half were links it added next to the products it named, almost all brands’ own pages and store listings that none of its searches returned. Even among the pages it found by searching, 7% (4–12%) ranked in Google’s top 10.
Google’s AI Overviews: closer, not the same
All 49 searches showed an AI Overview. They are built on Google Search and can run several related searches for one question [5], so they should sit closer to Google’s ranking, and they do: 38% (34–43%) of the 381 pages they cited ranked in the top 10. Ahrefs found 76% across a million overviews in 2025, counting each overview’s first three citations [6]. Counted the same way, ours was 47%.
Pages each AI cited that rank in Google’s top 10 for the same search
- AI Overviews38% (34–43%)
- ChatGPT3% (2–6%)
The overviews also lean on video: 23% of their citations were videos, mostly YouTube, a kind of page ChatGPT never cited. And the two AIs rarely agree with each other: 5% of ChatGPT’s citations were also cited by the AI Overview for the same search.
What to do with it
- Don’t report Google rankings as AI visibility. For ChatGPT they predict almost nothing. Measure the answers themselves.
- Treat product and pricing pages as GEO pages. Half of ChatGPT’s links go there. Say plainly what the product does, who it’s for and what it costs.
- Get onto the lists ChatGPT reads. The review blogs and “best X” pages it cites are the way into an answer that names competitors. Our guide to GEO vs SEO turns this into a plan.
- Keep doing SEO for Google’s own AI. Ranking still feeds the AI Overview, and video earns a place there that it doesn’t in ChatGPT.
Sources
- 1.Only 12% of AI cited URLs rank in Google’s top 10 for the original prompt · Ahrefs · checked Nov 6, 2026
- 2.Google Organic SERP API pricing · DataForSEO · checked Nov 6, 2026
- 3.Binomial proportion confidence interval (Wilson score) · Wikipedia · checked Nov 6, 2026
- 4.Overview of OpenAI crawlers · OpenAI · checked Nov 6, 2026
- 5.AI features and your website · Google Search Central · checked Nov 6, 2026
- 6.76% of AI Overview citations pull from the top 10 · Ahrefs · checked Nov 6, 2026
Cite this study
Menzio Research. (2026, November 6). Only 3% of the pages ChatGPT cites rank in Google’s top 10. Menzio. https://getmenzio.com/research/chatgpt-citations-vs-google-rankings
Which pages do AI answers cite in your category?
Run three buyer questions on ChatGPT, Gemini and Google AI Overviews: who gets named, and the pages the answers lean on.
Free · no signup · ChatGPT, Gemini and Google AI Overviews
Keep reading
Changes to this page
- November 6, 2026: First published.