Quick answer: Across 54 Google searches and 48 AI Overviews, we matched every page Google’s AI Overview cited against Google’s own organic top 10 for the same query. About two-thirds of the citations, 68 percent, went to pages that did not rank in the organic top 10 at all. Even when we counted a whole domain as ranking if any of its pages showed up, 60 percent of citations still came from sites off page one. When a cited page did rank, it usually ranked high, at a median position of 3. And ranking bought no guarantee: only about a third of top-10 results earned a citation. Getting cited by an AI Overview is a separate contest from ranking, and you can win it without a page-one position.
Most advice about AI search quietly assumes that ranking and citation are the same game. Rank well, the thinking goes, and the AI Overview will pull from your page. So we tested it directly. For every query, we lined up the sources Google’s AI Overview cited against the organic results Google served for that same search, and measured how much they actually overlap.
The overlap is far smaller than the assumption requires. The pages an AI Overview quotes are mostly not the pages ranking beneath it, which means a whole tier of sites that cannot crack the organic top 10 can still get cited in the answer box above it.
What we found
We ran 54 Google searches and an AI Overview appeared on 48 of them, an 89 percent trigger rate. We recorded every source each AI Overview cited and the organic top 10 for the same query, then measured the overlap across 376 individual citations.
- 68 percent of AI Overview citations did not rank in the organic top 10. Only 32 percent of cited pages also appeared as an organic result for the same query. The answer box reaches well outside the page-one list beneath it.
- 60 percent came from domains that were nowhere on page one. Even under the loosest possible test, counting a domain as ranking if any of its URLs appeared in the top 10, most citations still came from sites Google did not rank for that search.
- When a cited page did rank, it ranked high. Among the citations that also appeared organically, the median organic position was 3, and 56 percent sat in the organic top 3. Position one through three is where ranking and citation actually meet.
- Ranking in the top 10 did not earn a citation. Only about 35 percent of the organic results we recorded were cited by the AI Overview above them. Two in three page-one pages were passed over.
- Pricing queries rarely triggered an AI Overview at all. Four of the six searches with no AI Overview were plain pricing lookups, such as Ahrefs and Mailchimp pricing. When the searcher wants a number from the vendor, Google tends to step back.
- About 15 percent of citations went to places you cannot control. Reddit, YouTube, Medium, and LinkedIn accounted for roughly one in seven citations, surfaces where a site owner cannot publish a canonical version of their own answer.
Why doesn’t ranking predict citation?
Because an AI Overview is answering a slightly different question than the ranking system is. The organic list ranks whole pages by their overall fit for a query. The overview assembles a specific answer and reaches for the cleanest extractable passage on each sub-point, wherever it happens to live. A page can be the best single result overall and still lose the extraction contest to three other pages that each nail one part of the answer more cleanly. That is how you end up with citations sitting outside the top 10 while two-thirds of the top 10 goes unquoted.
This lines up with the rest of our data. In our schema markup study, the structured-data signal everyone optimises for did not separate cited pages from uncited ones. In our freshness study, cited pages were barely newer than the ones Google ignored. The levers people assume control AI citations keep turning out to be weaker than the quality and structure of the answer itself.
Does this mean rankings no longer matter?
No, and the top-3 finding is why. When a cited page did also rank, it clustered at the very top of the results, so a genuine position-one through position-three placement still carries real citation value. What the data undercuts is the softer belief that being somewhere on page one, at position six or eight, puts you in line to be quoted. It mostly does not. Ranking helps most when it is dominant, and it does surprisingly little from the middle of the first page.
How do you get cited without ranking?
You make individual passages easy to lift. The pages that get pulled into an overview from outside the top 10 tend to answer the exact sub-question in a self-contained way: a direct definition, a clean list, a specific number with its context, placed near the top of the section rather than buried under preamble. That is the practical opening here. A site that cannot outrank Semrush or a decade-old authority can still write the single clearest paragraph on a narrow sub-topic and earn the citation for it.
If you want to see how a given post scores on the signals that actually predict citation, our AI Citation Grader weighs answer-first structure and extractable formatting alongside six other factors, and points you to the specific fixes worth making.
Where ranking and citation still overlap
The disconnect is not total, and it would be a mistake to read it that way. Roughly a third of citations did come from ranking pages, and those pages skewed to the top of the results. The honest picture is two overlapping circles rather than two separate worlds: a dominant organic position is one reliable path to citation, and clean, extractable answers on narrow sub-topics are a second path open to sites that will never hold that position. Most sites have more room to move on the second path than the first.
How we ran the study
We ran 54 Google searches in a clean, US-localised, logged-in Chrome session, spread across informational, commercial “best of”, transactional, and news-style queries in the blogging and SEO space. Forty of these searches were drawn from our earlier schema and freshness datasets, collected the same way; 14 were harvested fresh for this study with a full organic top 10 captured for each. For every search we isolated the AI Overview block, recorded each external source it cited, and recorded the organic top 10 for the same query. We then matched citations to organic results two ways: by exact URL and by domain. A cited page counted as ranking only if its URL, or at minimum its domain, appeared in the organic results we captured. Six searches returned no AI Overview and were kept only in the trigger-rate figure. This is a correlational snapshot of one moment in AI Overview behaviour; AI Overviews are personalised and shift often, so treat the exact percentages as directional and the pattern as the durable finding.
Cite this data
Blogging Titan, The Citation-Rank Disconnect: Do AI Overviews Cite the Pages That Rank? (2026). Overlap between AI Overview citations and the organic top 10 across 48 Google AI Overviews.
| Metric | Value |
|---|---|
| AI Overview trigger rate | 89 percent (48 of 54 queries) |
| Citations not ranking in organic top 10 (by URL) | 68 percent |
| Citations from domains off page one | 60 percent |
| Median organic position of citations that did rank | 3 |
| Organic top-10 results that earned a citation | 35 percent |
Sample: 376 citations across 48 Google AI Overviews, 54 queries total, July 2026. Data by Blogging Titan, released under CC BY 4.0. BibTeX key: bloggingtitan2026citationrank.
Frequently asked questions
Do Google AI Overviews cite the pages that rank in the top 10?
Mostly not. In our study of 48 AI Overviews and 376 citations, 68 percent of cited pages did not appear in the organic top 10 for the same query, and 60 percent came from domains that were off page one entirely. Ranking and citation overlap only partially.
Can you get cited by an AI Overview without ranking on page one?
Yes, and it is common. Two-thirds of the citations we measured came from pages that were not ranking in the top 10. AI Overviews reach outside the organic list for clean, extractable answers, which gives lower-authority sites a genuine opening.
If ranking does not guarantee citation, what does?
Extractability. The pages pulled in from outside the top 10 tend to answer a specific sub-question in a self-contained passage, such as a direct definition, a short list, or a number with its context, placed high in the section. That structure predicts citation better than raw ranking position.
Does ranking position matter at all for AI citations?
It matters most at the very top. When a cited page did also rank, its median organic position was 3 and most sat in the top 3. A dominant position still carries citation value; a mid-page-one position at rank six or eight does much less.
Which searches don’t trigger an AI Overview?
Plain pricing lookups were the clearest gap in our data. Four of the six queries with no AI Overview were pricing searches such as Ahrefs and Mailchimp pricing, where Google tends to defer to the vendor’s own page and skips the assembled answer.
Want this done for you?
We audit your pages for the extractable-answer signals that earn AI Overview citations, even on queries you cannot rank for, and rewrite the sections that are close. See how our GEO citation audits and done-for-you service work.
View our servicesRelated reading
- Schema markup won’t win you AI citations: a 138-page study
- AI Overviews don’t reward fresh content, they punish stale content
- How to optimise your blog for Google AI Overviews
- Score your own post with the AI Citation Grader
- AI Overview and GEO statistics (2026): every data point, free to cite
Study published July 2026. Based on first-party analysis of live Google AI Overviews; figures are directional and reflect AI Overview behaviour at time of collection.