Ranking first no longer guarantees an AI citation because Google stopped pulling AI Overview sources primarily from the page you rank on. It now splits your query into several sub-queries, searches each one separately, and cites the pages that appear most consistently across all of them. You can win the original query and still lose every sub-query.
If your rankings are holding but your traffic is falling, this is one of the reasons. The relationship between where you rank and whether an AI answer names you has weakened dramatically in about a year, and the mechanism behind it is now publicly documented rather than speculative.
This post covers what the data actually shows, why credible studies disagree on the exact number, what caused the shift, and what it changes about how you should approach SEO. It is the detailed version of one section from our complete guide to AI search visibility.
What the data actually shows
Two Ahrefs studies bracket this change, and the methodology matters as much as the headline.
July 2025. Ahrefs analysed 1.9 million citations across 1 million AI Overviews, looking at the three most visible citations in each response. It found that 76.1% of cited pages also ranked in Google’s top 10 for the same query. Only 14.4% did not rank in the top 100. At that point, the working assumption across the industry was reasonable: rank well, and AI visibility largely follows.
March 2026. Ahrefs re-ran the study at more than double the scale, analysing 863,000 keyword SERPs and 4 million AI Overview URLs. They ran it two ways.
Ahrefs noted that its parsing methodology improved between the two studies, letting it detect more of the citations appearing in AI Overviews. That means part of the apparent drop may reflect better measurement rather than pure behaviour change. The direction of travel is not in doubt. The exact magnitude carries a real asterisk, and anyone quoting “76 to 38” without that context is overselling the precision.
Why it broke: query fan-out
This is the part most coverage skips, and it is the part that actually explains everything.
Google has publicly confirmed in its Search Central documentation that AI Overviews and AI Mode use a technique called query fan-out. When a user searches and an AI response is triggered, the system does not simply summarise the page you are looking at. It splits the original query into multiple related sub-queries, runs searches for each of them, and then assembles an answer from the pages that appear most consistently across those sub-query results.
This reframes the whole problem. You are no longer competing for one keyword. You are competing across a cluster of related questions you cannot see, and the winner is whoever shows up most consistently across all of them.
Ahrefs’ own interpretation aligns with this. They noted that AI Overviews appear to be relying less on direct search results and more on the sources surfacing in fan-out query results, and pointed to a specific trigger: as of January 2026, AI Overviews are powered by Gemini 3, having previously relied on earlier Gemini models. A more capable model may be expanding queries more aggressively or more broadly, pulling from related result sets where fewer pages rank for the original query.
Depth of topical coverage now beats a single perfectly optimised page. If you have one strong page on a subject and your competitor has eight covering every angle of it, the competitor appears across more sub-query results and gets cited more often, even if your one page outranks all eight.
The YouTube finding nobody expected
Buried in the same study is a result worth its own section. Among AI Overview cited pages that did not rank in Google’s top 100 for the query, 18.2% were YouTube URLs. Those YouTube URLs accounted for 5.6% of all AI Overview citations across the entire dataset.
Ahrefs reports that YouTube is currently the most cited domain in AI Overviews, and that its citation share has grown 34% over the last six months. A separate Ahrefs study of 75,000 brands found that mentions on YouTube, in video titles, transcripts and descriptions, were the strongest correlating factor with AI Overview visibility of any variable they measured.
Video is not a nice-to-have for AI visibility. It is currently the single strongest correlating signal Ahrefs has identified. If you have never considered YouTube part of your search strategy, this is the data point that should change that. Correlation is not causation, and Ahrefs is careful to frame it as a correlation, but the size of the effect is hard to ignore.
Why credible studies disagree on the number
You will see three different figures quoted for this, and the gap is wide enough that it is worth understanding rather than picking whichever suits.
| Source | Figure | What it measured |
|---|---|---|
| Ahrefs, July 2025 | 76.1% | Top three most visible citations per AI Overview, 1.9 million citations. Focusing on the most prominent citations skews toward strong organic performers. |
| Ahrefs, March 2026 | 37.9% | All citations across 863,000 SERPs, with improved parsing that catches more citations than the earlier study could. |
| BrightEdge, Feb 2026 | ~17% | All cited sources across nine industries. BrightEdge separately reports that around 54% of citations come from pages ranking somewhere in the results, just not the top 10. |
That last row is the key to reconciling them. BrightEdge’s 17% and 54% figures are not contradictory: roughly 17% of citations come from top 10 pages, while around 54% come from pages ranking anywhere in the index. Ahrefs’ March 2026 data shows the same shape, with 37.9% from the top 10 and a further 31.2% from positions 11 to 100, or 69% ranking somewhere.
All three studies agree on the structure of the finding, which is what matters:
- A minority of citations come from top 10 pages. Somewhere between 17% and 38%, depending on methodology.
- A large share comes from pages ranking deeper. Positions 11 to 100 contribute heavily.
- A meaningful share comes from pages that do not rank at all. Around a third, in the Ahrefs data.
What this does not mean
It would be easy to read the above and conclude that rankings no longer matter. That conclusion is wrong, and acting on it would be expensive.
Ranking position remains one of the strongest predictors of citation. AirOps reported in March 2026 that 43.2% of pages ranking first in Google are cited by ChatGPT, roughly 3.5 times the rate of pages ranking outside the top 20. Ahrefs’ earlier study found the median organic position of the top-cited URL in an AI Overview was position 2.
Ranking in the top 10 is neither necessary nor sufficient for an AI citation. It is not necessary, because roughly a third of cited pages do not rank in the top 100. It is not sufficient, because most top 10 pages are not cited. But it is strongly helpful, roughly tripling your odds. Treat rankings as a powerful input, not as the deliverable.
What to do about it
1. Build topical depth, not single pages
Because fan-out rewards consistent presence across related sub-queries, a cluster of eight connected pages on a subject outperforms one comprehensive page. This is the practical argument for hub-and-spoke content structure, and it is now a citation argument rather than just a ranking one.
2. Take YouTube seriously
The strongest correlating factor Ahrefs found with AI Overview visibility was YouTube mentions in titles, transcripts and descriptions. A modest, consistent video presence covering your core topics is disproportionately valuable.
3. Stop measuring only rankings
If you report on positions alone, you will miss both the problem and the progress. Track citations and share of voice across AI engines alongside your rank tracking. They move independently.
4. Keep the SEO fundamentals
Ranking still roughly triples your citation odds, and domain authority remains a strong predictor. SE Ranking found sites with 32,000 or more referring domains were 3.5 times more likely to be cited by ChatGPT than sites with 200 or fewer. This is not the moment to defund SEO.
5. Refresh existing content
Ahrefs’ analysis of 17 million citations found AI-cited content is measurably fresher than typical organic results. Updating strong existing pages is often higher return than publishing new ones.
Are you ranking but not being cited?
It is the most common pattern we find in audits: solid rankings, flat or falling traffic, and competitors named in AI answers instead. We will show you exactly which prompts you appear in, which you do not, and which sources the answers are pulling from.
Get an AI visibility auditFrequently asked questions
Does ranking #1 on Google still matter for AI visibility?
Yes, substantially, but it is no longer decisive. AirOps found 43.2% of pages ranking first in Google are cited by ChatGPT, around 3.5 times the rate of pages ranking outside the top 20. However, Ahrefs found that in March 2026 only 37.9% of AI Overview citations came from top 10 pages, and around 31% came from pages not ranking in the top 100 at all. Ranking is a strong input, not a guarantee.
What is query fan-out?
Query fan-out is a technique Google has confirmed it uses for AI Overviews and AI Mode. When a search triggers an AI response, the system splits the original query into multiple related sub-queries, runs a search for each, and assembles its answer from pages appearing consistently across those sub-query results. This is why a page can rank first for the original query and still not be cited.
Why do different studies report different numbers for citation overlap?
Because they measure different things. Ahrefs’ July 2025 study examined only the three most prominent citations per Overview, which skews toward strong organic performers, producing 76.1%. Its March 2026 study measured all citations with improved parsing, producing 37.9%. BrightEdge measured all cited sources across nine industries and reported roughly 17% from the top 10, while separately noting around 54% come from pages ranking somewhere in the index. All three agree on the shape of the finding.
Why is YouTube cited so often in AI Overviews?
Ahrefs found that among AI Overview citations from pages not ranking in Google’s top 100, 18.2% were YouTube URLs, and that YouTube is currently the most cited domain in AI Overviews. A separate Ahrefs study of 75,000 brands found YouTube mentions in video titles, transcripts and descriptions were the strongest correlating factor with AI Overview visibility of any variable measured. Video appears to be surfacing through fan-out sub-queries where it ranks even when it does not rank for the original query.
Should I stop investing in traditional SEO?
No. Ranking first roughly triples your citation odds versus ranking outside the top 20, and domain authority strongly predicts citation, with SE Ranking finding sites with 32,000 or more referring domains 3.5 times more likely to be cited by ChatGPT. The correct interpretation is that SEO is now a prerequisite rather than the whole job. AI visibility is an additional layer built on top of it.
Sources
- Ahrefs, “Update: 38% of AI Overview Citations Pull From The Top 10”, 2 March 2026. 863,000 keyword SERPs, 4 million AI Overview URLs.
- Ahrefs, original AI Overview citation study, July 2025. 1.9 million citations across 1 million AI Overviews.
- Ahrefs, brand visibility correlation study, 75,000 brands (YouTube mentions as strongest correlating factor).
- Ahrefs, citation freshness study, 17 million citations, July 2025.
- BrightEdge, AI Overview rank overlap analysis, February 2026.
- Google Search Central, documentation on AI features and query fan-out.
- Google, AI Mode and AI Overviews model update announcement (Gemini 3, January 2026).
- AirOps, ChatGPT citation rate by Google ranking position, March 2026.
- SE Ranking, referring domains and ChatGPT citation likelihood.
Muhammad Asjad Khan is a seasoned digital marketer and the Chief Operating Officer at Digital Age, where he leads strategy, innovation, and client success. With years of experience in SEO, content marketing, and performance-driven digital campaigns, Asjad is passionate about helping brands grow their online presence and achieve measurable results. When he’s not optimizing websites or scaling marketing strategies, he’s exploring the latest trends in tech and digital media