For years the SEO deal was simple: rank on page one and you get seen. That deal is breaking. In March 2026, Ahrefs analyzed 4 million AI Overview citations across 863,000 SERPs, using its Brand Radar tool. It found that 37.9% of URLs cited in AI Overviews also appeared in the top 10 results for the same query. In July 2025 the figure was 76.1%. So in eight months, the share of AI citations coming from page one roughly halved. The main reason has a name, query fan-out, and Google itself confirms it. If you publish anything online, it's worth understanding.
What query fan-out actually is
When you type a question into Google's AI Mode, it doesn't run your exact words as a single search. In Google's own description, AI Mode uses its query fan-out technique, "breaking down your question into subtopics and issuing a multitude of queries simultaneously on your behalf." One Google engineer put it more plainly: AI Mode is basically doing a dozen searches for you in the time it takes to do one.
The technique isn't limited to AI Mode. Google's developer documentation says both AI Overviews and AI Mode may use a "query fan-out" technique, issuing multiple related searches across subtopics and data sources, to develop a response. At the far end, Deep Search uses the same query fan-out technique but taken to the next level.
Here's an example. Say someone asks "best project management tool for a five-person agency." Behind the scenes, the system might also search for pricing comparisons, integrations with common apps, reviews from small teams, and alternatives to the market leader. It then writes one answer from all of those results and cites pages from each branch. Your page can get cited because it answered one of those side questions well, even if it never ranked for the question the person typed.
Why the top-10 overlap collapsed
Fan-out explains the Ahrefs numbers. If the AI is pulling from a dozen searches, the ten blue links for the original query are only one source out of many. Ahrefs found that 62.1% of citations came from pages ranking at position 11 or worse, including 31.0% from pages not in the top 100 at all. The company's own reading is that some AI Overview citations come from pages that surface in SERP features or sub-query results rather than from the regular rankings for the exact query.
YouTube shows up a lot in that outside-page-one group. Among the AI Overview cited pages that didn't rank in Google's top 100 results for the same keyword, 18.2% were YouTube URLs.
The two Google surfaces don't even agree with each other. Ahrefs measured a 13.7% URL overlap across 540,000 query pairs, and found the answers themselves about 86% semantically similar. In other words, two Google surfaces reach nearly the same answer while citing almost entirely different pages. (For what a citation is actually worth in clicks once you get one, see our breakdown of the AI Overview citation click gap.)
Why you can't just target the hidden queries
The obvious reaction is to find the fan-out queries and write a page for each one. Two problems make that harder than it sounds.
First, you can't see them. Google does not publish the literal fan-out queries. Second, the queries are often too niche to show up in keyword research. AirOps analyzed 15,000 prompts and found that 32.9% of cited pages that ranked in Google's top 20 appeared only for a fan-out query, not for the original prompt, and that 95% of fan-out queries had zero traditional monthly search volume. Your keyword tool will show those terms as worthless even when they're the ones earning citations.
They also change from one run to the next. In Surfer's study, 10,000 keywords produced 33,000 fan-out queries, and only about 27% repeated across runs of the same prompt, while 66% appeared once. If you build a page for one exact sub-query, you're targeting something that may never come up again. (This is the same randomness behind why AI answers change when you ask twice.)
What actually correlates with getting cited
The useful signal is breadth, not one precise target. In Surfer's analysis of 173,902 URLs, pages that ranked for the main query and at least one fan-out query were 161% more likely to be cited in AI Overviews. The study also found a strong Spearman correlation of 0.77 between the number of fan-out queries a page ranks for and its likelihood of being cited.
Two caveats. Surfer sells SEO software, and as one analysis points out, the finding is a correlation, not proof that covering sub-queries causes citations. Also, regular SEO still matters. Google's docs say fan-out retrieval runs on the same core ranking systems as classic Search. Pages that rank for many related questions tend to be thorough pages on sites Google already trusts, and that's the advantage.
It's also mainly a Google behavior right now. Ahrefs' October 2026 study of ChatGPT found that only 14.4% of responses contained fan-out queries, averaging about 1.6 each. If you're focused on Google's AI features, fan-out matters a lot. For ChatGPT, it matters less.
What to do with this
You don't need a new tool or a new acronym to act on any of this. Here's what follows from the data:
- Answer the side questions on the page itself. For any important topic, list the 5-10 things a careful buyer would also want to know, like cost, alternatives, who it's not for, setup time and common problems. Then answer each one in its own clearly labeled section.
- Write passages that make sense on their own. The AI may cite one section for one sub-query. A paragraph that only works if the reader saw the previous three won't get lifted.
- Build clusters, not lone pages. A main guide plus linked pages on comparisons, pricing and use cases covers more possible sub-queries than one long page aimed at a single keyword.
- Don't ignore "zero-volume" questions. If customers really ask it, the AI may search for it, even if your keyword tool shows no volume.
- Consider video for how-to topics. YouTube's large share of citations from outside page one suggests Google treats video as its own source pool.
- Measure more than rank. If you only track position for one head term, you'll miss most of where AI visibility comes from.
FAQ
Is query fan-out the same as keyword research?
No. Keyword research finds what people type. Fan-out is what Google's AI searches for on their behalf, and most of those sub-queries show no search volume and change between runs. Think of it as covering the questions around a topic, not as a list of keywords to hit.
Does ranking in the top 10 still matter for AI Overviews?
Yes, but it's no longer enough. About 38% of AI Overview citations still come from top-10 pages, so classic rankings remain the biggest single source. The other 62% come from further down or from sub-query results, which is where covering related questions helps.
Can I see which fan-out queries Google ran for a search?
Not directly. Google doesn't publish them. You can make educated guesses by looking at the spread of sources an AI Mode answer cites. Each distinct type of source (a review site, a pricing page, a forum thread) usually points to a separate sub-search. Asking an AI assistant to break your topic into the sub-questions a buyer would ask also gives you a workable checklist.