We pulled the top 10-ranking pages for 10,000 keywords. 76% had AI Overviews, so we used Gemini to pull their fan-out queries (33,000 in total) and scraped the top 10-ranking pages for those, too.
Result? This pretty graph confirming that, yes, the more fanout queries you rank for, the more likely you are to get cited in AI Overviews:
How much more likely?
Well, the Spearman correlation here is 0.77. (For those of us who aren’t stats nerds, that basically translates to “pretty damn strong correlation”).
But before you obsess over ranking every fan-out query you can find, let’s take a look at a few more findings from our data. I’ll also share my opinion on how you should (and probably shouldn’t) act on this data…
Ranking for the main query is important, but you’re 161% more likely to get cited if you also rank for fanouts
Of the AIO citations that rank in organic search, 51.2% (i.e., most of them) do so for the main query and at least one fanout query, while only 19.6% rank solely for the main query.
Takeaway: If you want to stand the best chance of being cited in AIOs, ranking for both the main query and fanouts appears to be the game. And again, the more fanouts you can rank for, the better.
Given a choice between ranking only for fanouts or the main query, you’re 49% more likely to get cited by ranking for the fanouts
When AIOs cite organic results, they cite pages ranking only for fanout queries (not the main query) 29.2% of the time, and pages ranking only for the main query (no fanouts) 19.6% of the time.
Takeaway: Ranking for fanouts seems more likely to get you cited than ranking for the main query.
Most AIO citations don’t rank in organic search at all
67.82% of AIO citations in our sample didn’t rank in the top 10 at all—neither for the main query nor for any fanout queries.
However, the numbers shift slightly if you focus only on the top 3 citations (you know, the ones visible in Google without clicking “Show all”). Only 45.86% of these don’t rank at all, meaning the majority (54.14%) do.
I think part of this can be explained by the fact that we didn’t pull SERP data beyond the top 10. After all, many pages cited by AIOs probably just rank lower. But it’s also likely that AIOs draw from other sources as well.
Takeaway: Ranking for the main query and fanouts is clearly important, but "traditional SEO" not be the be and end all of earning AIO citations.
What should you do about all this?
First, a quick reality check: correlation ≠ causation. Our data only shows patterns, not that ranking for fanouts always improves your chances of getting cited in AIOs. (This is a limitation of most SEO data studies that you should be aware of).
Plus we showed that AIOs are likely pulling from sources beyond organic “blue link” results—so getting cited probably isn’t just about ranking.
Those caveats aside, you might be tempted to think: “let’s just scrape as many fanouts as possible and try to rank for them all”.
My two cents? This probably isn’t the best idea…
Besides the fact that pulling lots of fanout queries is slow and messy (especially if you want to do it across multiple LLMs), there are two more issues:
- You won’t account for personalization this way because fanouts vary depending on what’s known about the user. For example, if Google knows someone has a family and they search for "best electric car," the fanout might be "electric car safety ratings" for them but "fastest electric cars" for someone else.
- You would need to run queries multiple times because according to our data, only around 27% of fanouts stay consistent across multiple searches for the same prompt. This means that to get clean data, you’d have to cluster the outputs across multiple runs to understand the “core” fanout topics.
I don’t know about you, but I really don’t have the time or energy to even try to account for all of this stuff…
Better idea: just focus on building “topical authority”
The idea here is simple: If you can build a strong content base around important topics, you'll likely always have content answering the questions that AIOs (and other LLMs) want to know—regardless of who’s asking.
For example, let's say you want to be cited and recommended for "best electric car"…
If you see that the current fanout is "electric car safety ratings," then yes, technically you could create content to rank for this query and maybe increase your odds of getting cited. But what if another user gets a different fanout or the fanout differs between runs?
You have to play catch-up. And this game of catchup never stops because you’re always optimizing for a particular moment in time and a particular set of fanouts.
Did you know that you can see fanout queries in ChatGPT with Keyword Surfer, our free Chrome extension?
But if you take a more holistic approach and build “topical authority” around a topic, you have a much stronger chance of getting cited consistently—even when fanouts vary—because your content covers so much ground (i.e., lots of potential fanouts).
Plus, 90% of SEOs agree that “topical authority” is important for SEO…
… and this makes it important for GEO/AEO/LLMO/whatever we’re calling it too because most AIO citations rank in the top 10 for the query itself or its fanouts:
So This is exactly the use case for Topical Map in Surfer:
If you've never used it before, here's how it works:
- You give it a site or topic
- It finds semantically-related keywords (either to the topic you gave it, or the keywords you already rank for)
- It clusters them into topics
You can then literally just work your way through the suggestions, focusing on the topic clusters you care most about getting cited in AI for.
For example, the Topical Map for my site shows that I've covered lots of ground around the topic of "domain authority"… but I haven't covered anything about domain rating (i.e., Ahrefs' website authority metric):
If I click into the topic, I can see all the keywords in the cluster and start drafting in Content Editor in one click:
(Alternatively I can import an existing post and optimize that piece for the topic instead.)
In Content Editor, I also have Surfy, Surfer’s AI writing assistant, which helps me get content out the door faster by polishing my half-baked ideas into something people might actually want to read (and that AI actually wants to cite):
Here’s another idea…
You’ll remember that according to our data, you’re ~49% more likely to get cited by ranking for the fanouts than for the main query. And from what I’ve seen, fanouts tend to be lower difficulty than the main keyword.
So… why not start building topical authority by going after the low-difficulty topics/potential fanouts first?
For example, remember how my Topical Map suggested covering “domain rating” to build more topical authority around the topic of “domain authority?” That makes sense. But if you look at the difficulty of that topic, you’ll see that although it’s not super high, it’s not super low either (26/100):
But what I can do is filter my Topical Map for low-difficulty topics (e.g., < KD 20) to reveal easier potential fanouts around that same topic…
Following this approach, you can essentially tip-toe your way to topical authority. You start with the easiest topics, then tackle the slightly harder ones, and so on.
Just make sure to create LLM-friendly content for each topic you cover.
.avif)
