I think most of us would assume that Google’s AIO citations (like the ones pictured below) rank somewhere in the regular “blue link” results—either for the query itself or fan-out queries.
But according to our data, this isn’t always the case…
We pulled the SERPs for 10,000 keywords, filtered for those with AIOs (spoiler: it was 76% of them), then pulled the top 10 results for the keywords and their fan-out queries using Gemini.
The result?
67.82% of AIO cited pages didn’t rank in Google’s top 10, neither for the query itself or any of the fan-out queries.
However, when we filtered only on the top 3 citations (you know, the ones you actually see in the SERP without having to click the “Show all” button), this number dropped to 45.86%.
This means that the majority of the top 3 citations do rank in the top 10 for the main query, at least one of its fan-outs, or both.
But still, whichever data you look at, it’s clear that a lot of AIO citations don’t rank in the top 10 regular “blue link” results at all.
This seems odd, so what’s going on here? Where do they come from?
Let’s explore a few possibilities…
AIOs might be citing pages that rank beyond the top 10
For our study, we only pulled the top-10-ranking pages for the keywords and their fan-out queries. We didn’t pull any positions beyond that, so it’s possible (and in my opinion, likely) that many AIO citations simply rank in positions 11+.
This is definitely true for at least some keywords.
For example, the top AIO citation for “what is the best time to publish content on tiktok” is this page from Shopify:
If I check the rankings for that keyword in Surfer’s Rank Tracker, I see that page ranking in position #12:
It’s a similar story for “which adapter do i need for germany”…
The top citation is this page from IAmExpat.de:
According to Surfer, the page does rank in the “blue link” results—but at position #32:
Of course, both of these examples could just be a coincidence. It’s always possible that AIO didn’t pull these citations from “blue link” organic results but other sources instead. Which brings me neatly onto…
AIOs might be citing pages from other Google-owned properties
If you look at the most cited domains in AIOs, two of the top three are google.com and youtube.com:
In fact, these two sources combined are responsible for ~40% of AIO citations!
Given the fact that Google owns both of these sites, it’s possible that AIOs are pulling directly from these data sources (in addition to Google’s web index).
That said, there’s not really any evidence to suggest that this is what’s happening. None of Google’s patents related to how AIOs work talk about this. So although it’s a possibility, I think it’s more likely that these domains are cited frequently simply because they’re two of the biggest sites on the web and rank for millions of queries.
We have the same question as above when it comes to Reddit—the fourth most cited domain in AIOs. Although we know that Google cut a deal to use Reddit’s data to train its AI, we have no idea whether AIO citations somehow prioritise Reddit. Once again, I think it’s more likely that we see Reddit frequently cited in AIOs because the site is huge and ranks in “blue link” results for millions of queries.
AIOs might be citing hallucinated pages
Every LLM has issues with hallucination, so it’s highly likely that some of the URLs cited in AIOs simply don’t exist (and never did).
This data from Ahrefs backs this up:
You can see in the graph above that 0.86% of the URLs cited by Gemini have a 404 status code.
Note. Ahrefs actually ran this same study across two different data sources, and the percentage of hallucinated URLs cited by Gemini was only 0.21% in the other one.
But even if we ignore the fact that Gemini and AIOs aren’t exactly the same thing, this is a very low number. It might explain partly why we see many AIO citations that don’t rank in Google’s top 10, but it can’t be the sole reason.
AIOs might be searching for different fan-out queries to us
Right now, there’s no way to see exactly which fan-out queries AIOs actually search for. So, for our study, we used Gemini fan-outs as a proxy.
This is important because, even though Gemini powers AIOs, additional mechanisms may affect which fan-out queries an AIO actually searches for. If these differ from Gemini, some AIO citations from our study could actually be ranking for fan-out queries—even if our data doesn’t show it.
For example, here are the top three citations for “best time to visit asturias spain”:
The first two rank on the first page of Google, but the third doesn’t (and I don’t see it ranking anywhere in Google’s top 100 for this query either).
But maybe it ranks for a fan-out query?
According to Gemini, that doesn’t appear to be the case. When I checked, the page wasn't ranking for any of the fan-outs Gemini generated.
But that doesn’t mean it isn’t ranking for one of the fan-outs the AIO searched…
Unlike Gemini, the AIO could have searched for something like “things to do in gijon spain”, where the cited page ranks in the “blue link” results.
As you can see, the problem I’m highlighting here is a key limitation of all AIO studies. Keep that in mind when looking at any data related to AIOs, both from us and other brands.
How can I increase my chances of getting cited in AIOs?
If you want to get cited in an AIO, it seems that your best bet is to rank for your target query and its fan-outs. This is true because the majority (54.14%) of top 3 AIO citations rank in the top 10 for the main query, at least one of its fan-outs, or both.
Ranking for multiple fan-outs also massively increases your chances, according to our data:
But what else besides ranking can you do to improve your chances?
Here’s one method:
- Track AIO citations for important queries
- Find the most frequently cited pages across these queries
- Reach out and try to get featured on those pages
This is something you can do with Surfer’s AI Tracker.
For example, here are the most cited pages in AIOs for a bunch of queries I’m tracking related to the best SEO tools:
Maybe some of these pages rank in Google for my tracked queries or their fan-outs, maybe they don’t. It doesn’t matter because I know they’re already getting cited, so all I need to do is to try to get mentioned on them.
This is an especially smart move for “best [x]” queries, since being mentioned on cited pages can also lead AIOs to recommend you.
For example, Surfer is recommended in the AIO for “SEO software tools”...
… even though none of the cited pages come from our website.
That brings us to the bigger point: if you really want to increase your chances of being recommended and cited in AIOs, you can’t just rely on your own website. Your brand needs to show up across trusted, relevant sources—the ones Google’s AI already treats as credible.
Learn more about how to do this in our study on AIO sources.
.avif)