Top 10 Rankings and AI Overview Citations: What the Data Shows

Ahrefs published its study on March 2, 2026. It found that 37.9% of URLs cited in AI Overviews also appeared within the first 10 blocks of Google's results for the same query. Counting organic results only, Ahrefs put the figure at 37.1%.

The remainder split roughly evenly. Ahrefs found about 31% of cited URLs ranked at positions 11 to 100, and about 31% ranked beyond position 100.

Read that plainly. A top-10 page shows up in AI Overview citations far more often than its share of all ranking pages would suggest. But most cited URLs did not rank on page one for the query that produced the Overview.

One detail matters for larger sites. Ahrefs matched citations at the URL level. If Google cites a different page from your site than the one that ranks, the study counts that citation as coming from outside the top 10. That mechanically understates how often top-ranking sites, rather than top-ranking URLs, get cited.

For a full breakdown of the method, see our digest of Ahrefs' AI Overview top-10 study.

You will see the 38% figure quoted next to an earlier, much higher Ahrefs number as proof that rankings are losing influence. That comparison does not hold. Ahrefs changed its parsing methodology between the two studies, so the endpoints came from different instruments. The March 2026 figure stands on its own. The trend does not.

A second vendor disagrees outright. BrightEdge reported on September 18, 2025 that only 16.7% of AI Overview citations came from the top 10, while also claiming 54.5% overlap. BrightEdge published no sample size and did not define what "overlap" means.

One of these methods is wrong, or they measure different things. Do not cite them as agreeing.

What both do suggest is that a large share of citations come from pages outside the top 10. That rules out two popular claims at once. A top-10 ranking is not a guarantee of citation. And "rankings no longer matter" is not supported either.

It helps to split the process into stages, because rank only touches one of them.

Access and index eligibility

Google's AI features documentation says a page must be "indexed and eligible to be shown in Google Search with a snippet" to appear as a supporting link in AI Overviews or AI Mode. It adds that there are "no additional technical requirements." The same page warns that indexing and serving isn't guaranteed.

This gate is binary. In OtterlyAI's 2026 experiment, Google deindexed two test sites, and Google's AI surfaces almost never cited them. Bing kept the sites, and Microsoft Copilot kept citing them. An engine cannot cite what its index no longer holds.

Retrieval

Google says AI Overviews may use "query fan-out", issuing multiple related searches across subtopics. So the page Google retrieves may rank for a sub-query you never tracked, not the query the user typed. That is one reason organic ranking for a single keyword is a loose proxy for citation.

Retrieval also depends on whether an Overview appears at all. As of October 2026, AI Overviews often do not trigger, and Google says AI Overviews and AI Mode "may use different models and techniques" and return different links.

Generation and citation

Once pages are retrieved, the model chooses which to use and cite. Research such as C-SEO Bench found that a document's position in the model's context strongly affects citation. But that study moved documents to position one directly; it never queried a real index. No publisher can set that position. Our page on AI Overviews ranking signals grades the evidence for each stage.

In your control

• Index status. Whether Google has your page indexed. Check it in Search Console.
• Snippet eligibility. noindex, nosnippet, max-snippet:0 and data-nosnippet all keep content out of AI Overviews. Templates often apply them by accident. Bing honors data-nosnippet too, as covered in our Bing data-nosnippet digest.
• Googlebot access. There is no separate robots.txt token for AI Overviews. Googlebot rules are the control.
• The Search Console generative AI control. Inclusion is the default for all properties. It is an opt-out, it is still rolling out to a subset of owners, and child properties inherit it from the parent. A property can be excluded without anyone noticing. Our page on opting out of Google's AI features covers how it interacts with Search.
• Topical fit. Whether your pages actually cover the subtopics a buyer asks about.
• Specific facts on the page. A price, a number, or a one-line definition stated in the page gives the engine a sentence it can quote.

Not in your control

• Whether an Overview triggers for a query.
• Which fan-out queries Google runs.
• Which retrieved pages the model cites.
• Your brand's prominence relative to competitors, which is strongly associated with citation but is not a lever you can pull this quarter.

One Google-specific note: Googlebot renders JavaScript. Client-side rendering is a serious risk with ChatGPT and Claude crawlers, but it is not the main gate for AI Overviews.

Work from the hard gates outward. Each step removes a reason you cannot be cited, before you spend effort on things that only might help.

1. Confirm indexing. Run URL Inspection in Search Console on the pages that answer your buyers' questions. A page that is not indexed cannot be a supporting link.
2. Check snippet directives. Search your templates and HTTP headers for noindex, nosnippet, max-snippet and data-nosnippet. Remove any you did not intend.
3. Check Googlebot rules and the generative AI control. Make sure robots.txt does not block Googlebot on key paths. If your Search Console shows the generative AI control, confirm the setting on every property, including inherited child properties.
4. Keep doing normal SEO. Google's AI optimization guide says no special optimizations are needed for AI Overviews. Ranking is still one of the better proxies available, even if it is a decaying and contested one. Our guide to how AI search features affect organic results covers the overlap.
5. Cover the subtopics, with facts. Because of query fan-out, a page that answers the related questions has more chances to be retrieved. State specific facts in the page text rather than leaving them implied.
6. Measure the surface directly. Search Console's generative AI report shows impressions only, as of October 2026, with no clicks or position. To see whether an Overview appears for a query and which sources it cites, the free AI Overview Checker shows the current result.

• Reading 38% as a trend. The earlier Ahrefs figure came from a different method. Do not tell your team rankings are collapsing as an AI Overview input.
• Treating the top 10 as a guarantee. Ahrefs found most cited URLs ranked outside it. A page-one ranking with a nosnippet tag will not be cited.
• Writing rankings off entirely. Neither study supports it. Abandoning SEO to chase AI-specific tactics trades a weak proxy for tactics with less evidence.
• Adding schema to get cited. Google says structured data isn't required for generative AI search. Ahrefs' matched test found AI Overview citations fell 4.6% after pages added schema. Existing markup does no harm, but it is not a lever.
• Chasing domain authority. Domain Authority is a Moz metric. Google publishes no site-level authority score. Our page on backlinks and AI search separates what links do from what vendors claim.
• Tracking an "AI rank". Google exposes impressions only. No position data exists to check such a number against.
• Benchmarking against giants. Large brands dominate citations everywhere. Compare yourself to similar-sized competitors, or you are measuring brand size.

Two vendor studies disagree, and neither is replicated. That uncertainty points to a simple rule: spend first on the things Google documents, and treat everything else as a bet.

Google documents one requirement for AI Overviews: indexed and snippet-eligible. That is checkable today, it is fully yours, and failing it removes you regardless of rank. Fix it first.

Organic ranking comes next. It is associated with citation in the Ahrefs data, and it drives traffic from classic results whether or not an Overview appears. Keep investing in it for the queries that bring buyers.

After that, focus on topical coverage and clear, quotable facts. They do not promise a citation. They give Google more ways to retrieve your page and a sentence to use when it does.

Nobody outside Google can tell you exactly why one page was cited over another. Be wary of anyone who claims to.

Does ranking number one guarantee an AI Overview citation? No. Google says eligibility requires a page to be indexed and shown with a snippet, and that indexing and serving isn't guaranteed. Ahrefs found that most URLs cited in AI Overviews ranked outside the top 10 for the query. A top ranking makes citation more likely in that data, but the model still chooses among retrieved pages, and an Overview may not appear at all.

Can a page ranked beyond position 100 be cited in an AI Overview? Yes. Ahrefs found roughly 31% of cited URLs ranked beyond position 100 in its March 2026 study. One likely reason is query fan-out: Google may run several related searches, so a page can be retrieved for a sub-query rather than the one being tracked. URL-level matching also counts a site's non-ranking page as outside the top 10.

Can I track my position inside an AI Overview? Not reliably. As of October 2026, Google's Search Console report for generative AI features shows impressions only, with no clicks, click-through rate or position. Google also says third-party tools don't have access to its internal ranking data. Any "AI rank" number has no ground truth to check it against. Track whether an Overview appears and whether you are cited instead.

Does adding schema markup help AI Overviews cite my page? The evidence says no. Google states structured data isn't required for generative AI search. Ahrefs tracked 1,885 pages that added schema against matched controls and found AI Overview citations fell 4.6%. Those pages were already heavily cited, so whether schema helps a page enter consideration is untested. Keeping existing markup does no harm.

Should I trust the Ahrefs figure or the BrightEdge figure? Treat both with care. Ahrefs published its method and a 37.9% top-10 share, but matches at the URL level. BrightEdge reported 16.7% with no published sample size and an undefined overlap measure. They cannot both describe the same thing. For planning, rely on what Google documents: index and snippet eligibility are the hard requirements.