AI Search

AI Overview Citations: What 15 Studies Agree On, and Don't

Ask how often Google's AI Overviews cite pages that already rank in the top 10, and the published studies will give you answers that disagree by a factor of two. Digital Applied compared fifteen studies, from Authoritas in September 2024 to Wellows and Surfer in 2026, and found that the headline numbers differ mostly because the methods differ.

The headline gap

The starkest example comes from Ahrefs alone, which has published both 76.1% and 37.9%, eight months apart. On July 21, 2025, Ahrefs published an analysis of 1.9 million citations from one million AI Overviews and reported that 76.10% of cited pages ranked in the top 10, 9.50% between positions 11 and 100, and 14.40% nowhere in the results. On March 2, 2026, Ahrefs published an update built on 863,000 keyword SERPs and four million cited URLs, and found 37.9% of cited URLs appeared within the first ten SERP blocks, 31.2% in positions 11 to 100, and 31.0% beyond the top 100.

Ahrefs reads the pair as a real change, with its 2026 page saying Google's citation behaviour shifted. But the comparison also shows how much the counting method matters: the 2025 figure counted the top three cited pages against top-10 organic results, while the 2026 figure counted every cited URL against the first ten SERP blocks, including ads, featured snippets, and video packs. The same publisher, the same question, and two different answers, because the denominators differ.

What the other studies add

The other thirteen studies fill in the same picture from different angles. Surfer's 2026 study, on 10,000 keywords with the top 10 pulled for each query and for Gemini-generated fan-out queries, found 67.82% of cited pages outside the top 10 for any of them, and 45.86% when only the top three citations were counted. seoClarity analysed 362,000 US desktop queries that triggered an Overview on October 12, 2025, with 5.1 million citations, and its figures sit in the same broad range.

Eleven of the fifteen studies are vendor studies, and every one of those vendors sells a tool that measures the thing it studied, which the comparison records as an evidence tier. That does not make the numbers wrong, but it does mean they should be read with the method attached. The comparison was built from the primary study pages, read on November 5, 2026, and every row states the denominator and the rank range it used, because the figures are not comparable across rows without them.

Why it matters for SEOs

For SEOs, the comparison is a warning against quoting a single citation statistic as if it were a fact about Google. A claim that 76% of cited pages rank in the top 10 and a claim that 38% do are both true, of different measurements, and neither tells a site owner what to do. The useful question is not what share of cited pages rank, but what the cited pages have in common: they match the query, they are structured for retrieval, and they answer the question the Overview is answering.

For link builders, the comparison is also a reminder to check the method before citing the number. A pitch that quotes a citation statistic without its denominator will not survive a careful reader, and a strategy built on a single vendor's headline figure will be wrong in the details even if it is right in the direction. The direction is clear enough: citations are winnable without a top-10 ranking, and they are winnable by pages that answer the question well.

What to watch next

The comparison itself is the tool to keep. Digital Applied's fifteen-study table, built from the primary pages read on November 5, 2026, is the reference to check before quoting any citation statistic, and the next studies published will either confirm its pattern or complicate it. The question to ask of every new number is the same: what was the denominator, what was the rank range, and what was counted. A statistic without those three facts is a headline, not evidence, and the industry has enough headlines already. The studies that matter next are the ones that publish their method as clearly as their result.

Sources

Sources: Digital Applied, November 5, 2026; Ahrefs, July 2025 and March 2026; Surfer, 2026.

Related reading on this site: Google's November 2026 Core Update Starts Rolling Out, Google Makes Gemini 4 Hydrogen the Default Model in AI Mode, Reddit's Traffic Is Down 13%, Thanks to Google's AI Overviews, OpenAI Launches ChatGPT Wallet So Its Dots Agents Can Pay for You, Ahrefs: Walmart Is the Most-Mentioned Brand in ChatGPT, and AI Overviews Now Cover Over 80% of UK Keywords, SISTRIX Finds.

More from Linkbaits

Turn this playbook into links

Linkbaits helps you create content that earns citations and backlinks. Start with the toolkit and build your first linkable asset this week.

Start free trial