AI Overview SEO: Why Top Rankings Stopped Guaranteeing Citations

Ranking number one used to be most of the job. Now it is barely a third of it.

If you sell B2B services and a buyer searches a question your firm should own, Google increasingly answers it for them inside an AI Overview before they reach a single blue link. Every owner asks the same thing in response: does the SEO work still matter? It does. Google says plainly that its generative features sit on top of the same ranking and quality systems as regular Search, so "optimizing for generative AI search is optimizing for the search experience, and thus still SEO," per Google's own guide to generative AI features.

What changed is the payoff structure. Ranking a page in the top 10 used to make a citation in the AI Overview close to automatic. That link has weakened sharply, and the firms that adjust to the new mechanics will keep showing up in answers while their competitors quietly disappear from them.

Structuring content for AI Overviews is one of the technical layers inside our SEO and AEO leadgen service.

What actually changed: top rankings stopped predicting citations

In July 2025, roughly 76% of the URLs Google cited in AI Overviews also ranked in the organic top 10. By early 2026 that number had fallen to 37.9%, according to Ahrefs' analysis of 863,000 keyword SERPs and about 4 million AI Overview URLs, published in March 2026. The remaining citations split almost evenly between pages ranking 11 to 100 and pages that do not rank in the top 100 at all.

Read that again. Most of the pages Google now cites in an AI Overview are not the ones winning the classic search result. A page sitting at position 40, or one that barely ranks for the head term, can land in the answer box that a number-three competitor never reaches.

The cause is a retrieval method Google calls query fan-out. When a search triggers an AI Overview, Google runs more than the query you typed. It splits that query into a set of related sub-queries, runs each of them, and builds the answer from the pages that show up most often across those sub-query results. Ahrefs notes that since AI Overviews moved to Gemini 3 in January 2026, the fan-out appears to reach more widely, pulling from a larger pool of pages. Google describes the same mechanism in its documentation, alongside retrieval-augmented generation, as how it grounds answers in the index.

Query fan-out rewards coverage over single keywords

Most "AI Overview SEO" advice still reads like 2022 keyword strategy with fresh vocabulary: pick the term, write the page, push it to number one. Fan-out breaks that model. If Google answers one visible query by quietly running a dozen adjacent ones, the page that wins is the one present across the whole question cluster. A single trophy ranking no longer carries the answer on its own.

For a B2B service firm, this is the gap between writing one page on "fractional CFO services" and covering the real spread of questions a buyer works through on the way to that decision: what a fractional CFO costs, when to hire one instead of a controller, what the engagement looks like month to month, how it compares to outsourced bookkeeping. Each of those is a sub-query Google might fan out to. A firm that answers four of them with clear, distinct pages has four chances to become the recurring source. A firm with one keyword-stuffed pillar page has one.

The same logic holds whether you sell legal services, IT support, or accounting work. Map the questions a buyer works through before they sign, answer each one on a page of its own, and you give Google many places to find you across the fan-out instead of a single entry point that has to win every time.

This is also why thin, near-duplicate pages built to chase phrasing variations backfire. Google's guide warns directly against rewriting content for AI, "chunking" it into fragments, or generating pages at scale to cover every wording, and files those tactics under its scaled content abuse policy. The win comes from genuine topical depth. Each page answers a real question a buyer has, instead of spinning one answer into twenty thin variations.

Being cited is now the number that pays

Two pieces of research make the stakes concrete. The first measures the traffic loss. Pew Research Center tracked the real browsing behavior of 900 U.S. adults across nearly 69,000 searches and found that users clicked a traditional result on just 8% of searches that showed an AI summary, against 15% on searches without one. Clicks on the links inside the summary were rarer still, at 1% of all visits. When the AI answers the question, most people stop reading.

The second piece measures who survives that drop. Seer Interactive analyzed 3,119 informational queries across 42 organizations, covering 25.1 million organic impressions from June 2024 to September 2025. Organic click-through rate on queries that showed an AI Overview fell 61%. But brands cited inside the overview earned 35% more organic clicks and 91% more paid clicks than when they were not cited. Seer is careful to call this a correlation rather than proven cause, since authoritative brands may simply get cited more often. The pattern holds either way: on a query where an AI Overview appears, getting cited is what separates staying visible from going dark.

That reframes the goal. The old target was the number-one position and the click attached to it. The new target is inclusion in the answer itself, because the citation is what carries your brand into the moment a buyer decides who to call. Inclusion is also more reachable than dominance. Pew found that 88% of AI summaries cite three or more sources. You do not have to beat everyone. You have to be one of the handful Google trusts enough to quote.

What actually earns the citation

Google's guide is unusually direct here, and it lines up with what the citation data shows. A short list of things matters more than the rest.

Your page has to be eligible. It must be indexed and able to appear with a snippet in regular Search. There is no separate AI ranking system to game and no AI-specific markup to add. Structured data is not required for AI features, though it still helps you qualify for rich results.

Your answer has to be extractable. AI Overviews lift discrete claims, so a page that states the answer plainly near the top of the relevant section gets quoted more often than one that buries it inside a long narrative. This is the same discipline that wins featured snippets, and the overlap between snippet-winning pages and AI-cited pages runs high.

Your content has to be worth quoting. Google says it favors non-commodity content with a clear point of view, first-hand experience, and original analysis over recycled summaries of what already ranks. If your page repeats the consensus, Google has no reason to cite you instead of the higher-authority page that said it first. A proprietary number, a framework you actually run with clients, or a position the rest of the results avoid gives the model a reason to reach past the obvious sources.

Your brand has to read as a real entity. Google judges source authority partly through entity signals, like consistent naming across your web properties and named authors with verifiable credentials. A page from a firm with a clear identity and demonstrated expertise gets chosen over an anonymous one making the same claim.

Wire it together as one system

Most firms lose ground here because they treat search, content, authority, and measurement as separate projects owned by different people or vendors. Fan-out punishes that split. Topical coverage, entity consistency, extractable answers, and citation tracking only compound when they run off one plan aimed at one outcome.

I spent five years leading marketing at CFO Hub while it made the Inc. 5000 four years running, and the lesson that carried over is simple: growth came from one coordinated engine rather than a pile of disconnected campaigns. AI Overview SEO works the same way. The content cluster gives Google enough surfaces to fan out across. The entity signals tell it your firm is a source worth trusting. The measurement layer tells you which queries trigger an overview, whether you got cited, and what that does to booked calls, since average-position tracking alone no longer captures the layer that decides revenue.

A B2B service firm that builds those four pieces around a single goal keeps appearing in the answers its buyers see. One that ships isolated keyword pages and watches average position keeps losing citations it never knew it was eligible for.

To see where your firm stands today, the SEO and AEO checklist walks through the eligibility and structure points above, and our SEO and AEO service and full AI Marketing Department are built to run the whole engine as one system. For the same shift inside other AI surfaces, how to rank in ChatGPT and GEO versus SEO cover the adjacent engines.