Your Best-Ranked Page Might Be Invisible to Google’s AI

Earn a spot in the top 10, close the tab satisfied, and head out to happy hour.

That’s how it used to be. If you appeared high on the search results page, you felt confident about your page’s performance.

In the past, pages in Google’s top 10 provided most citations in AI overviews. However, this share has changed significantly in less than a year due to a new challenge: the query fan-out. While it’s important to acknowledge that AI Overviews can sometimes include errors, focusing on being cited rather than just ranked will help your strategy thrive through these changes.

Key takeaways

• Ranking and citation have split apart. The share of AI Overview citations that also rank in Google’s top 10 fell from about 76% in July 2025 to roughly 38% by March 2026.
• Query fan-out is the cause. Google’s AI breaks one question into many sub-queries and builds its answer from the pages that surface most consistently across all of them—not just the page that ranks for the typed query.
• Top-10 ranking still matters. A strong organic position remains the most reliable feeder into AI Overviews and the clearest authority signal Google has—it gets you into the candidate pool.
• Citation takes depth and credibility. Answer engine optimization (AEO) rewards self-contained sections, topic-level coverage, and E-E-A-T signals that let a model lift a clean, quotable claim.
• Track citations, not just rankings. Position tracking alone now misses most of the picture of where your traffic comes from.

What is a query fan-out?

Query fan-out is how an AI search system breaks one user query into several sub-queries. It collects information for each, then combines the results into one response. LLMs rely on it to produce richer answers.

When you ask a question in Google’s AI experiences, the system expands it before it answers. Behind the scenes, an AI model breaks your question into a set of related sub-queries—equivalent phrasings, follow-ups, broader framings, narrower specifications—and runs them all at once.

The AI Overview is then built from the pages that surface most reliably across that whole set. A page can rank first for its headline query, but it might not show up in the fan-out. This happens because the model looks at several related searches where other pages provide more detailed information.

Here is what that looks like in practice:

“How do I measure the ROI of our B2B content marketing program to prove its value to executives?”

Instead of running that one query and stopping, the LLM pulls the question apart into shorter searches drawn from inside it:

• measure content marketing ROI
• B2B content marketing metrics
• content marketing value
• prove content ROI to executives
• content program performance

That shift—finding answers based on the most consistent pages, not just the typed question—is what separates ranking from citation.

Why ranking still matters

Roughly half of Google searches already surface an AI summary, and McKinsey projects that figure will pass 75% by 2028. In a McKinsey survey of 1,927 US consumers, half now actively seek out AI-powered search, and it has become the leading digital source they use for buying decisions. With most searches headed toward an AI answer, the pages that get cited decide most of your traffic.

In July 2025, about 76% of pages cited in Google’s AI Overviews also ranked in the top 10 for the same query. Ahrefs’ March 2026 study looked at 863,000 keywords and around 4 million AI Overview URLs. It found that figure had dropped to about 38%.

The rest of the citations moved elsewhere on the web. Ahrefs found them split almost evenly: roughly 31% from pages ranking 11 to 100, and another 31% from pages ranking past 100 or not ranking for the query at all. Ranking and getting cited no longer go together.

Don’t throw out your plans to rank just yet. A 38% overlap is still a large minority, and top-10 pages remain the most reliable feeder into AI Overviews. A strong organic position is still the clearest authority signal Google has. Ranking well gets you considered. Getting cited takes more.

Think of it as two gates. Traditional SEO gets you into the candidate pool, and fan-out decides which candidates get quoted. A page that ranks and covers its topic with real depth clears both. A page that ranks for one keyword and stops there clears the first and stalls at the second.

What AEO actually asks of your content

This is where answer engine optimization enters. Structure helps: clear headings, self-contained sections, schema, and a direct answer near the top all make content easier for a model to parse and extract.

What you shouldn’t ignore is coverage and credibility. If the AI samples sub-queries, your content has to answer the surrounding questions in addition to the main query. That means depth over keyword breadth: one resource that resolves the real question and its natural follow-ups, written with enough specificity that a model can lift a clean, citable claim from it. The same E-E-A-T (experience, expertise, authoritativeness, and trustworthiness) signals Google has always rewarded are what make a passage worth quoting.

AEO is the same demand for good content with the stakes raised, where every section now has to stand on its own.

Where to spend your effort now

Fan-out rewards content that anticipates the questions a reader actually has, and anticipating those questions is editorial judgment.

Knowing which sub-questions matter, which framings are honest, where to be specific and where to be brief, what claim is worth stating cleanly enough to be quoted—that is the work of an experienced editor or subject-matter expert.

The brands that get cited consistently share a trait: their content carries a clear point of view and the depth to support it across a topic. Volume of output has little to do with it.

Strategies include:

• Build around topics, not single keywords. Map the cluster of questions a reader asks before and after the headline one, and cover them in depth within a page or a tight set of linked pages.
• Make every section independently citable, since a model will lift passages out of context.
• Stay the course on earning the ranking, because top-10 placement is still your surest path into the candidate pool.
• Watch where you are cited, not only where you rank. Position tracking alone now misses most of the picture.
• Invest in editorial depth and expertise. The thing that makes content worth citing holds steady: depth, judgment, and a point of view a reader, or a model, can trust. That’s editorial work, and it’s exactly what Contently’s network of expert editors and subject-matter writers is built to do.

Frequently asked questions

What is a query fan-out in AI search?

Query fan-out is the technique an AI search system uses to break a single user query into several related sub-queries—equivalent phrasings, follow-ups, broader framings, and narrower specifications. It runs them all at once, then builds its answer from the pages that surface most consistently across the whole set rather than from the one page that ranks for the typed question.

What is the difference between SEO and AEO?

SEO (search engine optimization) works to earn a high ranking on the results page, which gets your page into the pool of candidates an AI can draw from. AEO (answer engine optimization) works to get your content quoted in the AI answer itself, which depends on self-contained sections, topic-level depth, and E-E-A-T signals a model can lift a clean claim from. SEO gets you considered; AEO gets you cited.

Does ranking in Google’s top 10 still matter for AI search?

Yes. Even though the overlap between top-10 rankings and AI Overview citations fell to about 38% by March 2026, top-10 pages remain the single most reliable feeder into AI Overviews and a strong organic position is still the clearest authority signal Google has. Ranking gets you into the candidate pool; citation takes additional depth and credibility.

How do I get my content cited in Google’s AI Overviews?

Cover a topic—not a single keyword—deeply enough to answer the surrounding sub-questions a reader and the fan-out will ask. Structure each section to stand on its own with clear headings, schema, and a direct answer near the top, and write with enough specificity and demonstrated expertise (E-E-A-T) that a model can extract a clean, quotable claim.

What is E-E-A-T and why does it matter for AEO?

E-E-A-T stands for experience, expertise, authoritativeness, and trustworthiness—the signals Google has long rewarded in rankings. For AEO they matter because the same qualities that make a passage credible to Google are what make it worth quoting to an AI model: specific, well-sourced, expert content is the kind a model is most willing to cite.