Ahrefs Study: Google AI Overview Citations from Top-10 Pages Drop to 38%
By Charles Ryder
When Google rolled out AI Overviews in May 2024, early patterns were reassuring: rank in the top 10 and you were likely to get cited in the summary. Not always, but often enough that traditional SEO performance still felt like the main ticket to AI visibility.
That assumption is fading quickly.
Ahrefs’ newly updated research shows that only 38% of Google AI Overview citations now come from top-10 organic results. The connection between “ranking well” and “getting cited” has weakened sharply—and it doesn’t look accidental.
The headline stat: from ~76% to 38%
In Ahrefs’ earlier analysis, roughly three-quarters of AI Overview citations came from pages ranking in the top 10. The takeaway was straightforward: AI Overviews mostly repackaged what was already winning on page one.
In the latest update (published in early March 2026), Ahrefs expanded the dataset dramatically—analyzing 863K keywords and 4M AI Overview URLs. The overlap dropped to ~38% (reported as 37.9% overall and 37.1% for organic-only).
So where are the missing citations coming from?
- A large share now comes from pages ranking #11–#100 (the “page two to page ten” zone).
- Another large share comes from pages outside the top 100 entirely.
If your AI Overview strategy has been “rank top 3 and you’re set,” this data argues otherwise.
Why this is happening: Gemini 3 + query fan-out
Ahrefs points to two shifts that help explain the new citation behavior.
1) Google’s model upgrade (Gemini 3) is expanding the source pool
In January 2026, Google announced a major AI Overview update: Gemini 3 (Flash) became the default model powering Overviews globally. Google also said AI Overviews would use more sources per response (reported as 32% more). More sources naturally increases the odds that citations come from deeper in the index—especially on nuanced, long-tail, or multi-intent queries.
2) “Query fan-out” changes what “relevant” even means
This is the bigger strategic change.
Rather than relying primarily on the visible SERP for the original query, Google increasingly uses query fan-out: it breaks a search into related sub-queries, pulls results from those SERPs, then synthesizes an answer across them.
A simple example: a query like “Best puppy food” can trigger sub-queries such as:
- “puppy nutrition requirements”
- “best food for large-breed puppies”
- “AAFCO puppy food guidelines”
- “common puppy food allergies”
Your page can rank #1 for the head term and still miss citations if it doesn’t show up—or doesn’t satisfy intent—across those supporting sub-queries.
SEO consultant Ethan Lazuk summed it up bluntly: we’re shifting from optimizing for single keywords to optimizing for full user journeys—and the specific passages that answer the branching questions inside them.
What this means for SEO: rankings still matter, but they’re not the gatekeeper anymore
This isn’t an obituary for rankings. It’s a change in how much rankings filter visibility. In AI Overviews, rank is no longer the dominant gatekeeper for citations.
That changes execution in a few practical ways.
Your content must win “intent fragments,” not just head terms
AI Overviews pull from chunks: definitions, steps, comparisons, caveats, and clarifications. A page can be strong overall but still lose citations if it’s thin on the exact sub-questions the model needs to resolve. Sometimes the narrower page wins because it answers one fragment perfectly.
Topic clusters and coverage breadth matter more than ever
Fan-out tends to reward sites that offer:
- Topical depth (not just one “money page”)
- Interlinked supporting content that covers adjacent questions
- Consistent terminology and clearly structured answers
Zero-click pressure is increasing—citations become branding, not just traffic
AI Overviews often reduce clicks (multiple studies report meaningful CTR drops when Overviews appear). That makes citations feel less like a traffic lever and more like the new “impression”—brand visibility even when the visit never happens.
When clicks are compressed, repeated citation becomes a proxy for authority. Users may not click every time, but they notice which names show up again and again.
The surprising winner: YouTube’s citation growth
One of the more interesting changes in the Ahrefs update is YouTube’s rising share:
- YouTube accounts for 5.6% of all citations
- It represents a much larger share of citations from pages outside the top 100
- Ahrefs reports 34% growth in six months
This matches what many SEOs have been seeing in the wild: Google is increasingly comfortable citing video as a source—especially for how-tos, demonstrations, troubleshooting, and anything hands-on or experiential.
If you’re not treating YouTube as a core search asset—with strong titles, descriptions, and transcripts—you may be missing AI Overview visibility that competitors are picking up.
How I’d adapt strategy right now
If I’m advising a team based on this data, here’s what I’d prioritize immediately:
- Map fan-out sub-queries for your most valuable topics
Skip generic “related keywords” lists. Identify the likely sub-intents Google needs to cover: definitions, comparisons, steps, edge cases, and common objections. - Build passage-level clarity
Use tight sections, explicit headings, and direct answers. AI Overviews don’t cite “vibes.” They cite extractable blocks that match a specific intent. - Strengthen topical authority with clusters, not isolated posts
Build a hub-and-spoke structure where supporting pages can win different branches of fan-out—and where internal linking makes those relationships obvious. - Diversify formats (especially video)
If YouTube is being cited more often, treat it like a first-class channel—particularly for topics where showing beats telling. - Track citations as a KPI, not a curiosity
Rankings and traffic still matter, but they’re no longer the whole scoreboard. You need to know: Where are we cited? Where are we absent? Who replaced us?
FAQ
Does this mean getting into the top 10 doesn’t matter anymore?
No. Strong rankings still correlate with visibility and authority. The change is that top-10 status no longer guarantees citations—and a growing share of citations now comes from much deeper results.
Why would Google cite pages outside the top 100?
With query fan-out, Google may pull sources that best answer a specific sub-question, even if that page doesn’t rank well for the original head term. A niche page can be highly relevant for one fragment of intent.
What type of content is most likely to earn AI Overview citations?
Content with clear, self-contained passages: direct definitions, step-by-step instructions, comparisons, and well-labeled sections that answer common follow-ups and edge cases.
Conclusion: the new SEO reality is visibility across the fan-out
Ahrefs’ finding—top-10 pages dropping to just 38% of AI Overview citations—looks like a structural change. Google is building AI answers from a wider, more fragmented pool of sources, driven by Gemini 3 and query fan-out behavior.
The takeaway: rankings still help, but topical coverage is becoming the citation engine. If your content strategy can’t support the branching questions around a topic, you’ll struggle to show up in AI Overviews—even while “winning” traditional SERPs.
Because this landscape keeps shifting, it’s worth setting up a workflow to monitor AI-driven visibility and respond quickly. For teams that want to operationalize citation tracking and authority-building, AIuthority can help turn AI Overview volatility into a measurable, repeatable strategy.