Ranking #1 No Longer Guarantees an AI Overview Citation
For twenty years the SEO goal was simple: get to position one and the clicks follow. That contract is breaking. A new Ahrefs study of 863,000 SERPs and roughly 4 million citation URLs, updated in July 2026, found that only 38% of Google AI Overview citations now come from pages ranking in the top 10 — down from 76% just a year earlier. Rank #1 still matters, but it is no longer the ticket to being quoted in the AI answer that sits above your listing.
That single number reframes how AI search visibility works. The old model — win the top of the SERP, win the citation — has quietly split apart. Below we break down exactly what Ahrefs found, why the shift is happening, and what to change in your content strategy so you stay in the sources AI engines actually pull from.
What the Ahrefs data actually shows
Ahrefs re-ran an analysis it first published in 2025, this time across 863K SERPs and about 4 million citation URLs. The headline is a near-halving of how often AI Overviews cite the pages Google itself ranks highest:
| Where the cited page ranked | Share of AI Overview citations |
|---|---|
| Top 10 (page one) | ≈ 38% (down from ≈ 76% a year ago) |
| Positions 11–100 | ≈ 31.2% |
| Beyond position 100 (or not ranking for the query at all) | ≈ 31.0% |
Read that bottom row again: nearly a third of the sources an AI Overview quotes do not rank in Google's top 100 results for the query the user typed. In other words, Google's own AI is increasingly citing pages that its classic ranking system does not put anywhere near the first page. The connection between "ranks well" and "gets cited" has loosened dramatically in twelve months.
This sits alongside two other findings marketers keep quoting this summer: AI Overviews now appear on close to 48% of tracked queries, and separate Ahrefs data shows they reduce clicks to the top organic result by roughly a third to a half. So more searches trigger an AI answer, that answer steals more clicks, and the sources it credits are drawn from a much wider pool than page one. Every part of that chain works against a pure "rank #1" strategy.
Why the top 10 lost its grip
Ahrefs and the wider SEO community point to one mechanic more than any other: query fan-out.
Query fan-out changes which SERPs matter
Google has confirmed that when AI is triggered, it does not just answer the query you typed. It silently splits that query into a set of related sub-queries — a "fan-out" — runs each of them, and assembles the answer from the pages that show up strongly across those sub-searches. A page that ranks #40 for your exact phrase might rank #3 for one of the fan-out sub-queries, and that is what earns it the citation. The AI is effectively citing the winners of searches the user never saw.
That is why so many cited URLs sit outside the visible top 10. The Overview is not built from one SERP; it is built from a dozen. Optimising a single page for a single head term no longer maps cleanly onto how sources are chosen.
A more capable model reads more widely
The second factor is the model itself. Google made Gemini its global default for AI Overviews in early 2026, and a stronger model can synthesise across far more candidate pages, weigh niche or long-tail sources it would previously have ignored, and lift precise passages from deep in the results. More reading capacity means less reliance on the safe, top-ranked few.
Passages beat pages
Underneath both mechanics is the same truth we covered in our guide to optimizing for AI search: AI engines cite passages, not whole pages. A mid-ranking article with one crisp, quotable paragraph that nails a sub-question will beat a #1 page that buries the same answer in fluff. Ranking is about the document; citation is about the sentence.
What this means for your strategy
The instinct to read this as "SEO is dead" is wrong — and dangerous. Classic ranking still feeds the candidate pool, still wins the clicks that AI Overviews don't intercept, and still correlates with citation more than any other single signal. What changes is that ranking is now necessary but not sufficient. Here is how to adapt.
1. Optimise topics and sub-questions, not just head terms
Because fan-out rewards pages that appear across related sub-queries, breadth of coverage wins. Map the questions around your main keyword and make sure your content — or your cluster of internal pages — answers each one directly. A well-linked hub of pages that collectively own a topic outperforms one page chasing a single phrase.
2. Write self-contained, liftable answers
Lead every section with a one- or two-sentence answer a model can extract without context. Use clear H2/H3 headings phrased as the questions people actually ask, short paragraphs, comparison tables and numbered steps. If a skim-reader finds your answer in five seconds, so does the extractor.
3. Chase citation and referral signals, not just rankings
Rankings under-report AI visibility, and clicks under-report it further. Track a wider set of signals — manual prompt checks in AI Overviews and ChatGPT, and referral visits from AI surfaces. We unpack the measurement side in our breakdown of AI referral traffic in 2026, where ChatGPT alone now drives the overwhelming majority of standalone AI referrals.
4. Build authority beyond the page
When a third of citations come from outside the top 100, entity-level trust — consistent brand mentions, credible authorship, coverage on reputable third-party sites — does more heavy lifting than a single page's backlink count. The same off-site work that helps you rank helps a model feel comfortable quoting you. Our primer on improving your SEO rankings covers those fundamentals.
Rank #1 vs. AI citation, side by side
| Question | Winning the top ranking | Winning the AI citation |
|---|---|---|
| What competes? | The page, on one query | The passage, across fan-out sub-queries |
| Which SERP decides it? | The query the user typed | A dozen related sub-queries they never saw |
| How wide is the source pool? | Effectively the top 10–20 | Top 10, 11–100, and beyond 100 — roughly a third each |
| What earns it? | Authority + relevance + links | A clear, extractable answer on a covered sub-topic |
| How do you measure it? | Position and organic clicks | Citations, brand mentions, AI referral visits |
The mistakes to avoid right now
- Treating rank #1 as the finish line. It is the entry pool, not the prize. If your page one listing isn't structured to be quoted, a page ranking #40 will take the citation.
- Abandoning SEO because "AI killed it." Ranking still drives the ~52% of queries without an Overview and still feeds the citation pool. Pull back now and you lose both channels.
- Optimising one page for one keyword. Fan-out rewards topical breadth. A thin, isolated page can't appear across the sub-queries that decide the citation.
- Ignoring the measurement gap. If your dashboard only shows rankings and clicks, you are blind to whether AI is citing you at all.
The takeaway
The Ahrefs 76%-to-38% drop is the clearest signal yet that AI search has decoupled visibility from ranking. Google's AI now reads across a fan-out of sub-queries with a stronger model, and it pulls the best passage wherever it lives — page one, page ten, or off the SERP entirely. The response is not to abandon SEO but to widen it: own topics rather than terms, write answers a machine can lift, build authority beyond the page, and measure citations, not just clicks. Do that, and it stops mattering whether you rank #1 for the exact query — because you'll be the source the answer is built from.
