Last verified: 3 November 2026
To appear in Google’s AI Overviews and AI Mode, a page has to be indexed and eligible to show a snippet in ordinary Search, and it has to answer one of the narrower questions Google searches for when it builds the answer. Google says no special markup, text file or rewrite is needed; the pages it cites are found by its normal ranking systems plus query fan-out.
Key takeaways
- AI Mode breaks a question into subtopics and runs many searches at once, which Google calls query fan-out (Google, May 2025).
- Only 37.9% of pages cited in AI Overviews rank in Google’s top 10 for the same query; 31.0% rank beyond 100 (Ahrefs, March 2026, 863,000 keyword SERPs).
- AI Overviews are associated with a 58% lower average click-through rate for the top-ranking page (Ahrefs, February 2026).
- Adding schema made no meaningful difference to citations across 1,885 pages tracked against 4,000 controls (Ahrefs, May 2026).
What does Google need before it will cite a page?
The same things it needs for Search. Google’s guidance says pages must be indexed and eligible to show a snippet to appear in AI features. So the basics come first: the page is crawlable, not blocked by robots.txt or a noindex tag, and does not carry a nosnippet directive. Answers hidden behind tabs loaded by script, login walls or images are harder for any system to use.
Google is also clear about what it does not need. It does not use llms.txt files, and it does not require content to be rewritten into chunks for AI. It warns that creating pages mainly to target fan-out queries can breach its scaled content abuse policy.
Why do pages outside the top 10 get cited?
Because of fan-out. When Google splits a question into sub-questions, it takes citations from the results of those narrower searches, not only from the results page for the original query. A year before Ahrefs’ March 2026 update the top-10 share was 76%; it fell to 37.9% as fan-out grew.
That cuts both ways. A page on page three for your head term can still be quoted if it answers a narrower question well. A page ranking first can be left out if it does not answer the sub-questions cleanly.
Are AI Overviews and AI Mode the same?
They share Google’s index and ranking systems, but AI Mode uses heavier fan-out and draws on a more balanced mix of domains. Semrush’s study of more than 100 million citations found LinkedIn, YouTube and Reddit among AI Mode’s five most-cited domains (Semrush, November 2025). In Profound’s study, Reddit, YouTube and Quora led AI Overview citations. Forum threads and video carry real weight in both.
On 15 October 2026 Google said both will show more inline links, higher up in responses, and that it is testing a section of official sites at the top of AI answers. More links inside the answer means more value in being one of the pages it relies on.
What should you change on your site?
- Find the questions behind your head terms. List the sub-questions buyers need answered: cost, process, eligibility, comparisons, risk.
- Give each a direct answer. Open the page or section with one or two sentences that answer it, then the detail.
- Keep facts in text and tables. Not PDFs, images or script-loaded tabs.
- Merge thin and duplicated pages. One clear page per question is easier to cite than five similar ones.
- Corroborate off-site. Point digital PR and links at the same answer pages.
- Use video where it helps. Ahrefs found 18.2% of AI Overview citations from outside Google’s top 100 were YouTube URLs.
What does it look like in practice?
Stelrad reached a 62% share of voice in AI Overviews for radiator questions and 557 AI Mode citations across 231 pages after its advice guides were rebuilt as direct answers and thin pages were merged. Fast Loan UK had loan pages cited in AI Overviews for payday loan and quick loan queries.
Because AI Overviews reduce clicks, measure citations and share of voice alongside traffic, not instead of it. Our AI visibility service starts with exactly that baseline.
