You do not rank in an AI Overview. You get cited in one, and according to Google’s own documentation the only requirements are that the page is indexed and eligible to be shown with a snippet. There is no AI-Overview-specific markup, file, or submission.
That is an unsatisfying answer to sell, which is probably why so much of what is written about this topic says otherwise. As of September 2026, Google’s Search Central documentation on AI features states directly: “There are no additional requirements to appear in AI Overviews or AI Mode, nor other special optimizations necessary.” Everything useful therefore happens upstream of the feature, in whether your page is the best available thing to quote on the question.
In short
- Google states there are no additional requirements and no special optimisations for AI Overviews beyond normal Search eligibility.
- The page must be indexed and eligible for a snippet. That is the gate, and it is the same gate as ordinary Search.
- Kunzum measured Google AI Overviews citing 7.7 sources per answer across 120 answers in September 2026 — denser than ChatGPT or Claude.
- Google-Extended is a separate control governing training and grounding, not Search indexing.
- The workable levers are ordinary: be indexed, be quotable on the specific question, and be the source other pages state the fact from.
What Google actually says
Kunzum quotes this at length because it is repeatedly misreported. From Google’s Search Central documentation on AI features, as of September 2026: to be eligible, “a page must be indexed and eligible to be shown in Google Search with a snippet, fulfilling the Search technical requirements. There are no additional technical requirements.”
And on markup specifically: “You don’t need to create new machine readable files, AI text files, or markup to appear in these features. There’s also no special schema.org structured data that you need to add.”
If an agency proposal for AI Overview visibility is built on schema, llms.txt files, or an AI-specific technical audit, the largest engine in the set has published a denial of the mechanism. Kunzum publishes an llms.txt and uses schema, and neither is billed as the reason engines cite this site.
How densely AI Overviews cite
Google AI Overviews is a search surface, so unlike the chat engines it retrieves for every answer. What varies is how many sources it pulls. Kunzum logged this across both editions — 120 AI Overview answers in edition one and 270 in edition two:
| Engine | Citations per answer, edition one | Citations per answer, edition two |
|---|---|---|
| Google AI Overviews | 7.7 | 7.1 |
| Perplexity | 19.3 | 19.4 |
| ChatGPT | 3.6 | 3.5 |
| Claude | 2.7 | 2.1 |
Two practical consequences. AI Overviews has more citation slots per answer than ChatGPT or Claude, so the marginal source has a better chance of being one of them. And because the citation count is high while the answer stays short, most cited pages contribute a sentence rather than the answer — which is an argument for having one genuinely quotable passage rather than a comprehensive page.
What is actually workable
Given Google has ruled out the technical levers, what is left is unglamorous and is most of the method.
Be indexed and snippet-eligible. This is the only stated requirement, and it is worth verifying rather than assuming. A noindex left on a template, or a nosnippet directive, removes the page from consideration entirely.
Answer the specific question in one place. An AI Overview is assembled from passages. A page that states the answer in a single self-contained passage is usable; a page that builds to it over eight paragraphs is not, no matter how good the eighth paragraph is.
Be the source other pages state the fact from. In Kunzum’s edition one corpus, reddit.com was the most-cited domain at 293 citations, and editorial press accounted for 26 of 3,987. The places engines actually draw on are rarely the ones a traditional PR plan targets — the full source breakdown is published with its data.
Re-check rather than assume. Liu, Zhang and Liang’s work on verifiability in generative search engines found engines frequently state things their citations do not support, so being cited does not guarantee being described correctly. The only way to know what an engine says about you is to ask it, repeatedly.
Common questions
Can you pay to appear in AI Overviews?
Not organically. Google's documentation states that appearing in AI Overviews requires a page to be indexed and eligible for a snippet, with no additional technical requirements and no special markup. There is no organic submission or opt-in.
Does schema markup help you appear in AI Overviews?
Google's own documentation says no special schema.org structured data is needed. Markup remains useful for how a page is parsed and displayed, but Google has published a direct denial that it is the mechanism for AI feature eligibility.
How many sources does an AI Overview cite?
In Kunzum's September 2026 index, Google AI Overviews cited an average of 7.7 sources per answer across 120 answers in edition one, and 7.1 across 270 in edition two. That is denser than ChatGPT or Claude and sparser than Perplexity.
Can you opt out of AI Overviews without leaving Search?
Partly. Google documents Google-Extended as a control over training and grounding in some of its other systems, separately from Search indexing. Blocking Googlebot removes a page from Search entirely, including from the results AI Overviews draws on.
Sources
- Google Search Central, “AI features and your website”. Google. Checked 2026-09-12.
- Liu, Zhang and Liang, “Evaluating Verifiability in Generative Search Engines” (2023). arXiv:2304.09848. Checked 2026-09-12.
- Aggarwal and colleagues, “GEO: Generative Engine Optimization” (2023). arXiv:2311.09735. Checked 2026-09-12.
- Lewis and colleagues, “Retrieval-Augmented Generation for Knowledge-Intensive NLP Tasks” (2020). arXiv:2005.11401. Checked 2026-09-12.
Published 2026-09-12. Written by Narender Charan, who runs Kunzum.