How to Rank in Google AI Overviews: What Google Actually Says

To rank in Google AI Overviews, a page has to clear the same bar as regular search (indexed, crawlable, eligible to show a snippet) and then be the best source for one of the sub-questions Google's AI searches for while building its answer. Google's documentation says there is no separate AI Overview ranking system you can opt into and no special markup required. The practical levers are coverage of a topic's sub-questions, passages that answer those questions directly, and content with information other pages lack.

This guide is built on Google Search Central's AI features documentation and its May 2025 guidance, plus what follows mechanically from how AI Overviews retrieve sources.

How does Google AI Overview work?

An AI Overview is a generated summary shown at the top of some Google results, with links to the web pages that support it. Google began rolling them out to all US users in May 2024, powered by what the I/O 2024 announcement called "a new Gemini model customized for Google Search". AI Mode, a separate conversational search tab, uses the same underlying approach for longer questions.

The core mechanism is query fan-out. Google's documentation says AI Overviews and AI Mode may issue multiple related searches across subtopics and data sources to develop a response. Google's AI Mode announcement describes it as breaking a question into subtopics and issuing many queries at once on the user's behalf.

In practice, a query like "best CRM for a 5-person real estate team" can fan out into searches such as "CRM for real estate agents", "CRM pricing small team", "real estate CRM MLS integration" and "HubSpot vs Follow Up Boss". The model reads top results for those searches, writes the answer, and attaches supporting links. Google says this lets it show a wider and more diverse set of links than a classic results page.

Three things follow from that design:

1. You compete on sub-queries, not only the head query. A page that ranks well for one sub-question can be cited even if it is nowhere on page one for the query the user typed.
2. Passage quality matters. The model needs a chunk of text that answers a narrow question. A page with the right answer buried in paragraph nine is harder to use.
3. Normal ranking signals still gate entry. The fan-out searches run against Google's index with Google's ranking systems. Weak pages don't get retrieved.

What does Google say you need (and don't need)?

Question	Google's answer	Source
Is there an extra technical requirement?	No. The page must be indexed and eligible to be shown in Search with a snippet.	AI features and your website
Do I need special optimization?	No additional requirements or special optimizations are needed.	Same page
Do I need new files (like llms.txt) or AI markup?	No new machine-readable files, AI text files or markup needed.	Same page
Do SEO best practices still apply?	Yes, they remain relevant for AI features.	Same page
What kind of content does well?	Unique, non-commodity content that satisfies people.	Google Search Central blog, May 2025
Where do I see the traffic?	In Search Console's Performance report, within the "Web" search type.	AI features page
How do I limit what's shown?	nosnippet, data-nosnippet, max-snippet or noindex.	AI features page

That table is the honest core of "AI overview SEO". Anyone promising a secret AI Overview ranking factor is guessing beyond what Google documents.

How to rank in AI Overviews: step by step

1. Confirm the page is eligible

Check each target page for:

• HTTP 200 and indexable content, with Googlebot allowed in robots.txt
• Indexed status in Search Console's URL Inspection
• No nosnippet robots meta, and no max-snippet:0. Google's snippet controls documentation explains how these work, and its May 2025 guidance says more restrictive settings limit how your content is featured in AI experiences.
• Key text rendered in HTML, not only inside images or behind interactions

Note that Google's crawler documentation says Google-Extended does not affect inclusion in Google Search. Blocking it won't remove you from AI Overviews; allowing it won't add you.

2. Map the fan-out for your topic

List the sub-questions an AI Overview would need to answer for your target query. For a commercial query, that usually includes definitions, criteria, pricing, comparisons, alternatives, and "for [audience]" variants. Our free query fan-out tool generates a likely set. Then check which URLs rank for each sub-question today. Where a competitor ranks for four of six sub-questions and you rank for one, the AI Overview has more reasons to lean on them.

3. Write passages that answer one question each

Structure pages so each section can stand alone:

• A question-style H2 that mirrors a sub-query ("How much does a real estate CRM cost?")
• A direct answer in the first two to four sentences under it, with the number, the condition and the exception
• Supporting detail, examples and a table after that

This is ordinary good writing, but it matters more when a model is choosing a passage rather than a reader skimming a page. The heading analyzer helps check whether your headings describe the questions you answer.

4. Add information other pages don't have

Google's May 2025 guidance asks for "unique, non-commodity content". If ten pages say the same thing, the model can cite any of them, and your odds are low. Things that set a source apart:

• Original numbers: your own pricing, test results, benchmarks, survey data with method stated
• First-hand detail: screenshots described in text, setup steps you ran, edge cases you hit
• Clear comparisons with named options and criteria
• Up-to-date facts where competitors are stale

The GEO research paper (Aggarwal et al., KDD 2024) found a related effect in its test setup: adding statistics, quotations and citations to credible sources raised how visible a source was in generated answers, while keyword stuffing did little.

5. Keep structured data accurate

Structured data won't get you into AI Overviews on its own, but it helps Google read facts like product, price, rating, organization and author. Google's guidance is to make sure all marked-up content is also visible on the page and to validate it. Use the schema markup generator and schema validator to check yours.

6. Cover page experience, media and business data

From the same Google guidance: a page that is cluttered or slow disappoints visitors from AI results as much as from blue links; support text with high-quality images and video; and keep Merchant Center and Business Profile information current, because multimodal and local answers draw on them.

7. Link the cluster together

Internal links help Google find and understand every page in a topic cluster. If your "pricing" and "vs" pages are orphaned, they're less likely to be retrieved for the sub-queries they answer.

AI Overview SEO vs ranking in blue links

	Blue-link ranking	AI Overview supporting link
What's evaluated	The whole page for one query	Passages across pages for many fan-out queries
Entry requirement	Indexed	Indexed and snippet-eligible
Competitive set	Pages ranking for the query	Pages ranking for any sub-query
Win condition	Position	Being used and linked in the summary
Stability	Fairly stable	Changes by query wording, location and over time
Reporting	Search Console, by query and page	Mixed into Search Console "Web" totals; no separate AI Overview filter is documented

The main practical difference: you optimize topic coverage and passage clarity, and you measure presence across a prompt set instead of one rank.

How to track AI Overview visibility

Google counts AI Overview and AI Mode clicks and impressions inside the regular Performance report under the Web search type. It doesn't document a separate filter for them, so you can't isolate AI Overview traffic from Search Console alone.

Google also says clicks from results pages with AI Overviews are "higher quality", meaning users tend to spend more time on the site. That makes conversions and engaged sessions better success metrics than raw click counts.

To see whether you're cited, you need to check the results themselves: run your target queries, record whether an AI Overview appears, which domains it links, and whether your brand is named. Do it on a schedule, because AI Overviews change with query wording, location and time. Our AI Overview tracker does this with SERP data and also covers AI Mode (aeotime is our product). If you track other engines too, AI visibility tracking puts Google and ChatGPT results side by side.

What doesn't help

• AI-specific files or markup. Google says none are needed. An llms.txt file won't change AI Overview eligibility; see what is llms.txt.
• Keyword repetition. Fan-out rewrites the query anyway, and research on generative engines found keyword stuffing ineffective.
• Mass-produced pages. Google's guidance on generative AI content says generating many pages without adding value for users may violate its scaled content abuse policy.
• Blocking or allowing Google-Extended to influence Search. It doesn't affect Search inclusion.

Checklist for one target query

• Page indexed, returns 200, Googlebot allowed, no restrictive snippet settings
• Sub-questions listed; you have a section or page ranking for each
• Each section opens with a two to four sentence direct answer
• At least one piece of information competitors don't have
• Structured data valid and matching visible content
• Internal links connect every page in the cluster
• AI Overview presence and cited domains logged weekly

Start with the free query fan-out tool to see which sub-questions your pages need to cover.