How to rank in AI Overviews: what Google actually uses
Short answer
There’s no separate ranking system to game for AI Overviews. Google says a page only needs to be indexed and eligible to show a snippet in normal Search, and that no special files, markup or schema are required. AI Overviews and AI Mode then run several related searches (query fan-out) and cite pages that answer those sub-questions well. So the work is ordinary SEO done thoroughly: get indexed, don’t block your snippets, answer each sub-question plainly near the top of a section, and track citations by hand alongside the Web search type in Search Console.
Ranking in AI Overviews means being one of the linked sources inside Google’s AI summary. You earn it much like a strong organic listing, with one twist: Google searches for the pieces of an answer, not just the whole query.
This guide is for owners and marketing leads in the US, the UK and Australia who rank well but see competitors cited instead. Demand for this question is real. Semrush (October 2026) shows about 720 US searches a month for “how to rank in ai overviews”, with 110 each in the UK and Australia.
If you want the ChatGPT side, we cover it in how to rank in ChatGPT. For how the two disciplines relate, read GEO vs SEO. This piece sticks to Google.
How AI Overviews choose their sources
AI Overviews pick sources from Google’s normal index, using pages that are allowed to show a snippet. That’s the whole published entry requirement.
Google’s AI features and your website page says it plainly: “There are no additional requirements to appear in AI Overviews or AI Mode, nor other special optimizations necessary.” To be eligible as a supporting link, “a page must be indexed and eligible to be shown in Google Search with a snippet”.
The same page says you don’t need “new machine readable files, AI text files, or markup”, or special schema. Google doesn’t publish how it picks between eligible pages, so anyone quoting a precise formula is guessing.
What query fan-out means for your pages
Query fan-out means Google breaks one question into several related searches, then builds the answer from what those searches return. Google describes it as “issuing multiple related searches across subtopics and data sources” and says it produces “a wider and more diverse set of helpful links” than a classic results page.
So your page can be cited for a sub-question even when it doesn’t rank for the head query.
Take “kitchen renovation cost Melbourne”. The answer may also draw on labour rates, permits and timelines. A page that clearly answers the permit question has a shot at a citation, even if a big publisher owns the main ranking.
So map the sub-questions first, from People Also Ask and the follow-ups your sales team hears. Then give each one a section that opens with a direct answer.
Snippet controls: how pages lock themselves out
Your snippet settings decide whether Google can use your text in AI Overviews at all. This is where we find the most self-inflicted damage.
Google’s robots meta tag documentation says nosnippet applies to AI Overviews and AI Mode and “will also prevent the content from being used as a direct input” for them. max-snippet “will also limit how much of the content may be used as a direct input”.
| Control | What it does in normal results | Effect on AI Overviews and AI Mode |
|---|---|---|
noindex |
Page is dropped from Search | Not eligible at all |
nosnippet |
No text snippet shown | Content can’t be used as a direct input |
max-snippet:[number] |
Caps snippet length | Limits how much text can be used as input |
data-nosnippet (on an element) |
Hides that element from snippets | Google’s AI features page lists it as a way to limit what’s shown |
| Google-Extended in robots.txt | No effect on Search | Google says it doesn’t affect Search inclusion or ranking |
That last row matters. Google’s crawler documentation says “Google-Extended does not impact a site’s inclusion in Google Search nor is it used as a ranking signal in Google Search.” Blocking it is about Gemini model training, not AI Overviews.
In our experience, the usual culprit is a max-snippet:0 or sitewide nosnippet left over from a legal review or an old plugin. Check the rendered source, not just the CMS, because themes add their own tags.
What actually helps you get cited
Once you’re eligible, cited pages tend to answer a specific question more directly than the alternatives. That’s our reading of a system that lifts passages to build a summary, not a Google formula.
Here’s what we work on, in order:
- Indexing. Check the page in Search Console’s URL Inspection tool and make sure key content is in the rendered HTML.
- Answer-first sections. The first two or three sentences under each heading should answer it without needing the rest of the page.
- Sub-question coverage. One strong page answering five follow-ups beats five thin pages.
- Specifics. Prices, timeframes, named standards and real numbers give Google something worth quoting.
- Visible authorship and dates on pages where trust matters, such as health, money and legal topics.
- Structured data that matches what’s visible on the page. Google says it isn’t required for AI features, but it helps in normal Search.
Our view: AI Overviews reward the page a good salesperson would hand a prospect. If your page makes a buyer scroll past three paragraphs of brand story to find the price range, it’s not the page Google will quote.
A worked example: a UK conveyancing firm
This is a hypothetical, built from patterns we see.
A conveyancing firm in Bristol ranks fifth for “how long does conveyancing take”. The AI Overview cites three other sites and not them.
Step one is the snippet check. Their page source has max-snippet:50, set years ago. That alone caps what Google can use. They remove it.
Step two is fan-out coverage. The AI Overview mentions leasehold delays and chains. Their page says nothing on leasehold, so they add a section that opens with a direct answer and a range they can stand behind from their own files.
Step three: the page opens with “We’re a friendly, award-winning team”. They replace that with a two-sentence answer.
None of this guarantees a citation. It removes reasons not to cite them.
How to track AI Overview citations
Track AI Overviews two ways: Search Console for traffic, manual checks for citations.
Search Console counts AI Overview clicks inside the normal Web search type. There’s no separate AI Overview filter. Google’s Performance report help explains the counting: clicking an external link in an AI Overview counts as a click, standard impression rules apply, and “all links in the AI Overview are assigned that same position”. For AI Mode, position follows the same method as a normal results page.
So an AI Overview citation can show a high average position with a modest click-through rate. It looks odd until you know why.
For the citations themselves, keep a log:
- Pick 15 to 30 queries that matter commercially, including the sub-questions.
- Check them monthly from the countries you sell in. AI Overviews vary by location and change between checks.
- Record whether an AI Overview appears, who’s cited, and whether you are.
Our free AI visibility check runs 10 fixed questions across Google AI Overviews, ChatGPT, Perplexity and Gemini if you’d like a baseline without building the log yourself.
Common AI Overview myths
Most bad advice here sells a new product for an old job. Google’s own documentation contradicts all of these.
“You need an llms.txt file.” Google says no AI text files are needed. We cover where llms.txt does and doesn’t matter in our llms.txt guide.
“You need special schema for AI Overviews.” Google says no special structured data is required.
“Blocking Google-Extended removes you from AI Overviews.” Google says Google-Extended doesn’t affect Search inclusion.
“Mass AI-written pages will cover every sub-question.” Google’s spam policies name “using generative AI tools or other similar tools to generate many pages without adding value for users” as scaled content abuse.
“Position one guarantees a citation.” Fan-out means Google pulls from several searches, so a lower-ranked page that answers a sub-question better can be cited instead.
Quick answers
What should I fix first?
Check every key page for noindex, nosnippet and low max-snippet values in the rendered source. It’s the fastest way to go from ineligible to eligible.
Should I hire someone for AI Overview SEO?
If your pages are eligible and still not cited, the remaining work is content and authority over several months. Our SEO and GEO retainers for international clients start from US$1,500 a month, with a 6-month minimum.
Will AI Overviews kill my organic traffic?
It depends on your queries. Track clicks and impressions by query in Search Console and judge from your own data.
We run the same playbook on Arché Academy, Glen’s own tuition centre, before we use it on clients. If your site isn’t showing in Google at all, start with why your website isn’t ranking. For the wider picture of AI search beyond Google, read our guide to LLM SEO. Clients in the US, the UK and Australia can see how we work internationally or get in touch for a free consultation.
Google’s AI features change often. Documentation quoted here was checked in October 2026.