How Google Picks AI Overview Supporting Links

Google's AI features documentation states the rule directly: "To be eligible to be shown as a supporting link in AI Overviews or AI Mode, 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."

A supporting link is a cited page shown alongside the generated answer. The rule sets a floor, not a ceiling. The same documentation adds that "indexing and serving isn't guaranteed." Meeting the requirements puts your page in the pool. It does not put it in the answer.

Google also rules out a separate optimization track: "There are no additional requirements to appear in AI Overviews or AI Mode, nor other special optimizations necessary." As of October 2026, that page (last updated December 2025) is still Google's most direct statement on the topic. Our digest of Google's AI features documentation covers the rest of it.

The practical reading: if a page can't appear in normal Search with a snippet, it can't be an AI Overview source. Every diagnosis starts there.

Google says its AI features may use a "query fan-out" technique, issuing multiple related searches across subtopics. The sources in one AI Overview can therefore come from several searches, not only the one the user entered. Read the definition of query fan-out for the mechanics, and how AI Mode uses fan-out for the sibling surface.

Three things follow:

• Your page can be cited for a subtopic. A page about one part of a question can be pulled in by a fan-out search the user never saw.
• Nobody outside Google sees the fan-out queries. Google doesn't publish them, and Search Console doesn't show them. Any tracker running a fixed prompt list is measuring a proxy for that hidden input.
• AI Overviews often don't trigger at all. Google also says AI Overviews and AI Mode "may use different models and techniques" and can return different links. Track the two separately.

The next move is to think in subtopics, not keyword variants. Google says you don't need to chase every long-tail variation. The research that agrees on anything points to topical relevance as one of the most reproducible factors.

Ahrefs published an analysis on March 2, 2026 that found 37.9% of URLs cited in AI Overviews also appeared within the first 10 blocks of Google results. Counting organic results only, Ahrefs put the figure at 37.1%. The rest split roughly evenly: about 31% from positions 11 to 100 and about 31% from beyond position 100.

Read it with three caveats:

• It is an association. It shows overlap between rankings and citations. It does not show that ranking causes citation.
• Matching is URL-level. A large site cited through a URL that doesn't itself rank counts as no overlap, which penalizes big sites.
• It is not a trend. Ahrefs changed its parsing method from its earlier study, so comparing the two figures as a decline mixes two different instruments.

What it means for you: ranking well puts you near the pool, but in this dataset most cited URLs came from outside the top 10. Neither "rank top 10 and you will be cited" nor "rankings no longer matter" is supported.

Every documented control on Google's side works by removing a page, not by promoting it. These are the ones to know:

• robots.txt rules for Googlebot. There is no robots.txt token for AI Overviews or AI Mode. Google states that robots.txt directives for Googlebot are the control, because AI is built into Search.
• noindex. A page out of the index can't be a supporting link.
• nosnippet, max-snippet and data-nosnippet. Google documents these as the on-page controls for limiting how content appears, including in AI features. data-nosnippet works on a section of a page. These often arrive by accident, inherited from a template someone used to hide a cookie banner.
• The Search Console AI features control. Inclusion is opt-out: it is the default for all properties. It covers AI Overviews, AI Mode and generative AI features in Discover. It is still rolling out to a subset of owners, and child properties inherit from the parent by default. Our guide to Google's AI features control walks through it.

One token people confuse: Google-Extended. Google says it does not affect a site's inclusion in Google Search and is not a ranking signal. It governs training and grounding for Gemini Apps and Vertex AI. Blocking it won't remove you from AI Overviews, and allowing it won't add you.

Rendering is less of a risk here than elsewhere. Googlebot renders JavaScript, unlike most other AI crawlers, so client-side content is generally reachable for Google.

Google is clear about eligibility and silent about selection. As of October 2026, it has not published:

• How it chooses among eligible pages. There is no list of factors for supporting link selection.
• Any position data. Search Console's generative AI performance report shows impressions. It shows no clicks, no click-through rate and no position. A tool that reports your "AI Overview rank" has no ground truth to check that number against.

Google has also said what you don't need. Its guide to generative AI features says you don't need new machine-readable files, AI text files, markup or Markdown, because Google Search doesn't use them. That covers llms.txt. Google also says you don't need to write in a special way for AI and that there's no ideal page length.

On structured data, Ahrefs ran a matched test of 1,885 pages that added schema and found AI Overview citations fell 4.6%. Every page in that test was already heavily cited, so whether schema helps a page enter the pool is untested. Existing markup does no harm. It just isn't a lever for AI answers.

For how this compares across engines, see how AI search picks sources.

Work from the gates you control to the parts you can only observe.

1. Confirm the page is indexed. Use URL Inspection in Search Console. A page Google hasn't indexed can't be cited.
2. Check snippet directives. Look for noindex, nosnippet, max-snippet:0 and data-nosnippet in templates and headers. Make sure any you find are intentional.
3. Check robots.txt for Googlebot. A rule meant for another crawler can catch Googlebot.
4. Check the Search Console AI features control. If your account has it, confirm the property and its parent haven't opted out.
5. State key facts plainly in the HTML. A specific price, number or one-line definition gives the engine a sentence it can quote.
6. Look at who does get cited. When AI Overviews cite other sites for your topic, being named on those sites is the route into the answer. That is an association, not a promised effect.

Then observe. Run your buyer queries and note which trigger an AI Overview and which sources appear. The AI Overview Checker does this for a query. Results vary by query, location and day, so treat any single check as a snapshot.

Can I keep my page in Google Search but out of AI Overviews? Partly. There is no robots.txt token for AI Overviews. Google documents snippet controls such as nosnippet, max-snippet and data-nosnippet, but these also limit regular search snippets. The Search Console AI features control lets a property opt out of AI Overviews, AI Mode and Discover generative features while staying in Search. As of October 2026 it is still rolling out, so not every account has it yet.

Why isn't my top-ranking page cited in the AI Overview? Google says indexing and serving isn't guaranteed, and it doesn't publish how it chooses sources among eligible pages. Fan-out searches may surface pages for subtopics your page doesn't cover. Ahrefs' March 2026 data shows rankings and citations overlap only partly: 37.9% of cited URLs appeared in the top 10. A strong ranking makes you eligible and nearby. It doesn't secure a citation.

Do I need schema or an llms.txt file to be an AI Overview source? No. Google's guide to its generative AI features says you don't need new machine-readable files, AI text files or markup, because Google Search doesn't use them. Ahrefs' matched test found AI Overview citations fell 4.6% after pages added schema. Keeping existing structured data does no harm, but it isn't a documented route into AI Overviews.

Can I see which AI Overviews cite my site? Only partially. Search Console's generative AI performance report shows impressions for AI Overviews and AI Mode, with no clicks, click-through rate or position. It also doesn't show the fan-out queries Google ran. To see which sources appear for a specific query, you have to run that query and inspect the result, knowing it varies by location, language and time.

Does blocking Google-Extended remove me from AI Overviews? No. Google states that Google-Extended does not affect a site's inclusion in Google Search and is not used as a ranking signal. It controls whether your content is used for training and grounding in Gemini Apps and Vertex AI. For AI Overviews, the relevant controls are Googlebot's robots.txt rules, noindex, snippet directives and the Search Console AI features control.