Does Google-Extended affect my rankings?

Google states it does not affect inclusion or ranking in Google Search. It only governs Gemini training and grounding use.

Can I allow grounding but block training?

Google-Extended covers both, and there is no documented token that separates them. It is one decision.

How do I know if an answer was grounded?

Grounded answers show sources or links. An answer with no sources came from the model alone.

Engine guide: Gemini

How to get cited by Gemini

The Gemini app sits outside Google Search, and its controls work differently. A site can be fully visible in AI Overviews and still be unavailable to Gemini because of one robots.txt token.

The short answer

Gemini grounds answers in Google Search results, so your page must be indexed by Googlebot. Unlike AI Overviews, Gemini apps also honour the Google-Extended token: Google documents that disallowing it stops content being used for Gemini training and for grounding. To be cited by Gemini, keep Google-Extended allowed and make pages directly answer the question.

What ranked on 26 September 2026: Guides ranking for this query mostly copy advice written for AI Overviews. Almost none mention that Google-Extended affects Gemini grounding, which is the one control that behaves differently between the two products.

What is specific to Gemini

Gemini is a standalone assistant that decides for itself when to search. Its link display and its opt-out rules both differ from Search.

Google-Extended is a robots.txt product token, not a separate crawler. Googlebot does the fetching, and the token tells Google whether Gemini may use what it fetched
Disallowing Google-Extended affects Gemini apps and the Vertex AI Gemini API, including grounding, but not Google Search or AI Overviews
Gemini does not search for every prompt. Questions about recent events, local businesses or prices are the ones that reliably trigger grounding
Sources are shown through a sources panel and inline links, and the double-check feature can highlight statements against search results
Gemini is also embedded in Workspace and Android, where it may answer from a user's own files before the web

Things that sound right and are not

ClaimWe blocked Google-Extended to protect our content and nothing changed.
Nothing changed in Search, which is expected. The loss shows up in Gemini, where your pages can no longer ground answers. Many sites made this trade without realising it.
ClaimGemini and AI Overviews are the same thing.
They share a model family and an index, not controls or triggers. Test Gemini separately, with the same prompts you use elsewhere.
ClaimGemini names brands from training, so pages do not matter.
Ungrounded answers do come from training, and they carry no links. Grounded answers are where citations come from, and those read live pages.

Where to start

01Check robots.txt for a Google-Extended rule and decide deliberately whether grounding access is worth more to you than the training opt-out.
02Confirm key pages are indexed in Search Console, since grounding depends on the Google index.
03Write the pages Gemini would ground on, such as pricing, location, hours and comparisons, with the facts in plain text near the top.
04Test prompts that force grounding, for example ones that ask about this year or a specific city, and log which URLs appear.
05Recheck after any robots.txt change, because the effect only shows on grounded answers.

Questions

Does Google-Extended affect my rankings?: Google states it does not affect inclusion or ranking in Google Search. It only governs Gemini training and grounding use.
Can I allow grounding but block training?: Google-Extended covers both, and there is no documented token that separates them. It is one decision.
How do I know if an answer was grounded?: Grounded answers show sources or links. An answer with no sources came from the model alone.

Doing this yourself and want a second opinion on what the assistants say about you? Tell me what you are seeing, or read the AEO Playbook for the whole framework.