Gemini SEO
Gemini SEO is mostly a single decision, and it is made in robots.txt. Google-Extended is the token that governs whether content Google has already crawled may be used for Gemini, and everything else is ordinary page quality.
It is also the most misunderstood control in the category, because it does not behave like a crawler at all. Getting the mental model right saves you from both of the common mistakes: blocking it by accident, and expecting to see it in your logs.
SEO Health
Good
AI Readiness
Good
Entity: established
Google-Extended is a control token, not a crawler
Google's documentation is explicit that Google-Extended has no separate user-agent string of its own. Crawling is done with existing Google user agents, and the robots.txt token is used purely in a control capacity.
Two practical consequences follow. You will never find Google-Extended in a server log, so its absence there tells you nothing. And there is no extra crawl load to weigh — allowing it does not add a single request to your origin, because the fetch has already happened as Googlebot.
What allowing it does, and what it does not
Google-Extended governs two things at once, which is the detail most summaries get wrong. It controls whether your content may be used for training future Gemini models, and separately whether it may be used for grounding — supplying content from the Google Search index to the model at prompt time to improve factuality.
Disallowing it therefore has a cost beyond training. It also withdraws your pages from the grounding that makes Gemini answers factual and current. What it does not touch is Google Search: Google states plainly that Google-Extended does not affect a site's inclusion in Search and is not used as a ranking signal.
| If you set | Gemini training | Gemini grounding | Google Search |
|---|---|---|---|
| allow (or say nothing) | Content may be used | Content may be used | Unaffected |
| disallow | Content withheld | Content withheld | Unaffected — no ranking impact |
Do not confuse it with the Vertex agent crawler
Google publishes a separate token, Google-CloudVertexBot, which crawls sites on a site owner's own request when they are building Vertex AI Agents. Google documents that it has no effect on Google Search or other products.
The two are unrelated, and a rule written for one does nothing for the other. If your intention is to control Gemini, Google-Extended is the token; if you are not building Vertex agents, the second one is simply not your concern.
- Google-Extended — governs Gemini training and grounding. No user agent, robots.txt only.
- Google-CloudVertexBot — crawls for Vertex AI Agents at the site owner's request. No Search effect.
- Googlebot — the ordinary Search crawler. Blocking it removes you from Google entirely, which is a different and much larger decision.
The page work is the same as everywhere else
Once access is settled, Gemini rewards what every answer engine rewards: a page whose claims can be lifted cleanly and attributed with confidence.
That means a direct answer under each heading, structured data declaring what the page is, enumerable content in lists and tables, visible dates, and citations to primary sources. None of it is Gemini-specific, which is the point — the work compounds across engines rather than being spent on one.
How the score is calculated
The score is the result of 38 published checks run against your page as a crawler receives it, each with a fixed weight and a public source.
- SEO Health is 100 points across six categories: indexability and crawlability, on-page elements, content depth and quality, delivery signals, security and trust, and link architecture.
- AI Readiness blends page readiness (60%) with entity authority (40%): how easily a passage can be quoted, and whether AI engines can tell who the brand is.
- Every check, weight and threshold is published on the methodology page.
Frequently asked questions
- Will blocking Google-Extended hurt my Google rankings?
- No. Google states that Google-Extended does not impact a site's inclusion in Google Search and is not used as a ranking signal. It is a separate control that governs Gemini training and grounding only, so the Search consequence of disallowing it is nothing at all.
- Why can't I see Google-Extended in my server logs?
- Because it has no user-agent string. Google documents that crawling is performed with existing Google user agents and the robots.txt token is used only in a control capacity. Looking for it in logs will always come up empty, whatever your setting.
- Does allowing Google-Extended increase crawl load on my server?
- No. The fetch has already happened as Googlebot; the token governs what may be done with content Google already holds. Allowing it adds no additional requests, which removes the usual performance argument for blocking a crawler.
- Is Google-Extended the same as the AI Overviews control?
- No, and conflating them is a common error. Google-Extended is documented as governing training and grounding for Gemini Apps and the Vertex API. It is not described as a control over inclusion in Google Search, which is where AI Overviews appear.
Check your own page
Paste any URL for its SEO and AI Readiness scores and the first fix, in about 15 seconds. No account needed.
Read next
Sources
- Google Search Central — Google common crawlers
- Google Search Central — Introduction to robots.txt
- schema.org — Getting started
Last reviewed .