AI search guide · Gemini
How Gemini finds and cites sources.
What Google documents about grounding Gemini in Google Search, the control that decides whether your pages can be used, and what checked studies say it cites.
Zach ChmaelChecked September 30, 2026 · from Google’s documentation and checked studies
Free AI visibility check
The short answer
Gemini, in brief.
The Gemini app can ground its answers in Google Search: it’s given content from the Search index at the moment someone asks. Whether your pages can be used that way is decided by one robots.txt token, Google-Extended, which also governs training future Gemini models. In Profound’s data, 69% of Gemini’s citations went to company websites, against 47% for ChatGPT.
Sources
Where it finds sources
Google-Extended decides whether content Google crawls from your site may be used for grounding in Gemini Apps: providing content from the Google Search index to the model at prompt time, to improve factuality and relevancy.
Google Search Central, Google’s common crawlers · read September 30, 2026The same token governs whether that content may be used to train future Gemini models. Google-Extended has no crawler of its own: crawling is done by Google’s existing crawlers.
Google Search Central, Google’s common crawlers · read September 30, 2026Google says Google-Extended doesn’t affect a site’s inclusion in Google Search, and isn’t used as a ranking signal there.
Google Search Central, Google’s common crawlers · read September 30, 2026
Crawlers
The crawlers to know
| User agent | What it does | robots.txt |
|---|---|---|
Googlebot | Crawls for Google Search, whose index Gemini grounds its answers in | Obeys it. Blocking it keeps Google from crawling your pages for Search, the index Gemini grounds in. |
Google-Extended | A robots.txt token for Gemini training and grounding in Gemini Apps | Obeys it. Disallow it and Google won’t use your content for either; your place in Search is unaffected. |
Source: Google Search Central, Google’s common crawlers · read September 30, 2026
If you want Gemini to be able to ground answers in your pages, leave Google-Extended allowed: one token covers both grounding and training, and Google’s documentation describes no way to allow one and refuse the other.
Google Search Central, Google’s common crawlers · read September 30, 2026Citations
What it tends to cite
- Vendor study47% vs 69%
ChatGPT drew 47% of its citations from company websites, the lowest share of the major engines Profound tracked; on Gemini, company websites made up 69%.
Company websites here are any company’s own site, competitors’ included, on the prompts Profound tracks.
Profound, Where do AI citations come from? · Jul 30, 2026
11.84 billion citations from ChatGPT, Claude, Google AI Mode, AI Overviews, Gemini, Grok, Copilot and Perplexity, across 8,061 active Profound categories, April 16–July 16, 2026.
Stability
How much its answers change
- Preprint0.3–0.5
In repeated runs of the same query, the sites two answers cited overlapped by a median of about 0.3 on Gemini, 0.33–0.40 on SearchGPT and 0.5 on Perplexity, on a scale where 1 means the same sites.
Measured by domain (the Jaccard index). The author works at IQRush, which sells measurement verification for AI search.
Ronald Sielinski, arXiv, Quantifying Uncertainty in AI Visibility: A Statistical Framework for Generative Search Measurement · Mar 9, 2026
Repeated queries on three consumer topics through the Perplexity, SearchGPT and Gemini APIs: daily for nine days, plus a run every ten minutes.
First steps
What to fix first
- 01
Leave Google-Extended allowed if you want Gemini to be able to ground answers in your pages.
- 02
Be in Google Search first: Gemini grounds in the Search index, so a page that isn’t indexed can’t be used.
- 03
Keep your company’s own pages complete and current: company websites made up 69% of Gemini’s citations in Profound’s data.
- 04
Measure with enough questions: in one study, measuring a site’s share of Gemini’s citations to within five points took about 40–50 queries.
Your company
See what AI says about you
Trovance sweeps Gemini for your buyer questions every week on Growth and Agency. See the plans
Go deeper
Read next
- Where Do AI Citations Come From Across Industries?Profound classified a proprietary 11.84-billion-citation corpus across eight models, with useful patterns and a hard causal limit.SEO/GEO8 min read
- Blocking AI Crawlers Is a Policy Choice, Not a Citation SwitchA purpose-specific worksheet for service businesses choosing search access, training preferences, and user-requested visits.GEO11 min read
- How to Track AI Citations Across ChatGPT, Perplexity, and ClaudeOne answer is noise. A frozen question set, a few phrasings run weekly on every engine, and a log that separates mention from citation turn it into a measurement.SEO/GEO11 min read
FAQ
Questions, answered.
How does Gemini decide what to cite?
The Gemini app can ground answers in Google Search, drawing content from the Search index when someone asks. Google’s page on its crawlers doesn’t say how Gemini chooses among those pages; in Profound’s data, 69% of Gemini’s citations went to company websites.
Does blocking Google-Extended hurt my Google rankings?
No. Google says Google-Extended doesn’t affect inclusion in Google Search and isn’t used as a ranking signal. It decides whether Gemini can train on your content and ground answers in it.
Is Gemini the same as AI Overviews?
No. AI Overviews and AI Mode are features of Google Search, governed by Googlebot and snippet controls. The Gemini app is separate, and whether it can ground answers in your content is governed by Google-Extended. The Google profile covers Search’s AI features.
Rather ask a person? Text us.
Get understood. Get cited. Get chosen.
Text Trovance
Or, in iMessage, send this note to +1 (839) 205-9508:
Hi Trovance, show me how AI sees my company.
You choose to send it. Trovance replies, and sends the updates you ask for, over iMessage through our messaging provider. Reply STOP anytime. Tapping or sending never bills you.