To get cited by Gemini, be indexed and competitive in Google Search for the short queries Gemini runs, leave Google-Extended allowed in robots.txt, and get into the few sources each answer uses. Google says Gemini Apps ground answers in content from its Search index, and in our scans Gemini cited fewer sources per answer than any other engine, about three and a half.
How does Gemini find sources?
Gemini's web sources come from Google Search. Google's crawler documentation describes grounding in Gemini Apps as providing content from the Google Search index to the model at the moment someone asks. There is no separate Gemini crawler to allow: Google crawls with its existing user agents.
The Gemini-specific control is a robots.txt token, Google-Extended. It decides whether content Google crawls from your site may be used to train future Gemini models and to ground answers in Gemini Apps and in Grounding with Google Search on Vertex AI. Google says it has no effect on inclusion or ranking in Google Search. So a site can block Google-Extended, stay in Google Search, including AI Overviews, and still drop out of Gemini's grounded answers.
| Control | Gemini Apps | Google Search, including AI Overviews and AI Mode |
|---|---|---|
| Googlebot allowed in robots.txt | Needed for your pages to be in the Search index Gemini draws on | Needed to be crawled for Search |
| Google-Extended disallowed | Your content is not used for grounding answers or for training future models | No effect on inclusion or ranking |
| Search generative AI control set to exclude (Search Console) | Not listed among the features it covers | Removes your links and content from AI Overviews, AI Mode and Discover's AI features |
Google's developer documentation shows the grounding step in more detail. With the Google Search tool switched on, the model first judges whether a search would improve its answer. If so, it writes one or more search queries, runs them, and composes an answer in which each cited passage is linked to the web page it came from.
In the Gemini apps, sources appear behind a Sources button or inline in the response. Google notes that not every response includes sources, and that when Gemini quotes a large amount of text from a web page, the page appears in the sources list. One more warning from Google's help center is useful: Gemini can misrepresent how it works, including how it cites sources. Test what it does rather than asking it.
What did Gemini cite in our scans?
Our scans cover the questions Cituna tracks about AI visibility software: 33 scans in the 30 days to 22 September 2026 (how we count). Gemini gave 552 answers containing 1,910 citations to third-party sites, about 3.5 per answer, the fewest of the seven engines. Perplexity averaged 20.2.
Its five most-cited sites were llmpulse.ai (111 citations), aiclicks.io (61), geoptie.com (61), reddit.com (55) and writesonic.com (53). Four of the five sell AI visibility software, according to their own homepages; the fifth is Reddit.
Set aside the vendors Cituna tracks as competitors and the engines' own pages, and look at the 100 third-party sites Gemini cited most. Roundups took 75% of their 533 citations. Articles took 12%, community sites 11% (almost all of it Reddit), and review sites 2%. The leading roundups were on writesonic.com (53), zapier.com (39) and integrate.io (27).
Gemini is a Google product, but it did not cite like Google's AI Overviews. YouTube, the site Google AI Overviews cited most in our scans (371 times), drew one citation from Gemini.
Names were a weak point. For "best Otterly AI alternative", Gemini cited voicescriber.com and voicedash.ai 20 times between them, for pages about alternatives to Otter.ai, a meeting transcription app with a similar name. The cross-engine study of what seven AI engines cite counts how often this happened elsewhere, and what Gemini cites has Gemini's full source list.
What does that mean for your site?
Every slot counts more. At about three and a half sources per answer, a Gemini citation is a large part of what the reader sees, and there are few of them. Being one of the obvious best sources for the question matters more than being one of many.
Google Search is the entry point. Gemini grounds on the Search index, so a page that is not indexed, or not competitive for the short queries Gemini writes, is unlikely to be retrieved. Standard SEO work is Gemini work.
Google-Extended is a real trade-off. Blocking it keeps your content out of Gemini training and Gemini grounding together. If being cited by Gemini matters to you, weigh that before adding the rule. How to set up robots.txt for AI crawlers covers the syntax.
Make your name impossible to misread. Gemini confused one product with another that has a similar name. Put your category next to your name in titles, headings and descriptions. What is an entity in search? explains why consistency helps.
Roundups decide most of the rest. Three in four citations to Gemini's top third-party sites went to lists, so being on the lists Gemini reads is the main lever outside your own site.
How do you get cited by Gemini, step by step?
- Check robots.txt for a Google-Extended rule. If one exists, decide whether keeping your content out of Gemini training is worth losing Gemini grounding.
- Confirm Googlebot can crawl your key pages, and use URL Inspection in Search Console to confirm they are indexed.
- For each buyer question, write down the short search queries a model would run for it.
- Check where your pages rank in Google for those queries.
- Improve the one page that should answer each question until it is a clear match for those queries.
- Add your category next to your product name in your homepage title and main headings.
- Find the roundups Gemini cites for your questions and ask their authors to include you. How to get into "best tools" roundups covers the pitch.
- Run your questions in the Gemini app, open Sources on each answer, and record which pages it links. Note answers with no sources too.
- Repeat monthly and compare. Do not ask Gemini to explain its choices; Google says it may get that wrong.
What to do first
- Look for a Google-Extended rule in your robots.txt and make a deliberate decision about it.
- Confirm your five most important pages are indexed in Google.
- Put your category next to your name in your homepage title.
- Record Gemini's sources for your top three questions as a baseline.