How to rank in Gemini and get named in its answers

What does it mean to rank in Gemini?

To rank in Gemini is to be the brand, place, product or page it names when someone asks it something in your category. In practice that's three different outcomes: a link in the Sources panel, a mention in the text with no link, or a card for a place or product. The work behind each is different, which is why generic "Gemini SEO" advice tends to miss. The audience is large: Alphabet said in July 2026 that the Gemini app had 950 million monthly active users, with daily users tripling over the year14.

It also helps to separate the products. The Gemini app (gemini.google.com and the mobile apps) is an assistant. AI Overviews and AI Mode are features inside Google Search. They share a model family (Google made Gemini 3 the default model for AI Overviews in January 2026)9, but they're controlled and reported differently. This guide covers the app. For how every engine's retrieval compares, see how AI engines choose their sources.

How does Gemini decide what to cite and recommend?

Gemini grounds answers in Google's own data. It decides whether a search would help, runs one or more Google queries, and writes from what comes back. For places it reads Google Maps, and for products, the Shopping Graph. When it doesn't search, it answers from training.

Grounding in the Google Search index

Google's crawler documentation describes grounding for Gemini Apps as "providing content from the Google Search index to the model at prompt time to improve factuality and relevancy"1. The Gemini API's grounding documentation describes the same pattern from the developer side: "The model analyzes the prompt and determines if a Google Search can improve the answer," then generates and runs one or more queries2.

So the index is Google's, and the crawler is Googlebot. If a page isn't in Google's index, Gemini can't ground on it. That's the biggest difference from Claude (Brave) and Perplexity (its own index): for Gemini, classic Google SEO is the entry ticket.

What Google-Extended actually controls

Google-Extended is the robots.txt token that decides whether Google may use your content to train future Gemini models and to ground Gemini answers. Google's common crawlers page is explicit that it "does not impact a site's inclusion in Google Search nor is it used as a ranking signal in Google Search"1. So blocking it doesn't remove you from AI Overviews, but it does take your pages out of what the Gemini app can ground on. Many publishers blocked it to opt out of training without realizing they were opting out of Gemini app citations too.

Places and products come from Maps and the Shopping Graph

For local questions, the Gemini Help Center says the app uses "public information from Google Maps" to give addresses, websites, ratings and opening hours4. Google's developer equivalent, grounding with Google Maps, draws on a "database of over 250 million places worldwide," including reviews and photos5. For a restaurant, clinic or local service, your Google Business Profile listing is the page Gemini reads.

For products, Google said in November 2025 that the Gemini app brings "product listings, comparison tables and prices directly from Google's Shopping Graph," which holds more than 50 billion listings6. Google's Shopping Graph explainer says listings are updated through information retailers share via Google Merchant Center or post on the web7.

How sources appear in the app

When sources are available, the Gemini app shows a Sources button at the bottom of the response or inline, which opens a panel of related links. If Gemini quotes a long passage from a page, it links that page3. Many answers show no sources at all.

How often does Gemini actually link out?

Less than any other major assistant. In Slate's test of 1,000 B2B software questions, run through the API on 3 August 2026, Gemini returned 0.07 links per answer, and only 16 of 1,000 answers contained a web address, all as mentions inside the text10. The app may behave differently from the API, but a separate panel by cloro, published July 2026, also found Gemini cited a source on just 41% of answers11. In Ahrefs' study of about 15,000 prompts, only around 12% of URLs cited by ChatGPT, Gemini and Copilot ranked in Google's top 10 for the same prompt, and Ahrefs noted that "ChatGPT and Gemini, on the other hand, don't always link out"12.

The 8-step plan to get named by Gemini

Steps 1 to 4 set up the baseline and the technical base. Steps 5 to 7 are the ongoing work. Each ends with a test.

Worked example: reading Gemini's sub-queries

This is a hypothetical company, "Haulwise," a transportation management system for small trucking fleets. We haven't run these prompts; the example shows the method and the decisions it leads to, not results.

The API call. Haulwise sends its prompt to the Gemini API with Google Search grounding switched on, in the request shape Google's documentation shows2:

curl "https://generativelanguage.googleapis.com/v1beta/models/gemini-3.8-flash:generateContent" \
  -H "x-goog-api-key: $GEMINI_API_KEY" \
  -H "Content-Type: application/json" \
  -X POST \
  -d '{
    "contents": [{"parts": [{"text": "best TMS for a 50-truck carrier"}]}],
    "tools": [{"google_search": {}}]
  }'

It reads groundingMetadata.webSearchQueries for the sub-queries and groundingChunks for the cited pages. Google bills grounded search, so this is for a few dozen prompts, not a daily tracker.

The sheet. One row per sub-query:

prompt_id | sub_query | haulwise_google_position | ranking_url | top_3_urls | haulwise_named_on_top_3? | action

Reading it. Each pattern points to a different move:

• If the sub-queries are narrower than the prompt (say, a TMS query about fleet size and another about pricing), Haulwise needs a page for each, not one "best TMS" page.
• If Haulwise ranks top three for a sub-query and still isn't named, it checks the snippet: a data-nosnippet wrapper or a JavaScript-loaded answer can keep the passage out of reach.
• If the top three results are listicles that skip Haulwise, the move is outreach (step 7), not another rewrite of its own page.
• If Gemini names Haulwise in the app with no link, the brand is in training data or in grounded pages. Haulwise logs it as a mention and keeps working the pages that rank.

What doesn't get you into Gemini

Blocking Google-Extended "for privacy" and expecting nothing to change

It changes one thing: Gemini app grounding on your pages1. If you sell to people who ask Gemini, that's a real cost for a modest training opt-out.

Assuming AI Overviews wins carry over

They overlap, because both run on Google's index. But the Gemini app shows links far less often, adds Maps and Shopping data, and is controlled by Google-Extended, which AI Overviews ignore. Track them separately; our AI Overview tracker comparison covers that side.

New files and special markup

Google's own AI optimization guide says you don't need new machine-readable files, AI text files or special markup to appear in its generative AI search features8. That guidance is about Search, but nothing Google has published says the Gemini app treats llms.txt differently.

Manufacturing mentions

The same guide warns that "seeking inauthentic 'mentions' across the web isn't as helpful as it might seem"8. Mentions matter because they sit on pages Google ranks, which is exactly what thin, paid placements rarely achieve.

How do you measure your visibility in Gemini?

Report three rates from your prompt set each month: link rate (runs where your URL is in Sources), mention rate (runs that name you) and card rate for places or products, each with run counts. Our guide to measuring AI visibility covers sample sizes, and our AI visibility tools comparison shows which trackers include Gemini.

For traffic, filter GA4 for sessions from gemini.google.com. Clicks from AI Overviews and AI Mode arrive as Google organic and can't be split out that way. The trend is worth watching: in Goodie's 2026 report, Gemini's share of AI referral sessions on a mostly B2B panel rose from 2.4% in May to August 2025 to 10.6% in March and April 202613. For a quick free read, the Arobis AI visibility checker shows whether AI engines name your brand for your category's questions.

How long does it take to rank in Gemini?

As long as it takes to rank in Google for the sub-queries, because that's the index Gemini grounds on. Unblocking Google-Extended or fixing a snippet restriction can show up within weeks, once Google recrawls. Business Profile and Merchant Center changes appear once Google processes them. Brand mentions in no-link answers move slowest, since they depend on other sites updating and on new model versions. Judge the work over a quarter, by bucket. The ChatGPT guide covers the engine where off-site mentions matter even more.

Frequently asked questions

Sources