Free Check of What Gemini Says About Your Brand

A Gemini brand check shows you three things: whether the Gemini app names your company for a buyer question, what it says about you, and which pages it leaned on to say it.

That third part matters most. When Gemini grounds an answer in Google Search, it can show the sources behind it. Those URLs tell you where its picture of your brand comes from. If the sources are a stale directory listing or a competitor's comparison page, you now know which page is shaping the answer.

What the check does not tell you is how Gemini will answer tomorrow, or for a slightly different question. Treat it as a dated snapshot. For an ongoing view, see tracking what Gemini says about your brand.

Google runs several generative surfaces, and they are controlled by different settings. Google's own documentation says AI Overviews and AI Mode "may use different models and techniques" and can return different links. The Gemini app is a third product again, and the Gemini API's Grounding with Google Search is a fourth, separately billed path.

For AI Overviews and AI Mode, Google states a page must be "indexed and eligible to be shown in Google Search with a snippet" and that there are no additional technical requirements. There is no robots.txt token that blocks AI Overviews or AI Mode on its own.

The practical move: run this check on the Gemini app, and check AI Overviews separately by searching the same questions on google.com.

You need a browser, a spreadsheet and three to five buyer questions.

1. Write the questions. Use the words a buyer would type, not your brand name. Say you run a payroll tool for restaurants: "What is the best payroll software for a small restaurant?" Add one question that names you directly, such as "What does [your company] do and what does it cost?"
2. Open a clean session. Go to gemini.google.com. Note whether you are signed in, which account you used, and which model the dropdown shows. Saved history and personal context can change answers, so keep the setup the same each time you rerun.
3. Ask each question in a new chat. A fresh chat stops earlier questions from steering later answers.
4. Record the answer. Paste the full text into your sheet with the date. Note whether your brand is named, where it appears in the list, and which competitors appear.
5. Record the sources. If the answer shows source links, copy every URL. If it shows none, write "no sources shown". That usually means the answer may have come from the model's training rather than a live search.
6. Mark the facts. Highlight every specific claim about you: price, plans, location, founding year, features. Mark each one right, wrong or out of date.
7. Repeat with one rephrasing. Ask each question a second way. If the answers diverge, you have learned how unstable this result is.

Sort what you found into four buckets, because each has a different fix.

• Named, facts right, your site cited. Nothing to fix. Rerun after any pricing or product change.
• Named, facts wrong. Look at the cited sources first. If a third-party page carries the wrong price, that page is the one to correct. The guide to correcting wrong facts in Gemini covers the steps.
• Named, no sources shown. The answer may come from the model's memory. Recent news will not appear there until a later model. A knowledge cutoff check helps you see which of your updates postdate what the model knows.
• Not named, other sites cited. Engines are citing review sites, listicles or forums instead of you. The evidence shows a strong association between being named on those cited sites and appearing in the answer, though no one can promise the effect.

One thing to state plainly on your own site: the specific facts you want repeated, such as a price or a one-line description, in text that loads without JavaScript. A fetchable, specific sentence is one the engine can quote.

Google-Extended is a robots.txt token, not a separate crawler. Google's crawler documentation says blocking it cuts your content from Gemini Apps and Vertex AI training and grounding, and that it "does not impact a site's inclusion in Google Search nor is it used as a ranking signal."

A rule like this would block it:

User-agent: Google-Extended
Disallow: /

If your robots.txt has this, or a blanket AI block that includes it, the Gemini app cannot ground answers in your pages. It can still describe you from other sites and from what it already learned. The question of whether blocking Google-Extended changes what Gemini says has its own page.

Two access points are simpler on Google than elsewhere. Googlebot renders JavaScript, so client-side rendering is less of a risk for Google surfaces than for crawlers that do not. And indexing still runs through Googlebot, so a stray noindex or nosnippet keeps pages out of AI Overviews and AI Mode regardless of your Google-Extended setting.

A hand check is honest about what it is: a few answers on one day.

• Retrieval is discretionary. Gemini decides per question whether to search. The Gemini API documentation says the model "automatically generates one or multiple search queries", so the searches behind an answer are not the words you typed.
• Models change. As of October 2026, Google lists several Gemini models in the app and does not state the free default. A model switch can change answers without anything on your side changing.
• Small samples mislead. Three questions cannot tell you your share of all buyer questions.
• Citation is not influence. A page can shape an answer without being shown as a source.

The same method works on other engines. Compare the result with a Perplexity brand check or a Claude brand check, and record each engine on its own row rather than blending them.

If you want a quicker first read, the free checker on this site runs up to three buyer questions, your own or ones written from your homepage, and scores the answers with the 99Visibility score. Like the manual method, it is a snapshot, not continuous tracking. Use it to decide which questions deserve a closer manual look, then keep your dated spreadsheet as the record you compare against after each fix.

Does blocking Google-Extended remove me from AI Overviews? No. Google says Google-Extended does not affect inclusion in Google Search, and AI Overviews and AI Mode are part of Search. There is no robots.txt token that blocks only AI Overviews. Their eligibility depends on Googlebot access, indexing and snippet controls such as noindex, nosnippet and max-snippet. Search Console is rolling out a separate generative AI control for some property owners.

Why does Gemini give a different answer each time I ask? Gemini decides per question whether to search the web, and when it does, it writes its own search queries. Small wording changes can trigger different searches and different sources. Model updates and account history can also shift answers. That is why the manual method uses fresh chats, a fixed setup and one rephrasing per question, and why any single result should be dated.

Can I check Gemini through the API instead of the app? You can call the Gemini API with Grounding with Google Search enabled and read the citations it returns. It is a useful, repeatable proxy. It is not the Gemini app, though. It is a separately billed retrieval path, and the model you call may differ from the one the app serves. Label API results as a proxy when you report them.

What should I do if Gemini cites a page with my old pricing? Fix the source first. If the page is yours, update the price in plain HTML text. If it is a third-party listing or review, contact the owner or update your profile there. Then make sure your own pricing page states the current price in a specific, fetchable sentence. Rerun the same question afterwards and record the date, since nobody controls when the answer changes.

Do I need server-side rendering for Gemini to read my site? Less than for other engines. Googlebot renders JavaScript, so client-side content is generally readable to Google surfaces. Many other AI crawlers, including those behind ChatGPT, Claude and Perplexity, do not execute JavaScript. If you care about more than Google, put load-bearing facts like prices and product names in the initial HTML.