How do I check if ChatGPT, Perplexity, or AI Overviews cite my brand?

By José Antonio Mijares | 2026-08-17 | 7 min read

Use the queries you already rank for. Ask the six surfaces. Log who is named. No dashboard required.

You do not need a dashboard to find out if an answer engine names you. You need the queries you already rank for, the six surfaces, and a log.

The six surfaces are ChatGPT, Claude, Perplexity, Gemini, Grok, and Google AI Overviews. Not Copilot. Reddit is not a sixth engine. Rank is not a cite.

If a URL sits on page one and the answers never mention you, that URL is probably not citable. Fix that page. Do not write a new blog on the same keyword.

Start in Search Console

Open Google Search Console. Pick five queries where you already appear. Write down the exact URL that ranks for each one.

Those five queries are the prompt list. Do not invent a second list of “AI keywords.” The buyer is already typing something Google records. Use that.

A collection that ranks for “best wool blankets” is the page you will test. Not a new article that repeats the phrase.

Ask the same prompt on all six

For each query, run it as the buyer would type it. Then run a brand query: “who makes [your product]” and “best [your category] store.”

Do this on each surface:

1. ChatGPT — same query, then the brand query. Are you named. Who else is named. What does it get wrong.
2. Claude — same two prompts. Claude is slower and more careful. Silence here still counts.
3. Perplexity — same two prompts. Then open the sources it lists. If you are not in the source list, you were not used.
4. Gemini — same two prompts. Overlaps with Search more than the others.
5. Grok — same two prompts. Recent pages can show up. Still a citation check, not a trend check.
6. AI Overviews — run the query in Google. Does the box appear. Are you in it. Which URLs are cited.

You now have thirty cells if you ran five queries on six surfaces. Named or not is enough. Do not invent a citation rate.

Read Perplexity’s sources

Perplexity shows its work. That is the useful part.

If it names a competitor, open the cited URL. That page has a fact the model could lift — a named product, a material, a process, a place, a date. Your ranking URL probably does not.

If it names no one in your category, the query may not be a citation query yet. Log that. Do not treat it as a win.

Fix the ranking URL

The page that already ranks is the page to fix. A new blog cluster on the same keyword does not make a product grid citable.

What a model can lift:

• A clear claim in the first sentence
• A named product, not “best sellers”
• A material, a size, a process, a place, a date
• The same names on About and on the PDP
• A visible FAQ whose text matches the FAQ schema

What it skips:

• A collection that is only cards and a filter bar
• A PDP with an empty body
• Two pages that contradict each other
• A blog that restates the collection title

Shopify stores hit this gap often. Why a store can rank and still never get mentioned is the store-side version of the same problem.

What the log is for

Write four columns: query, surface, named or not, who else is named. Repeat the same list later. The change you want is a name that was missing and is now there.

Jamak logs those six surfaces in Ezeo when the work is in scope. A sheet is enough if you are doing it yourself. The method does not change.

Do not buy a package that promises a citation percentage. You cannot optimize a number you invented.

Frequently Asked Questions

Q: Do I need a paid AI visibility tool to check citations?

No. Search Console gives you the queries. The six surfaces are public. A sheet is the report. A dashboard can store the log. It cannot replace the prompts.

Q: Which surfaces should I check?

Six: ChatGPT, Claude, Perplexity, Gemini, Grok, and Google AI Overviews. Not Copilot. Reddit is community and PR, not a sixth engine.

Q: What if I rank and I am never named?

The ranking URL is probably not citable. Add facts a model can lift. Do not start a new blog on the same keyword. See the Shopify version of this gap if the URL is a collection or a thin PDP.

Q: How often should I re-run the check?

After you change the ranking URL, run the same prompts again. Weekly while you are editing. Monthly when the log is stable. Do not add a new query list every week.

Q: Does appearing in Perplexity’s sources mean I will appear in ChatGPT?

No. The surfaces do not share one index. A source link in Perplexity is a good sign for that surface. Check the other five. Rank is still not a cite.

The work is citations on those six surfaces. That is GEO. Talk to us if you want the log run and the ranking URLs fixed. No rate card. No fake results.