To check whether your brand shows up in ChatGPT, write 20 to 30 questions your buyers actually ask (category and use-case questions like "best CRM for a small agency," not your brand name), turn web search on, ask each one in ChatGPT, and record whether you appear, where you rank, how it describes you, and which sources it cites. Run each question more than once, because ChatGPT gives different answers on different days. Repeat the whole pass monthly and check other AI search engines too, since ChatGPT is only one of several. This guide gives you the exact question list, the log to fill in, and how to read the results without fooling yourself.
The hard part is not running the queries. It is doing it in a way that produces signal instead of noise, and knowing what to do once you see where you stand.
How do I check if ChatGPT mentions my brand?
Ask ChatGPT the questions your customers ask, with web search enabled, and log the answers. Open a fresh chat (so earlier messages do not bias the result), make sure the web search toggle is on so ChatGPT pulls live sources, type a real buyer question, and read the response for three things: does it name you, where in the list you land, and what sources it links. Do not start by typing your own brand name. ChatGPT will happily describe a brand you ask about by name, which tells you nothing about whether it would ever recommend you on its own.
The reason web search matters is that ChatGPT answers two different ways. Without search it leans on training data, a fixed snapshot of the web from before its cutoff. With search on, it runs live queries through Bing and Google and builds the answer from current pages. The second mode is the one that reflects how a real buyer's question gets answered today, so that is the mode you test.
What to do: Before you test anything, confirm web search is enabled in your ChatGPT session, and start every query in a brand-new conversation.
Build a list of 20 to 30 real buyer questions
The questions you test should be the ones a buyer types when they do not yet know your brand exists. Category questions ("best project management tool for remote teams"), use-case questions ("what software helps a law firm track billable hours"), and comparison questions ("alternatives to [the market leader]") are what surface recommendations. A query containing your own name is not a visibility test, because the answer is guaranteed to mention you.
Spread the list across the different ways people phrase the same need. AI search engines break each prompt into a branching tree of sub-questions, so the same buyer intent can be reached through a dozen wordings. If you only test one phrasing, you see one branch. Cover the category, the use case, the buyer's constraints (budget, company size, industry), and the head-to-head comparison.
A workable set of 20 to 30 questions for a mid-market CRM might include:
- "Best CRM for a 10-person sales team"
- "Affordable CRM for a startup"
- "CRM with the best email automation"
- "What CRM integrates with [tool your buyers use]"
- "Alternatives to [the incumbent in your category]"
- "Easiest CRM to set up for a non-technical team"
What to do: Pull the questions from real sources, not your imagination: your sales call notes, your site search logs, the "people also ask" boxes in Google, and the threads in the subreddits where your buyers hang out. The closer your list is to real phrasing, the more honest the result.
Log what you find: appearance, position, sentiment, sources
For each question, record four things, because "did I show up" is the least useful of them on its own. Track whether you appeared at all, your position in the list if you did, how ChatGPT described you (sentiment), and which sources it cited to build the answer. The sources column is the one most people skip and the one that tells you the most, because it shows you exactly which pages ChatGPT trusts in your category.
A simple table is enough. One row per query, and columns for the run date, the question, appeared (yes or no), position, a short note on how you were described, the competitors named ahead of you, and the cited sources.
| Question | Run | Appeared | Position | How described | Sources cited |
|---|---|---|---|---|---|
| Best CRM for a 10-person team | Jun 24 | No | n/a | n/a | G2 listicle, two Reddit threads, a competitor blog |
| Affordable CRM for a startup | Jun 24 | Yes | 4th | "budget-friendly, fewer integrations" | Capterra, a roundup article |
The sources column is your roadmap. If the same three review sites, Reddit threads, and "best of" articles keep showing up across your queries and you are absent from them, that is not a mystery anymore. That is your work list. The full method for turning this into a number you can track over time is in how to measure AI search visibility.
What to do: Pay as much attention to where you are absent as to where you appear. The cited sources you do not appear on are the highest-impact places to build presence.
Run each query more than once
Ask the same question at least three times, on different days, because ChatGPT does not return the same answer twice. The model introduces variation by design, and its live search pulls slightly different pages each time, so the brand ranked second today can be absent tomorrow with no change on your end. A single query is a coin flip, not a measurement.
This is the mistake that wrecks most do-it-yourself checks. Someone asks once, sees a competitor, and concludes ChatGPT "hates" their brand, or asks once, sees themselves, and declares victory. Both are reading noise as signal. Only the pattern across repeated runs means anything: if you appear in four of five runs of the same question, you have real presence for that intent. If you appear in one of five, you got lucky once. The mechanics of why this happens are covered in why AI search results change.
What to do: For each question, log every run as its own row, then judge by the hit rate across runs, not by any single answer.
Check more than ChatGPT
ChatGPT is one AI search engine, and the others disagree with it constantly. Our research found that AI search engines pick a different top recommendation in about half of business queries, so being invisible in ChatGPT tells you nothing about Perplexity, Gemini, Claude, or Grok. Each one weights sources differently, which means your presence can vary wildly from engine to engine. A brand strong on Reddit may show up in Grok and Perplexity but be missing from Claude entirely.
Run the same question list through the other engines and keep a separate column or tab for each. The gaps between them are useful: if you appear in Perplexity but not ChatGPT, the difference is usually in which sources each engine leans on, which points you straight at the source types you are under-represented on.
What to do: Test at minimum ChatGPT, Perplexity, and Gemini. Use the same questions across all of them so the results are comparable.
A single check is noise. Make it a monthly habit.
One pass through your question list is a snapshot, and snapshots lie. AI search results shift week to week as sources get published, updated, and re-ranked, so the only way to know whether you are gaining or losing ground is to run the same list on the same schedule and compare. Honest measurement is a trend line, not a single reading.
Run the full pass monthly: same questions, same engines, multiple runs each, logged in the same sheet. Over a few months you will see whether your hit rate is climbing, which competitors are pulling ahead, and whether the new content and third-party presence you built is actually moving the answers. A practical cadence for this is laid out in what to do every month to stay visible in AI search.
What to do: Put a recurring monthly reminder on the calendar and keep every pass in the same file so the history accumulates. The history is the whole point.
The zero-effort shortcut
Running 30 questions times five runs times five engines by hand every month is roughly 750 queries, logged and read manually. That is real work, and it is why most people do the check once and never again. If you would rather skip the spreadsheet, a free scan runs a version of this pass for you and shows where you currently stand without you typing a single query.
For ongoing measurement, the manual method does not scale past a couple of months of discipline. An AEO platform automates the whole loop: it runs your buyer questions across multiple AI search engines on a schedule, logs appearance, position, sentiment, and the cited sources, and shows the trend over time so you are reading signal instead of a one-day snapshot. Loudmink does this visibility tracking across up to five AI search engines and reports where each answer's sources come from. Plans from $99/mo as of June 2026. The point of either approach, manual or automated, is the same: get from a single noisy check to a trend you can act on, which is the core of AEO.
Frequently Asked Questions
How do I check if my brand shows up in ChatGPT?
Open a fresh ChatGPT chat with web search enabled, ask the category and use-case questions your buyers ask (not your brand name), and log whether you appear, your position, how you are described, and which sources are cited. Run each question several times across different days, because answers vary, and repeat monthly to see a trend.
Why does ChatGPT give a different answer every time I ask?
ChatGPT introduces variation by design and its live web search pulls slightly different pages on each run, so the same question can name different brands on different days. This is why a single check is unreliable: you have to run each query several times and judge by how often you appear, not by one answer.
Should I just type my own brand name into ChatGPT to check?
No. Asking ChatGPT about your brand by name guarantees it describes you and tells you nothing about whether it would recommend you on its own. Test the category and use-case questions a buyer types before they know you exist, because those are the answers where recommendations actually happen.
Is checking ChatGPT enough, or do I need to check other AI search engines?
You need to check others. AI search engines disagree on the top recommendation in about half of business queries, so your standing in ChatGPT does not predict your standing in Perplexity, Gemini, Claude, or Grok. Run the same question list through at least ChatGPT, Perplexity, and Gemini.
How often should I check whether my brand shows up in AI search?
Monthly. A single check is a snapshot, and AI search results change week to week as sources are published and re-ranked. Running the same questions on the same schedule, with multiple runs each, is the only way to tell whether your visibility is improving or slipping.