To check whether your brand appears in ChatGPT, ask it the questions your customers would actually ask, rather than asking about your brand by name, and record whether you appear, how you are described, and which competitors show up instead. Because ChatGPT's answers vary between runs and differ from other platforms, test each question more than once and treat any single check as a sample rather than a verdict. This gives you a quick, useful baseline of where you stand.
In short
- Ask the questions customers ask, not your brand name, since that is what reveals real visibility.
- Record whether you appear, how you are described, and which competitors appear instead.
- Run each question more than once, since answers vary between runs.
- Treat manual checks as a sample; use a tracking tool for a reliable ongoing picture.
Why check directly?
Checking directly is the fastest way to get a concrete sense of whether AI describes your brand the way you would hope. It turns an abstract worry into specific evidence you can act on, and it costs nothing but a few minutes.
The value is in seeing reality rather than assuming it. You may believe your brand is well known in your category, but ChatGPT may not surface it for the questions that matter, or may describe it inaccurately, or may recommend competitors instead, and the only way to know is to look. A direct check gives you that ground truth for a handful of important questions, which is a sensible first step before any larger effort, as set out in the AI visibility how-to playbook. It also makes the problem tangible for others in your organization, since showing an actual ChatGPT answer that omits your brand is more persuasive than describing the issue in the abstract. The key is to check the right way, which means asking the questions a customer would ask rather than asking about yourself.
Which questions should you ask?
You should ask the questions your customers would actually ask when researching your category, not questions containing your brand name. Asking about your category is what reveals whether you surface when it counts; asking your brand name only tells you what the model already associates with a name it has been given.
The distinction matters a great deal. If you ask "what is [your brand]," ChatGPT will tell you what it knows about your brand, which is useful for checking accuracy but says nothing about whether you would be recommended to someone who does not already know you. The more revealing questions are the ones a potential customer asks before they know you exist: "what is the best [your category] for [use case]," "who are the leading [type of provider] in [location]," "how do I choose a [your product type]," and the comparison and recommendation questions that lead to decisions. These are where being present or absent actually shapes outcomes. Choosing good questions is the same skill as building a prompt set, covered in how to build a tracked prompt set, and starting with a handful of your most important customer questions is enough for an initial check.
How do you run the check and read the results?
You run the check by asking each chosen question in ChatGPT and recording three things: whether your brand appears, how it is described, and which other brands appear. Reading these three together tells you not just whether you are visible but whether that visibility helps you.
Work through it step by step. First, ask each question and note whether your brand is mentioned at all, which is your basic presence check. Second, if you appear, read how you are described, since the tone and accuracy of the description matter as much as the mention itself, the sentiment dimension covered in how does brand sentiment work in AI answers. Third, note which competitors appear, since their presence where you are absent, or alongside you, is a direct competitive signal, related to competitor mentions. It also helps to try each question both in a normal chat and in a mode where ChatGPT searches the web, since the two can give different answers and cite different sources, a behavior explained in how does ChatGPT choose its sources. Recording these observations for each question gives you a simple, structured picture of your ChatGPT visibility.
Why test more than once, and what are the limits?
You should test each question more than once because ChatGPT's answers are non-deterministic, meaning the same question can produce different answers at different times. A single check is a snapshot that may not represent the typical answer, so repeating gives a truer picture.
This points to the main limitation of manual checking. Because answers vary between runs, a one-off check can mislead in either direction, showing you present when you are usually absent or vice versa, so any conclusion should rest on several runs rather than one. Manual checking is also limited in scale: testing a handful of questions a few times each is feasible, but tracking many questions across many runs and across multiple platforms by hand quickly becomes impractical, which is exactly the problem a visibility tracking tool solves by running your questions automatically and averaging the results into a reliable trend, as explained in how to measure AI visibility. And ChatGPT is only one platform; your visibility can differ markedly on Gemini, Perplexity, Copilot, and the others, so a complete picture means checking across platforms, covered in how each AI platform surfaces and cites brands. Manual checking is a great way to start and to build intuition, but for an accurate, ongoing measure it is a sample, not a substitute for systematic tracking.