Answer
How do I check if ChatGPT recommends my brand?
Ask the way a buyer would: describe the problem instead of naming your company. "Which provider is best for X?" If competitors are named and you are not, that is your visibility gap. Asking "what does [your brand] do?" only proves the assistant has heard of you, which is a far easier test and tells you almost nothing.
Most people check this the wrong way, get a reassuring answer, and conclude they are fine.
The test that flatters you
Type your company name and ask what it does. You will almost always get a competent, positive summary — there is enough material about nearly any established business for a model to produce one. This tells you the assistant can retrieve information about you when handed your name. No buyer does that before they know you exist.
The test that tells you something
Describe the need in the buyer's words, with no brand names at all. "We are a mid-size insurer in Australia looking to modernise our claims platform — who should we talk to?" Then read what comes back: who is named, in what order, and which sources are cited.
Run it several times, and in a fresh session each time. Answers vary between runs, so a single result is an anecdote rather than a measurement.
What to record
Three things, every time: whether your brand was named at all, whether your site was cited as a source, and which competitors appeared instead. The third is the most useful and the one people skip — it converts a vague worry into a specific list of who is winning the questions you are losing.
Do it across assistants, not just one. ChatGPT, Claude, Gemini and Perplexity ground their answers differently, and being strong in one says very little about the others.
Related questions
Why do I get a different answer each time I ask?
Assistants are probabilistic and many now retrieve live search results, so responses vary between runs and over time. That is why a single check is an anecdote. Repeating the same fixed question set and recording the spread is what turns it into a measurement.
Does using my own account skew the result?
It can. Chat history, saved memory and personalisation all influence what you see. Use a logged-out or temporary session, and be aware your own browsing makes you an unusually poor test subject for your own brand.
Is checking manually good enough?
For a rough signal, yes — and it is worth doing today. It stops being sufficient when you need to track movement over time, compare across assistants, or show someone evidence, because manual checks are not repeatable and nothing is stored.