The first thing most owners do is type their category into ChatGPT and see whether they come up. It is a reasonable instinct and a poor instrument — for reasons that have nothing to do with the tool you would buy instead.
| AnswerRange | Asking ChatGPT yourself | |
|---|---|---|
| Platforms | ChatGPT, Google AI Overviews, Google AI Mode, Gemini, Perplexity | One |
| Sample | Dozens of questions, each repeated, per market | One question, once, in one phrasing |
| Your own history | Fresh sessions through the provider API; no memory, no personalisation | Your account's memory and your past prompts shape the answer |
| Language | Questions written natively in six languages | Whatever you typed, usually English |
| Who is cited instead | Ranked across the whole answer pool | Whoever came up this time |
| Change over time | Monthly, judged by a test | Memory of what it said last month |
| Cost | First measurement free | Free, and worth about that much as evidence |
Nothing here is a criticism of ChatGPT. A single conversation is not designed to be a survey, and a survey is what a visibility number needs.
Type the question and read the whole reply. The measurement tells you how often you are cited; reading one answer tells you how the assistant frames the category.
After a page goes live, one manual question with web search on will show whether the page is being read at all. It will not show whether the citation rate moved.
Run the measurement. It is free the first time and takes a few minutes.