To check whether AI recommends your company, ask relevant customer questions without inserting your company name, then inspect the resulting answers. One favourable answer is an example, not a visibility score.
Set up a small, useful first test
- Write a discovery question about your product category.
- Add a question with a genuine buying constraint, such as team size or budget.
- Add a recommendation question for a specific audience.
- Choose the country, language, model and search mode.
- Save every answer, including unfavourable results and failures.
For an illustrative payroll business, these could be “Which payroll providers serve small businesses?” and “Which payroll solution works for a company hiring its first employee?” These are examples, not measured results.
Read the answer before scoring it
Is your company named? Is it recommended, or only used as an example? Does the answer describe it positively, neutrally or negatively? Which other companies appear? Are any sources actually cited?
An answer that calls a provider “well known” is not necessarily telling the reader to choose it. Likewise, the first company mentioned in a paragraph is not automatically the number-one recommendation.
What if your brand is missing?
First check whether the selected questions fit what you sell and whom you serve. Then inspect competitor mentions and sources. The correct conclusion is “not observed in this sample,” not “AI does not know our business.”
Turn the check into a baseline
Keep prompts and settings stable for the next scan. In Propps, setup helps you approve questions and start a first scan; saved answers and a summary make the result inspectable. API measurements can differ from personalised consumer-app sessions.