Ask the questions a customer would ask
The most common mistake is asking an assistant about yourself by name. “Is Smith Plumbing good?” tells you whether the model can find you. It does not tell you whether a stranger who has never heard of you would be pointed your way. Both matter, and they are different questions.
- 01Write down three discovery questions with no business name in them, such as “What is the best plumber in Austin?” or “Can you recommend a family dentist in Leeds? Please name specific practices.”
- 02Write down one or two questions that name you, such as “Is Smith Plumbing a good choice for a water heater in Austin?”
- 03Open a fresh chat or a private window so earlier conversations and your own account history do not steer the answer.
- 04Ask each question and copy the reply word for word into a document, with the date, the app, and whether web search was on.
Read the reply, not just the yes or no
- Were you named at all? If yes, in which position of a list?
- Who was named instead? Those are the businesses the model found easier to support.
- Which pages did it cite, if it shows sources? Your own site, a directory, a review site, a local news piece?
- Did it say anything wrong: an old address, a service you do not offer, hours that changed?
- Did it refuse? “I couldn’t find reliable information about that business” is a useful result. It means the public web gives the model too little to work with.
Why one check is not enough
AI assistants are not deterministic. The same question can produce a different list an hour later, and the web pages a search-grounded model retrieves change too. Location, the app version, whether you are signed in, and how the question is phrased all move the answer. Treat each run as one sample.
If you repeat the same questions every week and write the results down, a pattern appears: are you named more often than last month, and by which questions? That pattern is worth far more than a single screenshot.
What our free scan does, and what it does not
AskedAboutYou automates the method above. It writes five customer-style questions from your business name, category and city, sends them to a search-grounded AI model with live web search (currently Perplexity Sonar), and shows each reply word for word with the model name and the time.
It is not a feed of what every ChatGPT user sees. Nobody outside OpenAI has that. It is a repeatable sample from a model that searches the same public web your customers’ assistants read. If you specifically want ChatGPT’s answer, ask ChatGPT the same questions by hand, using the steps on this page, and compare.
Questions
Can I pay ChatGPT to recommend my business?
Not as of this writing in the way you can buy a search ad. Some assistants are experimenting with ads, and policies change, so check the provider’s own announcements. Anyone who promises a guaranteed AI recommendation for a fee is promising something they do not control.
Why does ChatGPT name a competitor and not me?
Usually because the public web makes the competitor easier to support: clearer service pages, consistent directory listings, more third-party mentions, or reviews that describe the specific service. Look at the sources the answer cites to see which.
Is the free scan the same as asking ChatGPT?
No. It asks a different search-grounded model through its API and says so on every result. It is useful because it is repeatable and shows the reply verbatim, not because it copies any one consumer app.
Check what AI says about your business
The free scan asks five customer-style questions with live web search and shows every reply word for word, who was named, and which pages were cited. About a minute, no card.
Run my free scan