People ask ChatGPT for a restaurant or a plumber the way they used to ask a friend. The answer is not stable. Run the same prompt twice and the list can change.
What Frontage records
- The prompt, such as "best restaurant in Grayslake IL".
- Three samples, because one answer is not a measurement.
- Whether the business was named, and where it sat in the list when it was.
- Which competitors were named in the same answer.
What can make a mention more likely
Assistants tend to repeat businesses that are described the same way in many public places: a complete Google listing, reviews that mention the actual service, and a website with the same name, address, and phone. Frontage helps you keep those words consistent. It does not control ChatGPT, and it does not guarantee a citation.
If no OpenAI key is set, the check uses labeled demo samples so you can see the report before you connect a model.