How agencies can audit a client’s AI visibility without overselling it

Set up the audit

1. 01Agree with the client on a handful of questions their customers really ask, mixing discovery questions without the brand name and one or two that name the business.
2. 02Fix the wording, location and assistants you will use, and write them down so later checks are comparable.
3. 03Run each question in fresh sessions, with web search on where the app offers it, and save every reply word for word with date and app.
4. 04List every business named, the position, and every source cited.

Turn answers into findings

• Is the client named at all, and for which questions?
• Which competitors are named, and what do the cited pages say about them that the client’s pages do not?
• Are any details about the client wrong, and where does the wrong detail come from?
• Which third-party sources keep appearing, and is the client listed on them accurately?

Report it honestly

Explain that answers vary between runs and that no one outside the AI companies sees what every user is shown. Present results as a sample with a trend over time, not as a ranking. Avoid promising that the work will get the client recommended; promise the specific fixes instead, such as completed listings, new service pages and corrected data.

This framing protects the agency, too. When an answer changes for reasons nobody controls, the client already knows that can happen.

Use the free scan as a starting point

The AskedAboutYou free scan runs five customer-style questions for one business, category and city through a search-grounded model and shows the replies, the businesses named and the cited pages. It is a quick first sample to bring to a client conversation. Run it for each client separately, then follow up with your own manual checks in the assistants that client cares about.