AI visibility for local businesses and single-vertical practices

A local buyer asks an assistant a question with a place in it — a dentist in a named suburb, a med spa that offers two specific treatments, an HVAC contractor who can come out today — and the answer is assembled at that moment from pages the assistant can read. For a local business, almost none of those pages are yours. In our published sample, the measured clinic appeared on 1 of the 112 third-party sites the answers were built from, even though it was the most-named business in its own market.

That is the pattern these five pieces describe. Google's AI Overviews and the map pack are built by different machinery and can disagree: the clinic was cited in 4 of its 12 AI Overviews and 0 of its 25 AI Mode answers while ranking first on the map. Vertical guides repeat the same checks for dentistry and med spas, where the buyer questions and the sources differ in detail but not in kind. The self-check piece is the one to read first if you have never looked: five steps, an evening, no spend, and an honest line on what a self-check cannot tell you.

None of this is a guarantee that fixing the listed items will get you cited. It is a list of what the answers are built from, and which of those pieces you can influence.

Read by question

1. Google AI Overviews for local businesses — how the answer is assembled, and the four checks that decide whether you appear.
2. In the map pack but invisible in AI Overviews? — why the two rankings differ, and how to check which one you are missing.
3. Do AI assistants recommend your med spa? — one fully published measurement, every number from the report.
4. AI visibility for dental practices — what transfers from the measured samples to dentists, and what needs measuring in your own market.
5. ChatGPT SEO audit for local business: the five-step self-check — what you can check tonight, free, and what it will not tell you.

Where to go next

• Start free: the free check reads your site and shows the exact questions a full measurement would ask.
• See a real local measurement: the published sample is a real single-location run, pseudonymised, printed in full.
• Read the method: how the measurement works — including the map-pack probes taken from five points around the location.