Sample layout. Not a real client.
Trade, city: the problem in one line
- Starting point
- Rank grid before any change
- What changed
- The one or two changes that moved it
- Result
- Rank grid after, and the change in calls
When a homeowner asks ChatGPT who to call, it answers with a few company names. I test which names those are for your trade and your cities, then work on the things that can change them.
No long contracts. You talk to Simon.
My AC stopped blowing cold air. Who is a reliable AC repair company in Chula Vista?
Here are companies I would look at:
Scientific SEO
ChatGPT visibility is how often, and how accurately, your business is named when a local customer asks ChatGPT for a recommendation. There’s no ranking position to check, only what the assistant says when asked, and I can test that.
Sometimes ChatGPT answers from what it learned in training, which is a snapshot of the public web up to some earlier date. Sometimes it searches the web while answering and builds the reply from the pages it finds, usually with links to them. A question about who repairs AC units in Chula Vista often leads to a search, because the answer depends on current, local facts.
As far as anyone outside OpenAI can observe, how it picks the businesses it names in that case comes down mostly to the pages the search returned: directory listings, review sites, list articles, local news and company websites.
Your site is one source among several, and often not the first one read. The links shown under a local recommendation are frequently third-party pages: a review platform or a directory, sometimes an article written by someone you have never met. If your company is missing from those pages, or described on them with an old phone number or a service you dropped years ago, the assistant has poor material to work from. Your website still matters. It’s where the assistant, or the customer who follows a link, confirms what you do and which cities you serve.
The work is to be easy to find and easy to describe on the pages an assistant is likely to read:
This is the practical side of generative engine optimization (GEO). Most of it’s local SEO done properly, so I fold it into that work.
The same question asked twice can return different companies. The answer also shifts with how the question is worded, and with whether the assistant searched the web or answered from memory. A ChatGPT recommendation is therefore something no one can guarantee, and I’d be wary of a ChatGPT ranking sold as a package.
The prompts are worded as a customer would type them, such as the Chula Vista AC question above, with variations for each service and city you care about. I run them and record which businesses are named, in what order, what is said about each, and which sources are linked. That first record is the baseline, and it’s what the AI visibility audit produces. Then I change one thing and run the same prompts again about two weeks later. If your company isn’t named today, the record of who is named, and from which sources, shows where to start.
Every engagement starts and ends with the same measurement.
Sample layout. Not a real client.
Usually because the competitor appears on the pages ChatGPT read for that question and you don't, or you appear with less detail. The linked sources under the answer show which pages those were. I record them for each prompt, because that list tells you where your company is missing: a directory, a review site, a local article, or a service page of your own that doesn't exist yet.
Often, yes, by correcting the source, since the answer itself can't be edited. The wrong fact usually comes from an outdated listing, an old page on your own site or a directory with a previous phone number. Once that is corrected, I re-run the prompt over the following weeks to see whether the answer catches up. Some errors come from training data, a snapshot of the public web up to some earlier date, and take longer to fade.
Nobody outside OpenAI can say for certain. What can be observed is which pages are linked under an answer, and those sometimes include review platforms and directories that show ratings. So I treat reviews on several platforms as evidence an assistant may meet, and I check which review sources actually appear for your trade in your cities before recommending where to put effort.
No. Good local SEO already covers most of what makes a business easy for ChatGPT to name. Only the measurement is new, reported alongside your rankings and calls.
Tell me your trade and the cities you serve. I put your customers' questions to ChatGPT and show you which companies it names.
Prefer to talk? Call (619) 675-7678, 9am to 6pm.