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
I ask the AI assistants the questions your customers ask and write down every answer. You get a record of who was named, what was said about your company and which sources were used.
No long contracts. You talk to Simon.
Rain is coming through my tile roof. Who should I call for roof repair in El Cajon?
Here are companies I would look at:
Scientific SEO
An AI visibility audit records what AI assistants say when your customers ask them who to call: which businesses each assistant names, what it says about yours, and which sources it cites. It’s a measurement, not an opinion about your website.
I build them from the services you most want calls for and the cities you serve, worded the way a homeowner would type or say them, with no generic list behind them. For a roofer the set includes recommendation questions (“who should I call for roof repair in El Cajon”), questions about your company by name, and the cost and how-to questions people ask before they are ready to call. You see the list and can add to it before I run anything.
ChatGPT, Gemini, Perplexity and Google’s AI Overviews work differently and name different companies, so I record each on its own.
I run every prompt in a clean session, without earlier conversation that could tilt the answer. I also run it more than once, because the same question can produce a different answer an hour later.
| Recorded for every run | Why it matters |
|---|---|
| Each business named, in the order given | Shows who holds the spots you want |
| What the assistant says about your company | Catches wrong services, cities or phone numbers |
| The sources cited or linked | Shows which pages the answer was built from |
| The exact prompt wording and the date | Lets the same test be repeated later |
A plain-English report, with the full record behind it. The report says how often your company was named, by which assistant and for which kinds of question. It lists anything an assistant got wrong about you and where that error most likely comes from, then the competitors that appear most often and the sources their mentions rest on. Last comes a short list of changes in priority order, and the one I’d test first.
Many of those changes turn out to be ordinary local SEO items, such as a missing service on your Google Business Profile or a directory with an old number. When that is the case I point you to the relevant part of a local SEO audit so you aren’t sold the same fix twice.
What you hold at the end is a baseline taken over a short period. By itself it changes nothing and predicts nothing, and acting on it can’t guarantee that a business gets recommended. The report doesn’t imply otherwise. With the baseline in hand I can run a fair test: change one thing, then run the same prompts and compare. That repeat measurement is AI visibility tracking, and it only makes sense once the audit exists.
Every engagement starts and ends with the same measurement.
Sample layout. Not a real client.
You should, and treat what comes back as one sample. One answer on one day is a single reading from something that varies, and an assistant that knows your account or location may answer you differently from a stranger. The audit asks many questions, several times each, in clean sessions without earlier conversation that could tilt the answer, across four assistants. It writes down the exact prompt wording and the date, so the same test can be repeated and compared later.
An SEO audit measures where you rank in Google's results and on the map, and why. An AI visibility audit records what assistants say in their own words when asked who to call. The causes overlap heavily, so findings from one often explain the other, and the report labels any AI finding that is really a local SEO problem.
Not for the audit, which uses public assistants and public pages. I need access to your website and Google Business Profile only if you have me make the changes.
Then the report says that, and the list of fixes is short. I won't manufacture problems to justify more work. I'd still check that what the assistants say about you is accurate, look at the prompts where you're missing, and note which sources your mentions depend on, since a mention resting on one directory page is fragile.
Your trade, your main services and the cities you serve are enough for me to tell you which prompts I'd test and why.
Prefer to talk? Call (619) 675-7678, 9am to 6pm.