The method
We ask four AI assistants the same five questions.
Every number on a FoundByName report comes from one scan of the whole city, re-run every week. Nothing is estimated.
Step 1, the scan
The questions people actually type.
For each city and line of work, the same five questions go to ChatGPT, Claude, Gemini and Perplexity. Every business named in an answer goes into the record, with its position and the pages the assistant cited.
- 01“best your trade in your city, state”
- 02“top rated your trade in your city”
- 03“who is the best your trade in your city, state?”
- 04“most recommended your trade in your city”
- 05“your trade in your city with the best reviews”
One scan for the whole city
We never run a separate scan per business. Every report in a city reads from the same record, taken on the same day, so two businesses can compare their numbers.
Only the blue parts change
Your trade and your city fill the blanks. The shape of the questions stays the same for everyone, so a count in one city means the same thing as a count in another.
How we ask
Through each assistant’s own API, with web search on.
We send the questions the way a developer would: through the official API of each company, with the assistant allowed to search the web before it answers.
ChatGPT
OpenAI API, web search tool
Model, Sept. 2026gpt-5.4
Perplexity
Perplexity API, which searches on every question
Model, Sept. 2026sonar
Gemini
Gemini API, grounded in Google Search
Model, Sept. 2026gemini-3.6-flash
Claude
Anthropic API, web search tool
Model, Sept. 2026claude-opus-5
Every answer is saved as it came back, with the date, the model that wrote it and the pages it cited. Nothing is rewritten or estimated afterwards.
What this can’t show: the app on someone’s phone can answer differently. It may know where they are or what they asked before, and it may run a different model. So a report can’t show what every customer sees. The questions and the way we ask stay the same every week, so you can compare one week with the next.
Step 2, the count
A number you can check yourself.
Five questions on four assistants make twenty answers per market. Your report counts how many of them name your business, and shows where you landed in each list. All twenty are printed on the page, so you can count them yourself.
Question by question
Raphaelson & Levine, New York. The same screen as in the account.
| Question | ChatGPT | Claude | Gemini | Perplexity |
|---|---|---|---|---|
| who is the best personal injury lawyer in New York, NY? | #5 | #2 | ||
| most recommended personal injury lawyer in New York | ||||
| best personal injury lawyer in New York, NY | #4 | |||
| personal injury lawyer in New York with the best reviews | #2 | |||
| top rated personal injury lawyer in New York | #6 | #5 | ||
| Named you | 1 of 5 | 1 of 5 | 1 of 5 | 3 of 5 |
Named, and where
A blue number is your spot in that list. First or fifth, it counts the same in your total.
Named others, not you
That answer counts as zero for you. Most of the businesses we track sit here today.
Named nobody
The assistant answered without a single business. It happens, and we keep it.
Step 3, the work
We publish a clear page with your current details.
It lists your services and hours and gives a plain answer to the questions people ask. We build it from your website and Google listing. Then the same scan runs again every week and your report updates, including the weeks when nothing moved.
01
A clear page with your details
Live on foundbyname.com the day you pay, then on your own domain.
02
Your answers, on the page
One questionnaire. It becomes the questions people ask, and your answers.
03
Listings checked line by line
Phone, address, website and hours, across the sites the assistants read.
04
Asked again every week
The same twenty questions, a new count, whatever it shows.
The full report
Get the full report by email.
It asks four assistants five questions and counts every answer. We build it for your city and what you do, then email you the link, usually within a day.
- Every question, with the names each assistant gave
- Who got named instead of you
- The sites the assistants cite
Run an agency? See the white-label terms.