ScoredByAI measures how often ChatGPT recommends your clients in their city: the written answer and the map carousel, real confidence intervals, and reports you can put your agency’s name on.
Early access · built for local SEO agencies · US marketsShare of sampled ChatGPT answers naming each business · 90% CI
“Best dentist near me” is now a question people put to ChatGPT, and it answers with three names and a map, not ten blue links. Classic local SEO tooling can’t tell you whether your client is one of the three.
ChatGPT answers change from run to run. You can look great on Monday and be invisible on Tuesday, and neither run tells you the truth.
Your client can hold position #1 on Google and still never be named when someone asks ChatGPT for a recommendation in their city.
ChatGPT leans on a handful of sources per city and category. Without knowing which ones, “improve AI visibility” isn’t an action plan.
Paste a Google Maps link. We email you a report like this.
US businesses only.
One free check per email. By requesting it you agree to our Terms and Privacy Policy.
From “no idea where we stand” to a client-ready deliverable in three steps.
You paste a client’s Google Maps link and we resolve the city, the state and the category it competes in. Then we ask ChatGPT the way a customer would, across three phrasings and with the state in the question, on the real consumer interface rather than a developer API.
Mention rate with a 90% confidence interval, measured separately for the written answer and the places carousel. The kind of number that survives a client asking how you know.
We extract the sources behind the answers and rank the domains ChatGPT actually pulled from in that city. That list is the checklist you work through with your client.
Anyone can paste a prompt into ChatGPT. We built the measurement layer underneath.
Repeated sampling across three phrasings, confidence intervals on every number. Show clients evidence that survives questions.
Every scan becomes Report 1, Report 2, Report 3, each at its own permanent link. Your client opens the old one next to the new one and watches 17% become 31%: the difference between a one-off audit and a line item they renew.
We ask ChatGPT about Springfield, Ohio, not just Springfield. In our own field test the wording without the state made the model stop and ask which Springfield in 8 of 20 runs, and none of the 85 businesses it did name were in the target city. Ask by city alone and the report can be about someone else’s market.
Every recommendation comes from somewhere. We rank the domains the answers in that city actually cited, so the directories, publications and review sites your client should be on come from real citations, not a generic audit.
The map carousel is built from directory listings, and ChatGPT does not say which. We read it off the card itself and name it per business: Yelp, Google or TripAdvisor, with anything we cannot place shown as unidentified instead of guessed. So you see which profile actually feeds the answers in that market, and where the work pays off. See the Source column in the sample.
The written answer and the places carousel are different battles, and we measure both. Businesses are matched by map listing id, never by the spelling in the sentence, and a chain rolls into one line you can expand instead of five near-duplicates. You also see when your client was shortlisted for the carousel but not displayed.
Two reports from the same scan: you send one and keep the other. Both download as PDF and both keep their past editions.
The agency view: every business in the market with a 90% confidence interval on both surfaces, the directory behind each map card, the domains the answers cited, the search queries the model ran, and the thresholds behind every status. It’s how you answer the question your client asks after the good news.
See the agency viewA one-page report under your agency name and nothing of ours: their status, how often ChatGPT names them, the businesses it names instead, and how the number has moved since your last report.
See the client one-pagerBeta pricing, held for 6 months from the day you sign up. Not sure yet? The free check costs nothing and needs no account.
For consultants answering the first question every client asks: are we in ChatGPT?
For consultants proving ChatGPT visibility across their client list.
For micro-agencies packaging ChatGPT visibility as a service: ten clients at under $10 each.
Prices in USD, billed monthly by invoice. Client limits and plan contents may still change during the beta; your price does not, for those 6 months. Cancel any time in your portal: monitoring runs to the end of the period you have paid for, and that period is not refunded.
Create your account and add clients from your portal.
First reports arrive by email.
No card needed right now. Monitoring starts right away and a monthly invoice follows by email.