AI Visibility
A one-off check of what ChatGPT, Perplexity, Gemini and Google AI say when buyers ask for a business like yours.
The short answer.
An AI visibility check is a one-off read of what AI assistants say when someone asks them for a business like yours. We run 20 real buyer prompts through ChatGPT, Perplexity, Gemini and Google AI, log every answer, and send you a report. Nobody controls what these tools recommend, so nothing is guaranteed.
What we deliver.
A working dossier of the practice, every line is a thing we ship, not a thing we slide-deck.
We build 20 prompts out of your business, the area you serve and the competitors you name, phrased the way a customer types them rather than the way an agency writes keywords. They cover the services you want more work in, the questions buyers ask before they choose, and the blunt requests for a recommendation.
Every prompt is asked in ChatGPT, Perplexity, Gemini and Google AI, because these tools read different sources and disagree with each other more than most owners expect. One of them may already name you while another has never heard of you, and knowing which is which changes what is worth fixing first.
The first thing the report answers is blunt: across the prompts we ran, how often did your name come up, and in which tools. That mention rate is the number everything else hangs off, and for a local business seeing it for the first time the honest reading is usually lower than expected.
When a tool answers without you it still names somebody. We log every business that turns up in your place, how often each one appears and which assistants favour it, so you can see who currently owns the answer to the questions your buyers are already asking.
Being named is only half of it. We read what each answer claims about you: your hours, your services, the area you cover, your specialities, and anything stated as fact that is out of date or simply wrong. Wrong details do more damage than silence, because a customer acts on them without checking.
Where an answer shows its sources we record them, whether that is your own pages, a directory, a review site, a news story or a competitor's post. The pattern across those citations usually explains the result better than anything else, because it shows what these tools currently treat as evidence in your field.
Last, we look back at your site and your listings with the answers in hand: whether your pages state plainly what you do and where you do it, whether AI crawlers are allowed to read them, and whether your business details agree everywhere they appear. That is where the recommendations come from.
Request an AI visibility check.
Tell us the business, the city you sell in and the competitors you watch. We put a fixed set of buyer questions to the assistants your customers actually use, then write up what came back: the prompts, how each platform answered, where you stood beside the names you gave us, and what to change first.
One check, one report, no retainer and nothing to cancel. We cannot promise you a place in an AI answer: nobody controls what these tools recommend, and the same question can come back differently next week.
How we work.
- Step 01.
Build the prompt set from what you tell us
The form asks for your business, the area you serve and a few competitors you would hate to lose work to. From that we write the 20 prompts, mixing service questions, near-me questions, comparison questions and direct requests for a recommendation. The finished list travels with the report, so you can see exactly what was asked.
- Step 02.
Run every prompt across the four tools
Each prompt then goes to ChatGPT, Perplexity, Gemini and Google AI in a clean session, and we keep the full reply rather than a summary of it. Working this way makes the findings quotable: every claim in the report points back to a specific answer a specific tool gave on a specific day.
- Step 03.
Read who was named and what was cited
We go back through the saved answers and mark them up: where you appeared, where a competitor appeared in your place, what each answer asserted about the businesses it named, and which sources it cited or described. Counting across the whole set turns a pile of replies into a picture of where you stand.
- Step 04.
Work out what explains the result
This is the part that takes the longest. We look for what is driving the answers and what could plausibly change them, then write up the findings, the likely causes on your own site and listings, and a short ordered list of what to address first. Anything we cannot honestly promise is marked as such.
- Step 05.
Send the report and talk it through
The report arrives as a document you keep, holding the prompt list, the answers and the recommendations. We will walk you through it on a call if you want one, and you are welcome to hand the whole thing to whoever handles your marketing. Nothing is signed and no follow-on work is assumed.
Questions, answered.
How is the check actually run?
We ask 20 prompts in ChatGPT, Perplexity, Gemini and Google AI, in clean sessions so no account history steers the replies, and save what each tool says. Then we read those answers as a set and write up who was named, what was claimed and which sources the tools cited.
How are the 20 prompts chosen?
They come from what you tell us: the services you want more work in, the area you serve and the competitors you name. We phrase them the way buyers ask, mixing broad service questions, near-me questions, comparisons and direct requests for a recommendation. The full list is printed in the report.
What do I actually get back?
A written report covering all 20 prompts: which tools named you and how often, which competitors were named in your place, what the answers claimed about your business, which sources they leaned on, and what on your site or listings looks like the cause. A call to walk through it is included.
Why do two people asking the same thing get different answers?
Because these tools write a fresh answer every time instead of reading a fixed list. Location, account history, the exact wording and the model version all move the result, and the same prompt run twice can differ. That is why we run a whole set and read the pattern rather than one lucky reply.
A competitor gets recommended and we do not. What now?
Start with what the answers cited when they named them. It is usually a source these tools trust and can read easily: a complete Business Profile, a body of reviews, a directory listing, or pages that say plainly what the business does and where. The report names which of those is thin on your side.
Is this the same as rank tracking?
No. Rank tracking watches your position in a list of search results. This reads generated answers, which name a handful of businesses and cite a few sources, with no positions to count. A business can rank perfectly well in Google and still go unmentioned when someone asks an assistant.
How often is it worth running again?
When something has changed: you have acted on the recommendations, added a location or a service, or a competitor has clearly moved. The tools also revise their own models and sources without notice. There is no schedule and no plan to join, so ask for another check when it would tell you something new.