The question, verbatim
The exact wording asked, so you can repeat it live in front of a client and get a comparable answer.
The method
Real phones, in the real market, answering the real question.
You are going to be asked how this works by a client who is paying you for it. This page is the answer you can give them without needing us in the room.
Why the device matters
An AI answer is not one thing.
Ask the same question from two phones in two cities and you can get two different lists of businesses. The answer varies by location, by device, by account, and by the moment it was asked.
So a result pulled from an API, a headless browser, or a proxy with a location header attached is not the result your client’s customer would have seen. It is a plausible answer to a slightly different question.
We put a real consumer phone in the client’s market and photograph what it said. That is the whole difference, and it is the reason the captures survive contact with a sceptical client.
What actually happens to an account.
You add a client and we agree the query set
Service, market, and the buyer questions that actually matter for that business. The query set is fixed in writing before work starts, because it is also what the billing milestone is measured against. An agency should never be surprised by an invoice.
We capture the baseline on real devices
Real consumer phones, physically in the client’s market, running the real apps. We record what each engine said, when, and on which device — before any work is done. That baseline is what every later claim is measured against.
We work the third-party footprint
AI engines choose which businesses to name based largely on what the wider web says about them, not on the business’s own site. That is the surface we work. We do not touch your client’s website, and we do not interfere with the SEO you are already delivering.
We re-capture on the same standard
Same query set, same market, same device standard, repeatedly. A movement claim is only meaningful against an identical method — if the measurement changes, the result is not a result.
A milestone lands, and only then do you get billed
When the client is named in the agreed query set, we capture the evidence and that client joins your next invoice. Until then they cost you nothing, however long it takes.
What every capture has to carry.
If a capture is missing any of these, it does not go in a report and it cannot trigger a milestone.
The engine and the moment
Which of ChatGPT, Perplexity, Gemini produced it, and the timestamp. AI answers move; an undated screenshot proves nothing.
The market it was asked from
A real device in the client’s own market — the variable that changes the answer most and is faked most often.
Who was named instead
Not just whether your client appeared. The competitor list is usually the part that sells the retainer.
Three real pairs from the fleet
These describe what an AI engine answered on a given date, captured on a real device. They are observations of public AI output, not customer endorsements.
Where we stop
We publish proof, not the playbook.
You get the evidence standard, the methodology brief, and every capture. You do not get the exact prompt set or the device method, and neither does anyone else.
That is not us being cagey with a partner. A method that is published is a method that is gamed, and the moment it is gamed your client’s results stop being worth anything. The boundary protects your margin before it protects our moat.
Honest limits
What this does not do.
- We cover three engines, not ten. Claude and Copilot are not tracked today. When they are, they are included at no extra cost.
- We publish proof, not the playbook. You get the evidence standard and the methodology brief; the exact prompt set and device method stay ours.
- We are not an SEO replacement. This answers the AI-search question your clients are already asking you.
- We cannot promise a timeline. The median across our published, verified stories is 14 days, but a competitive category with little existing third-party coverage takes longer. Since you are not billed until a client is named, that affects when you start paying rather than what you pay.
- We do not guarantee a specific position or wording in an answer. AI output is generated, not ranked — anyone promising you a fixed placement is describing something that does not exist.
More questions in the partner FAQ, and the commercial terms are on partner pricing.
See it on your own accounts
Judge the method on your own accounts.
Bring two or three real client names to the call and we will walk the captures with you: the question, the engine, the timestamp, the market, and who was named instead.