White-label AI visibility reports for agencies
We ask ChatGPT, Google's AI Overviews, Gemini and Perplexity the questions your client's buyers actually type — in their own language — and turn the answers into a report you send under your own name. Monthly or weekly, one flat price per client.
Get a free scan for one clientNo signup. We run it by hand and email you the report within 2 days.
Invisible in 42 of 42 questions. Two engines cannot confirm the clinic exists; one answers about a same-named practice in Los Angeles. Three weeks later every competitor had gained ground and the clinic was still at zero.
How it works
That is the whole order form. We write ~50 buying-intent questions their customers actually ask, in the market's own language, and work out the competitors from the answers. Send us your own list instead if you prefer.
ChatGPT with browsing, ChatGPT without, Gemini and Perplexity through their official APIs, plus Google’s AI Overview read from the live search page in the client’s own country and language. We record who gets recommended, which sources the answers are built from, and what the engines say word for word.
A plain-language summary on page one, share of voice per engine, the questions where your client is invisible, the sources that decide the answers, and what to publish. Your logo, your name.
Two things worth knowing up front. Answers differ enormously between an engine that searches the web mid-answer and one answering from memory — every report says which mode each number came from. And a single run is a snapshot: the second run is where you can show a client what moved.
Engines
Every engine is in every plan. They disagree with each other constantly, and the disagreement is most of what a client needs to see. Google publishes no API for its AI Overview, so that one is read from the live search page in the client’s own country and language.
What a user is told when ChatGPT searches the web mid-answer.
The same question with no web access — what the model already believes.
The answer Google writes above its own results, read from the live search page.
Grounded in Google Search — often knows a business and still will not recommend it.
Answers built from cited sources — the clearest read on which pages decide them.
Model names change — engines get renamed and retired mid-quarter, and one we used was shut down with no notice in August 2026. Every report names the exact model behind each number on the day it ran, so a figure can always be traced back to what produced it.
Pricing
Most tools charge a platform fee and then meter you per prompt or per project, which punishes exactly the agency that has ten small clients rather than one large one. Here you pay per client, and that is the whole bill.
One report a month — the cadence agencies actually report to clients on.
A report every week, for clients who are actively moving the numbers.
Free scan
Tell us one client's domain and market. We run a one-off scan and email you the full report — yours to keep, and to show them. No card, no signup, no call.
Reports are produced by hand right now, by the person who built the tooling — you are buying the analysis, not a dashboard login. That is also why the free scan is a real one.
Questions
The share of a set of buying questions in which an answer engine names the brand. We ask around fifty questions a real buyer would type, count the answers that mention the client, and put competitors on the same scale. It is an aggregate percentage across many questions — never a “position in AI”. Repeat runs of the same question return different lists, so a rank would be fiction, and we do not publish one.
Yes. The AI Overview is the answer Google writes above the classic blue links, and for many searches it is the only thing a person reads. Google publishes no API for it, so we read it from the live search page, set to the client’s own country and language, alongside the four chat engines.
Five: ChatGPT with live web search, ChatGPT answering from memory alone, Google AI Overviews, Gemini and Perplexity. The gap between the browsing and the memory answer is usually the single biggest finding in a report — a small brand routinely scores double digits in one and zero in the other.
A rank tracker tells you where a page sits in a list of links. An answer engine returns no list: it returns one answer naming two or three businesses, and everyone else is simply absent. We measure whether the client is named at all, which sources the engine built the answer from, and what it says word for word.
Yes, and that is the point. Questions are written natively for the market — Ukrainian, Polish, Lithuanian and others — never machine-translated from an English keyword list. The engines recommend different businesses depending on the language they are asked in, so an English-only tool measures the wrong thing outside English-speaking markets.
Yes — white-label from the first client. Your logo, your name, no mention of us. A flat $15 per client per month for monthly reporting or $29 for weekly, with no platform fee and no per-prompt meter.