Coming soon · for founders
Your market research, measured — not imagined.
The real demand on Google, the competitors capturing it, and what AI assistants tell your future customers. Every number carries its source and its date.
The study is not on sale yet. Leave your email and we'll let you know when it opens.
- Search volumes actually pulled, not guessed
- Six families of AI assistants queried
- A source and a date on every number
Six sections, and where each number comes from
Demand
“How many people search for this every month?”
The full vocabulary of your market, query by query: monthly volumes, twelve-month trend, seasonality, and the split between “I'm researching”, “I'm comparing” and “I'm buying”.
measuredWeb competition
“Who is capturing that demand today?”
The sites that actually rank for those queries, their estimated organic traffic, how concentrated the market is, and the known players who are nowhere to be found on Google.
measured and estimatedAI recommendations
“What do ChatGPT, Claude and Gemini tell your future customers?”
The same questions put to six families of assistants, without naming any company: who gets recommended, on which criteria, at what perceived price, with which objections.
measuredCustomer voice
“What do the incumbents' customers complain about?”
Public reviews of the players in your market: frustrations, reasons to stay, reasons to switch. Real quotes, with their link and date — never what a model imagines people think.
sourcedEnvironment
“Which rules, which trends, which barriers to entry?”
Market context established by web-connected models: every claim comes with the URL that backs it. No link, no entry in the report.
sourcedTAM / SAM / SOM
“What market size do I put in my business plan?”
A calculator where the volumes are ours and measured, and the rates stay your assumptions — editable, and labelled as assumptions in the report.
your assumptionsEach section carries a signal — favourable, contested, unfavourable, or inconclusive when the measurement isn't enough to call it. No overall score out of 100: a market doesn't come down to a single number.
You frame it in two screens, the rest runs on its own
Your idea in one sentence
You describe what you want to launch, in which country and language. Nothing else to fill in.
Your market, as we understand it
We propose the segments, the incumbents and the customer profiles; you tick, correct and add. It's the only point where we ask anything of you.
The report
The data is pulled, the assistants are queried, the reviews are read. You get a dated report you can share and print.
How this differs from market research written by an AI
Ask an assistant for market research and you'll get a well-written, well-structured, plausible document. And an unverifiable one — none of its numbers were ever pulled.
| What an AI writes | What we measure |
|---|---|
| “A multi-billion market, growing fast.” | The real search volumes of your market, query by query, with twelve months of history. |
| A list of plausible competitors, some of them invented. | The sites that actually rank for those queries, with their estimated traffic and the share taken by the top 3. |
| “Customers are mostly looking for simplicity.” | Real customer reviews, quoted word for word, with their link and date. |
| Nothing about what assistants recommend — though your customers ask them. | The answers actually given by six families of assistants to your future customers' questions. |
| No source, no date, nothing you could measure again. | A source and a date per number, plus the coverage rate of the measurement, stated plainly. |
Before you sign up
Frequently asked questions
When will the market research be available?
We're building it. The sections above are the ones our engine already knows how to measure; we're assembling them into a study step by step. Leave your email to hear about it at launch — it commits you to nothing.
How is this different from market research written by an AI?
A language model produces plausible numbers: it doesn't measure them, and it can neither source nor date them. Here every number comes from a measurement — real search volumes, domains that actually rank, answers AI assistants really gave — and is shown with its source and its date.
Where do the demand figures come from?
From Google search volumes, pulled query by query from a professional data provider, with twelve months of monthly history. The study also states how much of your market's vocabulary our queries cover: an incomplete measurement is flagged as such.
What does “what AIs recommend” mean?
We put the questions your future customers will ask to the main assistants — ChatGPT, Claude, Gemini, Le Chat and others — without ever naming a company, then read every answer systematically. It's the measurement Opinion Radar already runs for brands, applied to your market.
Does my idea stay confidential?
Yes. A study belongs to the account that ordered it and is never published: the public examples are produced by us, on markets we pick ourselves. The idea you leave here is only used to shape the product; it is displayed nowhere.
What about a local business?
The first version targets online or nationwide projects: that's where demand measures best. If yours is a local business, say so in the form — that's what will tell us whether to build it next.
How much will it cost?
It will be a one-off purchase, no subscription: one study, one payment. The price will be announced at launch, and the waiting list will hear it first.
Get notified at launch
We're building the study section by section, measuring real markets as we go. Tell us yours: the ideas we receive here decide which markets we measure first.