Not engagement surveys.
No NPS, no scores, no "how do you feel about leadership."
Stop chasing answers through endless meetings. Ask Confr one hard question — "Why are sales dropping in the South?". Confr sends private chats on Microsoft Teams. Each person points to the next, until we reach the source. You get the ground-level truth in one written report — every line cited.
In a big company, the answer lives in hundreds of inboxes, spreadsheets and shift conversations. No one person has all of it. Ask openly and people clam up. Ask quietly and you never reach the people who actually know.
Every other number is green. Customers are still complaining.
A live example. The CEO of a 2,400-person retailer asks one question. Five stages later, the boardroom has an answer — with receipts.
A 2024 routing rule sends supplier deliveries to the Midlands first1. The national-stock banner reads the Midlands hub only2. South store managers stopped re-ordering when their shelves stayed empty3. Sales collapsed. Stock never followed.
Deep is the hard, political question that's been bouncing around for months. Standard documents what one person knows before they leave. Quick answers one specific question.
Every line links to evidence. Each finding carries a confidence score. We rank possible causes too, and tell you what we ruled out and why.
A 2024 routing rule sends supplier deliveries to the Midlands first1. The national-stock banner reads only the Midlands hub2, so South store managers stopped re-ordering3. Demand signal collapsed. Stock didn't follow.
A real-shape Deep-tier report from an anonymised customer. Click to read it the way the board would — table of contents, citation popovers, evidence gallery, dropped findings, and the anonymity stamp.
No fine print. No "trust the AI." Every promise leaves a stamp on the deliverable, so you can check it.
Every claim is checked three independent ways before it lands as a finding: our own rules, a second pair of eyes, and a test designed to prove it wrong. Three voices saying the same thing in process work; three checks in root-cause work. Same promise either way.
Every line in the report links to a quote or a file. Nothing without a source — and nothing comes from the model alone.
Every cause we set aside or rule out comes with the reason written down. No quiet dismissals. If we couldn't get to the evidence, we tell you what we'd have needed and where we got stuck.
No vendor recommendations. No autopilot. No "based on this, you should." We find what's true. The decisions stay with you.
Below 50 people, we can't keep voices private — there aren't enough people for anonymity to mean anything. Above that, we get more useful the bigger you are.
Big enough that nobody has the whole picture. Spread out enough that what one team knows, another doesn't. Political enough that asking openly skews the answer.
EU-built. GDPR by default. Designed with works council approval in mind: every investigation has an opt-in flow, an audit trail, and a refusals log that records categories — not the names of who refused.
If you have fewer than 50 employees, we'll tell you up front. The product won't help you.
No NPS, no scores, no "how do you feel about leadership."
It runs on demand, for one specific question, then stops.
If the question is about a named person, we refuse it.
We find the root cause. We don't pitch software.
There aren't enough voices to keep anyone private. We'll tell you up front.
A reviewer at Confr reads every report before it ships.
No seat licenses. No platform fees. No annual commit. You pay what the AI costs us, plus 30% so we can keep building.
Your first investigation is on us. Up to 10 interviews. Quick or Standard tier. Real report. No credit card.
We pass through what the AI costs us — priced per token — with a 30% margin on top. That's it. No subscription, no minimum, no platform fee.
Examples only — not fixed prices. You pay for the tokens your investigation actually uses, at the rate above. A bigger one costs more; a smaller one costs less.
Itemised on every invoice — the exact tokens your investigation used, plus 30%.
Tell us the question your last three steering committees couldn't crack. We'll run the investigation. You don't pay until the second one.