
A free, addictive game that teaches you to value a property like a pro. Real Paris listings, real sale prices pulled from DVF open data — AI-generated visuals over a real price engine. Twenty arrondissements, one question: tu tiens combien?
A free, addictive game that teaches you to value a property like a pro. Real Paris listings, real sale prices pulled from DVF open data — AI-generated visuals over a real price engine. Twenty arrondissements, one question: tu tiens combien?
DVF publishes every real Paris sale; nobody had made it playable. I built the price engine first, then wrapped it in a guessing game so the feedback loop teaches valuation instead of merely reporting it. Concept, design and build were mine.
A round is not one guess. It is seven phases — price, defects, strengths, energy rating, fees, era, then the reveal — and each one moves the number you are committing to. That is the design: valuation is not a single judgement, it is an accumulation of them.
The feed is deliberately social. Listings scroll like a vertical video feed, with likes and a running bankroll, because the format people already scroll is the format that teaches without feeling like homework.
Every price is real. DVF publishes every recorded sale in France; the game is a wrapper that turns that public record into a feedback loop.


The whole proposition in a single screen: real DVF sale prices, twenty arrondissements, free. The three proof points sit under the CTA rather than above it — the question earns the click first. The listing carries exactly what a valuer would work from — arrondissement, type, surface, floor, lift, balcony, view — plus the neighbourhood median per square metre. Then two candidate prices, and no way to move on without picking one.
No account, no onboarding, no tutorial. The headline states the game, the button starts it, and the legal note about AI-generated visuals sits where it belongs — visible, not hidden.


A second mechanic on the same listing: a price is shown, and the only question is whether the real one sat above or below it. A countdown bar runs and a speed multiplier climbs, so answering fast is worth more than answering safe. Ten markers, five columns — notary, agency, works, vendor file, energy survey — and the instruction that ten markers equal one hundred per cent of the hidden cost. The more a line really costs, the more it deserves.
The multiplier reached 2.3× before the call was made. Speed is scored because a professional valuer does not deliberate for a minute on a property they know — fluency is the point, not arithmetic.
The real sale price, the precision percentage, the gap and a letter grade — then the receipt underneath: final price against real net-to-vendor, bonus, bankroll, and every hidden cost itemised down to the diagnostics line.
A wrong answer is more instructive than a right one here, because the breakdown shows exactly which assumption was wrong. That is the feedback loop the price engine exists to serve — the game is the interface to it.
Sector: Real estate · Role: Concept · Design · Build · Data: DVF open sale prices
Designed, engineered & shipped by Roderic Andrews — Product Designer & Design Engineer
Everything on this page — the engineering, the design, the positioning — was built by one operator. That's the offer: a product designer who writes the code and wires the growth. No hand-offs, no telephone game, no agency margin.
Design · Engineering · Marketing · Intelligence · Video · Growth
I'll walk you through how it was built — live, with your questions.
Thirty minutes. The build, end to end. Replies same day.