Methodology
How we compute a BienCheck score
No black box here. This is exactly how public data turns into a score, a recommendation and a negotiation margin.
1. Official sources
DVF (sales), Géorisques (risks), ADEME (EPC), INSEE (territories). Nothing scraped off listings.
See the sources2. Per-town aggregation
For every French commune we compute robust indicators: median €/m², EPC mix, per-hazard risk scores.
3. Proprietary scoring
Four weighted pillars (market, risks, energy, liquidity) feed the overall score out of 100.
Score breakdown4. Confidence margins
When a town has few sales or few EPCs on file, we lower the confidence shown rather than inflate the score.
5. Refresh cycle
Public datasets update continuously. Towns are reprocessed in rolling batches to stay current.
Why a median, not an average?
For price per m², the median is less skewed by one weird sale (a 12,000 €/m² loft in a 2,800 €/m² town). We also show the number of transactions so you can judge how trustworthy the median is.
How we combine the pillars
Market 30%, risks 25%, energy 20%, liquidity 25%. Those weights aren't pulled out of a hat: they match the average resale-price impact we observe in the DVF database.
What we don't include (yet)
Noise, hyper-local air quality, detailed HOA (real service-charge amount), votes for major works. The data exists but isn't always open commune by commune. We'll add it as soon as it's available nationwide.
Displayed vs raw data
We always show the official figure when it exists for the town. When it doesn't, we clearly say "data unavailable". We never invent a default value, and we never show zero by default.