DVF France Scraper — French Real Estate Transactions API with Trustworthy €/m² (since 2014)

Search and analyze French real estate transactions from the official DVF (Demande de Valeurs Foncières) database — every property sale registered in France since 2014. Unlike raw DVF dumps, this DVF France Scraper collapses multi-lot mutations and computes a trustworthy price/m² (mono-dwelling only, outliers stripped), then adds comparables, AVM price estimation, an AI market summary, and DPE + risk enrichment. From $0.002 per transaction.

⭐ Bookmark this DVF France Scraper — Apify ranks actors by bookmarks, so it directly boosts the visibility of this scraper on the Apify Store.

What is the DVF France Scraper?

The DVF France Scraper queries the official DVF — Demande de Valeurs Foncières dataset, France's open registry of every taxed real-estate transaction since 2014. It returns sale price, surfaces, type, rooms, lots, GPS, and cadastral parcels — but it does not hand you the raw, defect-ridden rows. The DVF France Scraper runs a cleaning pipeline first:

• Multi-lot dedup — one valeur_fonciere is repeated across every parcel/lot of a sale in raw DVF. The DVF France Scraper groups by id_mutation into one clean row per real sale, summing distinct surfaces.
• Trustworthy €/m² — computed only on mono-dwelling mutations (a single apartment or house, no other building), price > 1 €, surface > 0, with outliers outside €200–30,000/m² stripped. Every other sale still ships, flagged m2_eligible: false.
• Median-based analytics — median price and median €/m² (robust to VEFA / en-bloc outliers), with a 10-year evolution and a YoY %.
• VEFA flag + coverage disclaimer — new-build-on-plan sales are flagged; the Bas-Rhin (67), Haut-Rhin (68), Moselle (57) and Mayotte (976) coverage gap is surfaced.

Backed by the free official files.data.gouv.fr/geo-dvf open data + ADEME + Géorisques APIs. No API key, no quota, no proxy.

Use cases

• Real estate valuation (AVM) — get a defensible €/m² estimate for any French address from clean nearby comparables, with a confidence tier.
• Investor deal sourcing — find every clean sale in a commune under a €/m² threshold, sorted by price/m², to spot under-market deals.
• Notary / agent comparables — pull the N most-similar recent sales for a price-comparable report.
• Market research — track median €/m² and volume per commune over 10 years, with an AI-written market brief.
• Due diligence — annotate each commune with its Géorisques risk profile (flood, seismic, radon) and its ADEME DPE energy-label mix.

Input

Only departement is required. Provide codeInsee for the fast path — the actor then downloads a single tiny commune file instead of the whole department.

{
  "departement": "92",
  "codeInsee": "92012",
  "typeBien": "appartement",
  "prixMin": 200000,
  "prixMax": 500000,
  "nbPiecesMin": 2,
  "nbPiecesMax": 3,
  "dateDebut": "2023-01-01",
  "dateFin": "2024-12-31",
  "sortBy": "prix_m2_desc",
  "maxResults": 200,
  "includeComparables": true,
  "computeAvm": true,
  "avmSurface": 65,
  "analyzeMarket": true,
  "enrichRisks": true,
  "enrichEnergy": true
}

Key filters: codePostal, rue, latitude/longitude + rayonKm (GPS radius), typeBien (appartement/maison/terrain/local/dependance), surfaceMin/surfaceMax, surfaceTerrainMin/Max, natureCulture, nbLotsMin/Max, excludeVefa.

Output

First record = analytics (median price, trustworthy median €/m², 10-year evolution, distribution) + any premium blocks (comparables, estimation_avm, analyse_ia, enrichissement_communes). Following records = one clean sale each:

{
  "id_mutation": "2024-123456",
  "date_mutation": "2024-09-12",
  "nature_mutation": "Vente",
  "vefa": false,
  "valeur_fonciere": 412000,
  "prix_m2": 6299,
  "m2_eligible": true,
  "type_local": "Appartement",
  "types_biens": ["Appartement"],
  "surface_reelle_bati": 65.4,
  "nombre_pieces": 3,
  "surface_terrain": null,
  "nombre_lots": 1,
  "nb_dependances": 1,
  "adresse": "12 RUE EXEMPLE",
  "code_postal": "92100",
  "commune": "Boulogne-Billancourt",
  "code_commune": "92012",
  "latitude": 48.8339,
  "longitude": 2.2400,
  "distance_km": 0.42,
  "nb_parcelles": 1,
  "id_parcelle": "920120000AB0123"
}

Pricing

PAY_PER_EVENT. Base data is charged per clean sale; premium features are opt-in and billed only when enabled.

Event	Price	When
transaction	$0.002 / mutation	per clean, deduplicated sale returned
comparables-report	$0.02 / run	includeComparables: true
avm-estimate	$0.05 / run	computeAvm: true
enriched	$0.01 / commune	enrichRisks / enrichEnergy: true
ai-market-summary	$0.03 / run	analyzeMarket: true (paid Apify plan)

Analytics, the bookmark CTA, and validation/empty records are never charged.

FAQ — DVF API alternatives

Why not just use a raw DVF dump? Raw DVF repeats one sale price across every parcel of a multi-lot mutation, so a naïve price ÷ surface gives an inflated, wrong €/m². This DVF France Scraper deduplicates by mutation and computes €/m² only on mono-dwelling sales — the number you can actually trust.

vs MeilleursAgents / SeLoger valuation? Those use proprietary black-box models. The DVF France Scraper gives you the raw declared sale prices plus a transparent comparables-based AVM you can audit.

Does it cover Mayotte / Alsace-Moselle? Mayotte (976) is not in the geo-DVF distribution, and Alsace-Moselle (57/67/68) uses the Livre Foncier, not the Cadastre — records there are not publicly available. The actor surfaces this in the analytics record.

Real-time? DVF is updated ~every 6 months by the DGFiP; the latest semester is usually ~6 months delayed.

---

⭐ Found this useful? Bookmark this DVF France Scraper — it's the strongest signal for Apify Store ranking.

Related actors — chain them with DVF on the commune / address / GPS join key

• ADEME DPE Scraper — energy diagnostics (class A–G) for the same properties.
• Géorisques France Scraper — flood / seismic / radon / industrial risk per address (ERRIAL).
• Pige Immo FR — current listings (LBC, SeLoger, PAP, Bien'ici) to compare ask price vs sold price.
• Géocodeur INSEE — address → INSEE code + GPS.

License: MIT · Author: dltik