How much should a property be discounted based on its DPE?

How does it work?

What 956,923 real transactions show

We matched almost a million sales with the energy certificate of the property sold. On houses, the price-per-square-metre gap between a G and a D rating reaches 22%. But the discount behaves very differently depending on property type and market.

Houses: a steady gradient from A to G

Median price per square metre by class, across 825,719 house sales. The regularity of the progression is what makes this result solid: it is not an isolated drop on energy sieves, it is a continuous scale.

Our benchmark: the Notaires study

The Notaires de France 2024 green value study isolates the DPE effect through hedonic regression — a method that neutralises other factors. It measures −25% on G-rated houses and −12% on flats. Our −22.2% on houses is very close, which supports the figure. On flats, however, their method is the reference, not ours: if you need one number for a flat, take theirs.

Where the discount bites hardest (houses)

This ranking covers houses only — for the reasons set out just above, a municipal ranking on flats would be meaningless. The gap widens in slack markets, where buyers have choice and can push down the price of an energy-hungry property. Across the municipalities our volumes allow us to publish, observed gaps exceed 30% in Tulle, Thiers or Mamers. Conversely, in a tight market an energy sieve still sells: the buyer has no alternative. Geography matters as much as the label.

Gap between the median of classes F and G combined and that of class D, in raw value, on house sales. 452 municipalities pass our publication thresholds (at least 50 reference sales and 30 energy sieves), out of 2,560 documented. Municipalities below those thresholds are not published: on small volumes, a 40% gap can be nothing more than a sampling accident.

The mistakes that ruin a negotiation

1. 1

   Applying the discount to an already-discounted price

   The calculator treats the price you enter as that of an equivalent D-rated property, then subtracts the class gap. If the seller has already cut the price because of the EPC, the same gap is removed a second time.

   What it changes: The reduction looks complete when part of it was already in the asking price — or the offer drops so low it is not even discussed. The useful reflex: first compare the asking price with recent sales of D-rated properties in the same area, then apply the discount only to that reference price.

2. 2

   Leaving the postcode field empty

   The postcode is optional, but without it the calculation falls back on the national average instead of the regional variant. Those variants exist only for classes E, F and G, never for whole buildings, and their coverage depends on property type: at most 34 departments for a G-rated house, 27 to 28 for an E or F house, and only 5 to 6 for a flat.

   What it changes: Everywhere else, the result stays a national average — which is what you should present it as in a negotiation. The markets where the label weighs least are precisely those where a national average turns against you.

3. 3

   Taking the EPC label as a certainty

   The calculator faithfully applies the letter you give it. Yet a certificate can contain internal inconsistencies between declared consumption, described insulation and the building’s characteristics, which make the label debatable.

   What it changes: A debatable letter distorts the whole negotiation, in either direction. Across our observable sub-sample of 689,279 certificates, 11.1% carry at least one major inconsistency once the sample is reweighted by department and vintage. That is a signal to check with the assessor before arguing, not a verdict on the certificate or its author.

4. 4

   Using an electric-heated EPC issued before 2026

   Since 1 January 2026, the primary-energy coefficient for electricity has fallen from 2.3 to 1.9. A home heated by electricity and surveyed before that date therefore carries a label computed with the old coefficient.

   What it changes: The same home can gain a class without any works at all once its certificate is updated. Entering the old letter inflates the discount claimed, and the argument collapses as soon as the other side notices.

What to do with this figure