How the numbers are made

What these numbers cannot tell you

Start here, because it matters more than the sources. A commune that scores 90 for you has cleared the things we can measure. Those are not the things that decide whether a move works.

• Whether the street is pleasant to walk down at eight in the evening.
• Whether the mayor is about to approve a development behind the house you are looking at.
• Whether the doctor listed three kilometres away is taking new patients. Most are not.
• Whether the English speakers we counted are people you would want to know.
• Whether the school is good, as opposed to close.
• What the village is like in February, when the shutters are closed and the café shuts at two.

We can rule out the places that fail on price, climate, healthcare or distance. That is worth doing: it turns 34,735 into a shortlist of twelve. It is not the same as knowing where to live, and nobody should treat a score here as a decision. Go, in low season, and stay a week.

How a score is built

Thirteen dimensions. For each one, every commune is ranked against every other commune and placed on a 0–100 scale by its rank, not by its raw value. Your weights decide how much each dimension counts; the score is their weighted average. Move a slider and the whole country re-ranks.

Ranking by position rather than by value is deliberate. House prices run from €200 to €12,000 a square metre, and the top end would otherwise swallow everything else. A rank says how a commune sits among the others, which is the question you are actually asking.

There are two ways to be ranked, and the choice changes the answer. Ranked against all of France, every city looks unsafe next to thirty-four thousand villages, and every village looks short of shops. Against places of similar size, villages are compared with villages and cities with cities. Neither is the truth. The switch is in the profile panel and it is worth using both ways before you trust a result.

Housing is scored one of two ways, and you choose which. Buying ranks what homes sell for, with the property tax owners pay on top. Renting ranks rents instead and leaves property tax out, because in France it is the owner's bill. The budget follows the same switch: a price per square metre when you buy, a monthly rent for the size of home you set when you rent.

Must-haves work differently. They do not affect the score at all: they remove communes from the map. A commune either clears your maximum drive to a hospital or it does not.

Where each number comes from

All of these are French public datasets, published under open licence. None of it is bought, scraped or estimated by a third party we cannot audit.

Travel times are estimates, and we say so

Every drive time on this site, whether to a hospital, a station, an airport or the sea, is a straight-line distance converted to minutes by a curve, with a penalty in mountains. It is not a route. It does not know about estuaries, one-way systems, or the fact that the far side of a lake is an hour away. Trouville is three kilometres from Le Havre and an hour's drive.

It also picks the nearest target as the crow flies, which on a peninsula or in a valley is often not the one you would actually reach first.

We are replacing it with real routing over the OpenStreetMap network. Until that ships, treat every minute figure as an order of magnitude: good for comparing two communes, wrong for planning a morning.

Estimated prices, and how to spot them

Of 34,735 communes, 14,029 have enough recorded sales for a reliable median. 19,101 have too few, and take the median of their eight nearest neighbours. 1,605, which is all of Alsace and Moselle, have no usable sales data at all: those three departments keep a separate land registry that never entered the national file. Their prices are modelled from local rents, and the model's typical error is 15%.

A price estimated from neighbours is a reasonable guess for a village next to a town with an active market. It is a poor guess for an isolated village. If a price matters to your decision, check it against a local agent before you act on it.

Numbers we removed rather than guess

Five fields are empty on this map today. Each was showing something, and each was wrong.

• Distance to a care home was a copy of the distance to a hospital, for every commune in France, because the establishment list did not identify care homes and the code fell back without saying so.
• Frost days was a formula applied to the winter average, not a count. It gave Bordeaux, Nantes and Toulouse zero frost days. They record about thirty.
• Clay soil came from prefectural hazard listings, which are an editorial choice rather than a soil map: Seine-et-Marne came out flagged at 100%, Haute-Garonne, among the most exposed departments in France, at 0%.
• Earthquake zone was never populated.
• Pharmacies nearby counted within a radius that worked out at three kilometres from the village centre, so 88% of communes read "none close", including Chamonix, which has five.

They will come back when they are measured rather than inferred. An empty field is annoying. A plausible wrong number is worse, because you will believe it.

What we fixed in September 2026

We audited the whole dataset against reference values and found errors we had been publishing. They are listed here because a methodology page that only describes the intended behaviour is worth very little.

• Paris, Lyon and Marseille were shown as 13, 11 and 19 minutes from a hospital. The establishment file codes those three cities by arrondissement, and every one of their hospitals was dropped in the join.
• 1,060 communes in Alsace and Moselle were priced from their nearest neighbours across the Vosges. Colmar was €1,690 a square metre, Obernai €1,000.
• Bordeaux was 13 minutes from the sea, and Blaye was a coastal commune, because the coastline file follows the Gironde inland. The coast is now the outer edge of the communes themselves.
• Montpellier was 35 minutes from a railway station. So were Lisieux, Menton, Gap and Le Puy. That was 33 of the 170 station towns we checked by hand.
• 36 communes had no shape on the map at all, including Île-de-Sein and Rochefort-en-Terre.
• Lyon and Bordeaux were listed as airports with US flights. Neither has one.
• A budget left at "No limit" still ruled out Saint-Jean-Cap-Ferrat and Val-d'Isère, whose prices run past the end of the slider, and a sunshine minimum left at "Any" ruled out three communes below 1,500 hours. A slider at its end now rules out nothing.

Tell us when a figure is wrong

There is a button on every commune for this. It sends us the commune, the figure you were looking at and its value, so we can find it. We read them.

The part this map cannot do is the part we do by hand: what a place is actually like, what Americans usually miss about it, what to check before booking a trip. Ask us at easyfrancenow.com.