leboncoin Scraper

A leboncoin scraper that returns French classified ads with the price and its currency, the seller type and company number, the exact location down to coordinates, every photograph, and the full structured attribute set the seller filled in. Promoted adverts the site injects into each page are labelled rather than passed off as matches.

France's dominant classifieds site, from cars to phones to property, on one input. No login, no API key, no browser.

What you get

• The two injected adverts per page named, not silently mixed in. leboncoin drops two promoted ads into every result page, and on a search for "iphone" they are cars. They are excluded by default and, if you ask for them, arrive flagged as is_injected_ad. This is the single most common way a leboncoin dataset ends up with an unexplained €37.599 row in a phone price distribution
• Deduplicated across pages. Page 1 and page 2 of the same search overlap, because those promoted rows repeat. Rows are returned once
• The full attribute bag the seller filled in, keyed by the site's own field names, whatever the category. On a car that is make, model, trim, registration year, mileage, fuel, gearbox, power, import status and the vehicle history report link. On a flat it is the surface area, the room count and the energy rating
• Mileage with its unit and price with its currency on every row. Vehicle rows carry vehicle_mileage and vehicle_mileage_unit as separate fields, so the number joins cleanly against any other market without a units audit first
• Professional sellers identified properly, with the shop id and the French company registration number, so a dealer's whole inventory can be grouped and cross-checked against the company register
• Location to coordinates: town, postcode, department with its id, region with its id, latitude and longitude
• Every photograph at full size, not just the thumbnail
• The filters the site says it applied, reported in the run summary next to the ones you asked for, so a filter that quietly did nothing is visible instead of invisible
• A refused run and an empty search are never the same answer. One exits successfully saying nothing matched; the other says the site did not serve the page

Why scrape leboncoin

leboncoin is where France buys and sells: 755.936 cars, 877.772 properties for sale, 509.679 phones and connected devices, and over 23 million home and garden listings as this was written. For price research, sourcing, or tracking a dealer's stock in France, there is no second choice.

It also publishes more structure than most classifieds sites. Sellers fill in typed fields per category, and those fields come back with the row rather than having to be mined out of free text. A car ad arrives with its mileage as a number, its gearbox as a value and its registration year as a year, which is the difference between a dataset and a pile of French prose.

Input

Field	Type	Required	Default	What it does
query	string	no	–	What to search for, as you would type it on the site
category	select	no	all	Cars, property, phones, fashion and 16 more
url	string	no	–	A leboncoin search URL to read instead of the fields below
price_min / price_max	integer	no	–	Price bounds in euros
seller_type	select	no	any	Private sellers, professionals, or both
locations	string	no	–	d_75 for a department, r_12 for a region
sort	select	no	relevance	Newest, oldest, price up, price down
include_injected_ads	boolean	no	false	Return the promoted rows too, flagged
limit	integer	no	50	Ads to return, 1 to 3500
retries	integer	no	5	Retry attempts per page

Output

{
  "position": 1,
  "list_id": "3270620909",
  "url": "https://www.leboncoin.fr/ad/voitures/3270620909",
  "title": "Golf GTD",
  "price": 14900.0,
  "price_currency": "EUR",
  "price_was": null,
  "category_id": "2",
  "category_name": "Voitures",
  "published_at": "2026-09-15 10:22:41",
  "is_injected_ad": false,
  "seller_type": "private",
  "seller_name": null,
  "seller_siren": null,
  "city": "Dax",
  "zipcode": "40100",
  "department": "Landes",
  "department_id": "40",
  "region": "Nouvelle-Aquitaine",
  "region_id": "2",
  "country": "FR",
  "latitude": 43.7102,
  "longitude": -1.0533,
  "images_count": 9,
  "vehicle_brand": "Volkswagen",
  "vehicle_model": "Golf",
  "vehicle_registration_year": "2016",
  "vehicle_mileage": 160000,
  "vehicle_mileage_unit": "km",
  "vehicle_fuel": "Diesel",
  "vehicle_gearbox": "Automatique",
  "vehicle_power_hp": "184 Ch",
  "attributes": { "brand": "Volkswagen", "regdate": "2016", "mileage": "160000 km", "...": "..." }
}

Use cases

Used-car pricing in France. Set category to cars, add a price band, and every row comes back with make, model, registration year, mileage, fuel and gearbox as typed fields. That is a valuation dataset straight out of the run, and seller_type separates the private market from the trade market, which price differently for the same car.

Tracking a dealership or a shop. seller_store_id groups a professional's entire inventory and seller_siren ties it to the company register. Run it on a schedule and what arrives, what leaves and what is being discounted is visible without a single manual page view.

Regional price differences. locations takes a department or a region, and every row carries its own department, region and coordinates. Running the same query across departments gives you the geographic price surface for anything on the site.

Property research. Property ads carry their surface area, room count and energy rating in the attribute bag, with coordinates on the row. That is enough to build a per-square-metre map of a commune without a second data source.

Clean price distributions. Because the injected promoted rows are excluded by default, the prices you get are the prices of things that matched your search. On a phone search that removes two cars per page, which is the difference between a median that means something and one that does not.

How it compares

	this actor	clearpath/leboncoin-api	silentflow/leboncoin-scraper-ppr
Per 1.000 ads	$1,10	$1,49	$1,50
Actor-start fee	$0,001	$0,009	$0,00005
Injected adverts labelled	yes	not stated	not stated
Deduplicated across pages	yes	not stated	not stated
Full attribute bag per ad	yes	yes	not stated
Company registration number	yes	not stated	not stated
Coordinates on the row	yes	not stated	not stated
Applied-filter echo in the summary	yes	no	no
Detail pages and seller profiles	no	yes	no

Honest about the other side: clearpath has 67 monthly users to our zero, a 28-input schema, and it also fetches ad detail pages and seller profiles, which this actor does not. If you need the full advert body and the seller's other listings, that is a real capability gap. What it does not tell you is which of its rows are promoted adverts rather than search results.

Pricing

Two events. ad costs $0,0011 per advert returned, which is $1,10 per 1.000. A $0,001 start fee covers the run itself. All pricing is pay-per-event, so you pay for rows you actually receive, and a run the site does not serve bills the start fee only. No per-compute-unit charges.

Limits and gotchas

• 3.500 ads per query. The site itself caps a search at 100 pages of 35, and the run summary reports both how many ads matched in total and how many pages were read. For a bigger harvest, split the query by department or by price band.
• locations needs a department or a region. d_75 and r_12 work. A bare town name with a postcode is accepted by the site, echoed back as applied, and matches nothing; the actor rejects that shape rather than handing you an empty dataset.
• Some filters the site accepts do nothing. The shipping filter is one of them: it is echoed back as applied and leaves the result count unchanged. It is not offered here. The summary always reports the filter set the site said it applied, so you can check any filter you reach through a pasted URL.
• A price of €1.000.000 is a real advert. Sort by price descending on any category and you will find placeholder and joke prices near the top. They are the seller's numbers, not a parse error.
• The attribute bag differs by category. A car and a sofa do not carry the same keys. The vehicle fields are promoted onto the row only when the ad actually has them, so a phone never comes back with an empty mileage column that reads as an unknown odometer.
• Prices are euros. price_currency says so on every row rather than leaving you to infer it from the domain.
• Ad bodies in the search result are short. leboncoin returns a truncated body on result pages. For the full advert text you need the ad page, which this actor does not fetch.

FAQ

Can I scrape leboncoin without an API key? Yes. There is no credential field in the input.

What is is_injected_ad? leboncoin injects two promoted adverts into every result page, and they frequently belong to a different category from the one you searched. This field marks them. They are left out by default, so the ads you receive are ads that matched.

Why do I get fewer rows than pages times 35? Because pages overlap. The promoted rows repeat across pages, and rows are returned once each. The run summary reports how many pages were read and how many ads came back.

Can I search one department only? Yes, with locations set to d_ followed by the department number, for example d_75 for Paris. A region works the same way with r_.

How do I follow one professional seller? Group by seller_store_id, which is stable, and cross-check seller_siren against the French company register.

What happens if a run returns nothing? The summary says which of three things happened: nothing matched, the location expression matched nothing, or the site did not serve the page. They are never reported the same way.

Related Actors

• Marktplaats Scraper: the same job for the Netherlands
• OLX Scraper: classifieds across Central and Eastern Europe
• mobile.de Scraper: the German vehicle marketplace