🇫🇷 Sirene French Company Scraper — SIREN, NAF, address, directors

Turn the official French company register into a clean B2B list. Pick an activity code, a department or a creation-date window and get one row per company: SIREN, SIRET, NAF/APE code, head-office address with GPS, headcount band, VAT number, creation date and the registered directors from the RNE.

Straight from the French government's own open API — no login, no API key, no scraping wall.

What does this actor do?

It searches the Sirene base (every registered French company) and returns flat, joinable rows:

• 🏢 siren / siret — the 9-digit company id and the 14-digit head-office id
• 🏷️ companyName / legalName / tradeName — full name, legal name, trading name
• 🧾 vatNumber — the intra-community VAT number (FR…)
• 🏭 nafCode / nafSection — activity code, e.g. 10.71C bakery, 62.01Z software
• 👥 employeeBand — the INSEE headcount code decoded to a readable range (10-19, 250-499). Roughly half of all companies have no band on file and come back as Non renseigné
• 📍 addressFull, postalCode, city, department, region, latitude, longitude
• 📅 dateCreated — the registration date, so you can build a new-business feed
• 🚪 dateClosed + isActive — or flip it around and build a business-closures feed
• 👤 directors[] / primaryDirector — registered officers with role and birth year
• 🗂️ establishments[] — optionally the establishments that MATCH your search, each with its own SIRET + GPS. The API returns at most 10 per company and only the sites matching your filters, so this is not a full site list for multi-branch companies (use establishmentCount for the true total)
• 🔗 profileUrl — the public annuaire-entreprises page

Example input

{
  "nafCodes": ["56.10C"],
  "departments": ["75", "13"],
  "createdAfter": "2026-01-01",
  "employersOnly": true,
  "maxItems": 500
}

Every restaurant with a site in Paris or Bouches-du-Rhône registered since January that has employees. Note: the department filter matches any establishment, but each row carries the company's HEAD-OFFICE address — so a chain with a Paris branch and a head office elsewhere is returned with its out-of-region address (about 13% of rows on a single-department search).

Who is this for?

French SDR and growth teams building territory lists, expert-comptables and insurers prospecting newly registered companies, banks and merchant-services vendors, KYB and credit teams needing firmographics, and researchers who want a SIREN-keyed dataset they can join to other open data.

Field coverage — measured, not promised

Measured on the actor's own full run of one search — every bakery registered in Paris, 2,334 companies (run LsvSyg4tkZgnDgu2m). Two address rows below are quoted from the 200-company default sample of that same search (run AK5gC3NfusdINsLs3) and are marked as such. We do not headline the 200-row sample: the API returns its biggest companies first, so a first-page sample reads 10-20 points high.

Field	Fill	Note
siren, siret, companyName, addressFull, nafCode, dateCreated, isActive, legalForm, profileUrl	100%	the core row
postalCode, city, department, region	99.5%	on the 200-company sample
latitude, longitude	99.0%	government geocoding misses a few addresses (200-company sample; 99.8% on the 2,334-company run)
isEmployer	99%
employeeBand	100% present / 59% a real range	INSEE returns Non renseigné for the rest (200-company sample; 48% a real range on the 2,334-company run)
directors, legalName	86%
primaryDirector	79%
companySize	73%	INSEE has not assigned a catégorie d'entreprise to every record
vatNumber	62%	VAT is not issued to every micro-entreprise
employeeBandYear	51%	INSEE does not date every headcount band
tradeName	26%	most companies simply never register a separate trading name
acronym	1%	a sigle is genuinely rare
dateClosed	0% / 98%	null by definition on active companies; 39 of 40 on a companyStatus: "closed" run
establishments	off by default	populated for most filter-based searches when includeEstablishments is on; empty for some searches (e.g. a bare SIREN lookup), and capped at 10 sites per company by the API

Honest limits

• A single search is capped at 10,000 companies. That is the API's ceiling, not ours — it reports total_results: 10000 for any broader query. To go past it, slice the search: one run per NAF code, per department, or per creation-date window. The actor already runs each activity code × department combination as its own search.
• Date filters need a companion. The API rejects a date-only query, so createdAfter must be paired with an activity code, department, postcode or search term.
• Non-diffusible records are skipped. INSEE lets a company opt out of redistribution; those rows are excluded unless you explicitly set includeNonDiffusible. The run log tells you how many were skipped.
• Directors are personal data. They come from a public register, but GDPR still applies to what you do with them. Birth dates are deliberately not emitted — only the birth year — because a lead list has no use for them.

Pricing

Pay per result — you are charged for the companies you actually receive, and nothing else.

What	Price
Per company returned	$0.004

A 1,000-company territory list costs $4.

FAQ

Do I need an INSEE API key? No. This uses the open recherche-entreprises.api.gouv.fr service, which needs no key and no account.

Can I get companies created last month? Yes — set createdAfter and createdBefore, together with an activity code or department. That is the standard new-business lead feed.

Can I find companies that shut down? Yes. Set companyStatus to closed and you get a closures feed with dateClosed populated.

What is a NAF code? The French activity classification (APE). 10.71C is bakery, 56.10C fast food, 62.01Z computer programming. Pass as many as you like — each is searched separately.

How do I get more than 10,000 companies? Split the search. One run per department, or per NAF code, or per creation-date window. Each combination gets its own 10,000-row window.

Are the addresses geocoded? Yes, latitude and longitude come from the government's own geocoding — 99.8% fill on a 2,334-company run; a handful of addresses the register cannot geocode come back null.

Does it return every establishment, or just the head office? Head office by default. Set includeEstablishments to add the establishments that matched your search, each with its own SIRET and GPS — the API returns up to 10 per company, not the full branch list. establishmentCount always carries the true total.

Is the data up to date? Each row carries lastUpdated from the register. Sirene is refreshed continuously by INSEE.

Can I filter out one-person shell companies? Set employersOnly to keep only companies registered as having employees.

What happens if my search returns nothing? The run ends successfully with a status message explaining why, and you are charged nothing.

Is scraping this legal? The Sirene database is French open data, published for reuse under an open licence. This actor reads only the official public API and honours the register's own non-diffusion flag. You remain responsible for using the data lawfully, including GDPR where director names are involved.

Data source and fair use

Data comes from the French government's open company-search API over the INSEE Sirene base and the RNE. Only publicly published register records are returned, the register's own opt-out flag is respected, and no login or paywall is bypassed.