The SIRENE file is probably the most underutilized B2B database by French growth teams. Maintained by INSEE, it officially lists every active company and establishment in France, with daily updates and 100% free API access. Yet, most SDRs continue to buy third-party databases without realizing that their upstream source is SIRENE. This guide breaks down the mechanics: record structure, API access, ICP filters, GDPR compliance, and how to enrich it with Apollo, Pharow, or Dropcontact.

TL;DR: SIRENE in 30 seconds

The SIRENE file is the official directory of French companies maintained by INSEE. Free API access, 39M establishments, daily updates. Upstream source for 90% of FR prospecting tools.

Key points
  • SIRENE = Information System for the Directory of Companies (INSEE, 2026).
  • Volume: ~39M establishments, of which ~13.3M active at the beginning of 2026 (INSEE).
  • Free access: Sirene API v3.11 on api.insee.fr and stock files on data.gouv.fr.
  • Usable fields: SIREN, SIRET, NAF/APE, employee headcount bracket, creation date, geolocation.
  • B2B prospecting legality: legitimate interest GDPR, under strict conditions (CNIL).

What is the SIRENE file and who maintains it?

The SIRENE file is the official directory of French companies managed by INSEE since 1973. It contains 39 million establishments (active and closed) and approximately 13.3 million active establishments at the beginning of 2026 according to INSEE statistics. Each entity receives a unique identifier, SIREN or SIRET, upon registration.

The acronym stands for Système Informatique pour le Répertoire des Entreprises et de leurs Établissements (Information System for the Directory of Companies and their Establishments). Concretely, it is the upstream source that feeds INPI, court registries, URSSAF, and most FR B2B data tools. When Pappers or Societe.com display a company profile, they consume SIRENE in the background.

SIREN vs SIRET: the confusion to avoid

The SIREN identifies the legal entity with 9 digits. The SIRET identifies a physical establishment with 14 digits (SIREN + 5 NIC digits). A multi-site company has one SIREN but several SIRETs. This distinction changes everything for geographical targeting: prospecting a retail brand at the SIRET level, yes; at the SIREN level, you only target the head office.

How to access the SIRENE file for free?

Access has been entirely free since the directory's open data release in January 2017, following the Law for a Digital Republic of 2016. Three channels coexist depending on your volumetric needs: the Sirene API for targeted queries, complete stock files, and the unitary search interface.

Sirene API v3.11: for structured queries

The Sirene API allows 30 queries per minute for anonymous access and up to 500 per minute after creating a developer account. It exposes the /siren, /siret, and /informations endpoints. The query syntax supports logical filters (AND, OR) on all fields: NAF code, headcount bracket, department, active status.

Stock files on data.gouv.fr

To ingest the entire database, data.gouv.fr publishes a complete monthly dump (StockEtablissement, StockUniteLegale) in CSV format. Expect ~12 GB uncompressed for the establishment stock. This is the preferred solution for building an internal data warehouse and cross-referencing with other signals (fundraising, recruitments, web traffic).

What data does a SIRENE record contain?

An establishment record contains about a hundred exploitable variables, of which the official INSEE documentation lists 95 fields for StockEtablissement and 47 for StockUniteLegale. The majority are declarative (updated by the company) but some are calculated by INSEE, such as the legal category or the employee headcount bracket.

Critical fields for growth

  • SIREN/SIRET: universal join key with any other FR dataset.
  • NAF/APE Code (5 characters): French activity nomenclature, ~730 codes, mappable to SIC or NACE.
  • Employee headcount bracket: 13 brackets (from 0 employees to 10000+), based on social security declarations.
  • Creation date: valuable for targeting young companies (SaaS tool purchase signal).
  • Geocoded address: municipality, department, region, Lambert 93 coordinates.
  • Administrative status: active (A) or ceased (C). Always filter on A for prospecting.

What SIRENE does not contain: turnover, net profit, executives, emails. For this data, you need to cross-reference with INPI annual accounts (National Register of Companies) or enrichment APIs.

How to use SIRENE for B2B prospecting?

SIRENE becomes a growth lever when used as an ICP filtering layer before enrichment. According to a HubSpot study published in 2025, 50% of prospects are not suitable for the product. Filtering upstream on SIRENE halves the downstream enrichment cost and mechanically improves cold email response rates.

Typical workflow for a B2B sales team

Step 1: define the ICP using NAF codes (e.g., 6201Z computer programming, 7022Z business consulting). Step 2: extract the corresponding SIRENE subset (headcount + department + active status filters). Step 3: enrich with Pharow, Apollo, or Dropcontact to retrieve decision-makers, emails, and intent signals. Step 4: route to the cold email sequence or LinkedIn Ads.

The ICP filters that really work

Three profitable combos observed across dozens of campaigns: (1) NAF + headcount bracket 10-49 = SaaS midmarket sweet spot; (2) creation date < 24 months + IDF department = early adopters; (3) NAF retail + multi-establishments (>5 SIRET) = ABM large accounts. Incorrectly filled NAF codes remain a classic pitfall. Always validate on 50 records manually before scaling.

Is prospecting on SIRENE legal under GDPR?

Yes, under conditions. The CNIL authorizes B2B prospecting based on legitimate interest, provided the message is related to the recipient's professional function. No prior opt-in is needed for a professional email, unlike B2C. But the opt-out must be visible and functional on every message.

Three non-negotiable rules: inform the prospect of the data source from the first contact, offer a functional unsubscribe link, and process any deletion request within 30 days. SIRENE being a public database, you are covered for the legality of collection. Regulatory friction shifts to downstream enrichment, especially when you retrieve scraped nominative emails.

Personal data in SIRENE: the case of individual entrepreneurs

Since April 2022, INSEE applies a right of opposition for individual entrepreneurs and micro-entrepreneurs. Approximately 2.5 million natural persons have checked the non-disclosure box. You must exclude these records from any commercial prospecting, under penalty of CNIL sanction. Filter on the field statutDiffusionUniteLegale = O (yes, disclosable).

What are the alternatives to the SIRENE file in 2026?

SIRENE remains the upstream source, but its raw interface deters non-technical users. Several players have built UX layers on top, all relying on the same INSEE data. According to le Journal du Net, the three most consulted interfaces by FR sales professionals are Pappers, Societe.com, and Infogreffe.

Pappers, Societe.com, Infogreffe: what to choose?

  • Pappers: modern interface, freemium API, integrated annual accounts. The most used by sales in 2026.
  • Societe.com: market history, but outdated UX and many hidden paid contents.
  • Infogreffe: official source for court registries, perfect for Kbis and legal acts, but priced per document.
  • Direct Sirene API: the only option if you are building a product or a data warehouse with >100k queries/month.

The right trade-off depends on volume. Below 10k records/month, free Pappers is sufficient. Above that, the direct INSEE API remains unbeatable in terms of cost (zero) and freshness (D+1). For a data-driven growth marketing strategy, direct SIRENE ingestion creates a lasting advantage.

FAQ

How to download the complete SIRENE file for free?

Go to data.gouv.fr, SIRENE database section. Download StockEtablissement.zip (~3 GB compressed) and StockUniteLegale.zip. Monthly updates, CSV UTF-8 format. No registration required. For daily delta, subscribe to the StockEtablissementHistorique feed.

What is the difference between the SIRENE file and the trade register?

SIRENE lists all companies (traders, artisans, liberal professions, employing associations), i.e., 13.3M active establishments. The trade and companies register (RCS) only contains traders and commercial companies, about 6M entities. SIRENE is broader, the RCS is more legal with acts and statutes.

Can you cold email contacts extracted via SIRENE?

SIRENE does not provide nominative emails. You must enrich via Dropcontact, Kaspr, or Apollo. B2B cold emailing is legal under legitimate interest GDPR, provided there is a visible opt-out and targeting by function. The CNIL recommends mentioning the source from the first message.

How many companies are created each year according to SIRENE?

INSEE recorded 1,111,200 company creations in 2024 in France, of which 64% were micro-enterprises. This flow feeds SIRENE in near real-time via the CFE. For B2B SaaS, the cohort of commercial companies (excluding micro) remains the activatable segment, i.e., ~400k annual creations.

Does the Sirene API have a request limit?

Yes: 30 requests/minute anonymously, 500/minute with a developer account on api.insee.fr. No daily quota. For volumes >1M records, prefer downloading the monthly stock from data.gouv.fr then local ingestion. API latency goes up to 200-400ms per request, so it's unsuitable for massive scraping.

From raw data to qualified pipeline

SIRENE is not a turnkey solution; it's a raw material. Its true value emerges when you couple it with an enrichment layer (emails, intent signals) and a clean activation workflow. The growth teams that win in 2026 are not those who pay the most for their B2B data, but those who ingest it at the source and qualify it finely.

If you want to transform this database into a qualified pipeline, the challenge shifts to ICP targeting, contact sequence, and compliance. Our team supports B2B scale-ups on this exact stack, from SIRENE extraction to the operated cold email sequence. To go further, browse other articles on the Uclic blog about data-driven outbound.