French Companies Scraper

Returns: one row per French company with siren, name, siret, legalForm, nafCode, status, address and sourceUrl. Input: {"departements":["75"],"nafCodes":["62.01Z"],"maxItems":25} Best for: company research, local market maps, and building lists by industry and location.

{
  "siren": "380474494",
  "name": "EURO INFORMATION DEVELOPPEMENTS (EID)",
  "siret": "38047449400016",
  "legalForm": "5710",
  "nafCode": "62.01Z",
  "status": "A",
  "city": "STRASBOURG"
}

What does French Companies Scraper do?

French Companies Scraper searches the public French business registry search API and returns structured company records. It accepts company names, SIREN or SIRET identifiers, address text, departments, postal codes, NAF activity codes, legal form codes, workforce bands, and company size categories. The Actor combines the supplied filters, walks the API's pages, and writes one dataset row per unique SIREN. It does not need a login.

The source is the official recherche-entreprises.api.gouv.fr search endpoint. This Actor is an independent tool and is not affiliated with the French administration. The sourceUrl on each row records the exact API request that returned that company, so downstream work can trace a record to its source.

Search filters apply to the API result, which may match an establishment other than a company's head office. The output's siret, address, postalCode, city, latitude, and longitude describe the head office (siege) supplied in the API response. For example, a company found through a Paris department filter can have its head office in another department. A department match can describe a local establishment while the displayed address identifies its head office.

Non-diffusible records are skipped. Missing directors, coordinates or finances appear as null, never invented.

What data can you get?

Each output row has a stable id, equal to its SIREN, and the endpoint sourceUrl. The remaining fields map directly to the API's company object or its head-office object. Codes stay as codes, which makes joins to other French open datasets straightforward. Financial data is grouped by reporting year exactly as returned.

Field	Meaning
id	Stable company key, equal to siren.
sourceUrl	Search API URL that supplied this row.
siren	Nine-digit company identifier.
name	Full company name.
siret	SIRET of the head office.
legalForm	INSEE nature_juridique code.
nafCode	Company's primary activity code.
headcountBand	Employee-count band code when supplied.
companyCategory	Size category such as PME, ETI, or GE.
creationDate	Company's creation date.
status	Administrative status code, A (active) or C (closed).
address	Full head-office address.
postalCode	Head-office postal code.
city	Head-office city name.
latitude	Head-office latitude when available.
longitude	Head-office longitude when available.
establishmentCount	Number of establishments reported for the company.
vatNumber	Intra-community VAT number, when the registry lists one.
openEstablishmentCount	Number of currently open establishments.
closureDate	Date the company ceased activity, when closed.
lastUpdated	Registry timestamp of the last update.
acronym	Registered acronym (sigle), when present.
nafCodeNaf25	Main activity code in the NAF 2025 nomenclature, when supplied.
headcountBandLabel	Plain-language label for headcountBand (for example "50 to 99 employees"), when the band is known.
matchingEstablishments	Establishments that matched your filters (for example the Paris sites of a company headquartered elsewhere), with SIRET, address, NAF code and status. null when the API returns none.
isSocialEconomy	True or false for the social and solidarity economy flag.
isAssociation	True or false for the association flag.
collectiveAgreementIds	IDCC collective agreement codes, when listed.
directors	Director objects from the API, when available.
finances	Year-keyed financial figures such as ca and resultat_net, when available and requested.

A legal form code identifies a legal form in the INSEE classification; the Actor does not guess a human-readable label. A workforce band is also a code, not an exact employee count. Financial figures can be absent even for a valid company. Director records can include people or legal entities, with different properties for each kind. Consumers should handle null fields and inspect the director objects they use.

How to use

Open the Actor's input form and enter at least one search query or filter. A common search is a department plus a NAF code. Set maxItems to the number of unique companies you want. Run the Actor, then open the dataset to view or export the rows. If you only need a short sample, use a small maxItems value; the Actor stops as soon as that many records have been saved.

For a name search, use searchQueries. It accepts names, SIREN or SIRET strings, and address text. For location research, use departements or postalCodes. For industry research, use nafCodes. Multiple entries in a list are searched separately and combined in one output. When you supply different filter lists, the Actor searches their combinations. The same SIREN is returned once even if several combinations match it.

activeOnly defaults to true and sends the API's active-status filter. Set it to false if you also want closed company results. For an exact nine-digit SIREN or fourteen-digit SIRET query, the API performs a direct lookup and ignores the other filters, including active status. The Actor logs a warning in that case. includeFinances controls whether the Actor includes the finances field in its output; when false, finances is null. It does not turn missing figures into zero. The endpoint includes directors and available finances in its normal response, so the Actor does not need to visit a company detail page for each row.

The search endpoint caps large result sets. The Actor requests at most 25 entries per API page, using a smaller fixed page size when maxItems is below 25 and follows pages until it reaches maxItems or the API result boundary. The Actor saves qualifying records from the first page before splitting, and stops immediately if that page satisfies maxItems. For a very broad search that reaches the API's 10,000-result cap, it splits by department, then by postal code when necessary. A very narrow postal-code query that still reaches the cap should be split further by a NAF code in the input. Requests are paced at no more than two search calls per second in the normal path, below the API's documented limit.

Input

Input	Description	Default
searchQueries	Company name, identifier, or address terms.	Empty list
departements	French department codes, for example 75 or 69.	Empty list
postalCodes	Five-digit French postal codes.	Empty list
nafCodes	NAF activity codes, for example 62.01Z.	Empty list
legalForms	nature_juridique legal form codes.	Empty list
headcountBands	Employee band codes.	Empty list
directorNames	Director or officer names to search (e.g. Dupont).	Empty list
minRevenue / maxRevenue	Latest reported revenue range in euros. Companies without published figures are excluded when set.	Not set
minNetResult / maxNetResult	Latest net result range in euros (negative values allowed).	Not set
companyCategory	PME, ETI, or GE.	No category filter
activeOnly	Restrict search to active companies.	true
includeFinances	Include available financial data.	true
maxItems	Maximum unique rows, from 1 to 10,000.	50
proxyConfiguration	Apify proxy settings if needed.	No proxy

This example searches active software-development companies with a matching presence in department 75 and returns up to 25 unique companies:

{
  "departements": ["75"],
  "nafCodes": ["62.01Z"],
  "activeOnly": true,
  "includeFinances": true,
  "maxItems": 25
}

The Actor requires at least one nonempty search or filter list, or a revenue/net-result bound. Company category and output options alone are insufficient. maxItems is an exact ceiling: if seven qualifying unique records exist and you set it to seven, the dataset contains seven rows. If fewer qualify, the dataset contains the available rows. A well-formed search that matches nothing finishes successfully with an empty dataset. Its final status suggests removing a location, NAF, director or financial filter, trying an exact SIREN/SIRET, or setting activeOnly to false for closed companies. Single-digit department codes are padded with a leading zero, and Corsican codes are accepted in either letter case (for example, 1 becomes 01 and 2a becomes 2A). Department entries may contain comma-separated or semicolon-separated codes. Unsupported department targets are skipped with a clear log message; other department targets continue, and a run containing only unsupported departments succeeds with zero results. Invalid postal or NAF formats still fail input validation instead of producing placeholder records.

Output

This is an excerpt of a real item from the current local prefill run (departements: ["75"], nafCodes: ["62.01Z"], maxItems: 25). Address fields describe the head office, even when a different establishment matched the filters.

{
  "id": "380474494",
  "sourceUrl": "https://recherche-entreprises.api.gouv.fr/search?departement=75&activite_principale=62.01Z&etat_administratif=A&page=1&per_page=25",
  "siren": "380474494",
  "name": "EURO INFORMATION DEVELOPPEMENTS (EID)",
  "siret": "38047449400016",
  "legalForm": "5710",
  "nafCode": "62.01Z",
  "headcountBand": "51",
  "companyCategory": "GE",
  "creationDate": "1991-01-01",
  "status": "A",
  "address": "4 RUE FREDERIC-GUILLAUME RAIFFEISEN 67000 STRASBOURG",
  "postalCode": "67000",
  "city": "STRASBOURG",
  "latitude": 48.600490458,
  "longitude": 7.7573257675,
  "establishmentCount": 61
}

Rows are company-level, not one row per establishment; the head-office SIRET is not necessarily the establishment that matched a location filter.

How much does it cost?

This Actor is charged per result saved to the dataset. A result is charged once when its row is written. Repeated SIRENs across pages or search combinations are skipped and do not create additional result rows. The maxItems limit is enforced before writing and charging another row. Check the Actor's Pricing tab for the current terms; this README deliberately does not quote an amount because Store terms can change.

An empty run has no result rows. Reducing maxItems is useful when validating a new filter combination, because the Actor stops collecting after the requested number of unique rows. API pages can contain duplicates or non-diffusible records, so the Actor may request more than one page to gather the requested count. The dataset row count, rather than the number of search requests, describes the output you receive.

Use with AI agents (MCP)

Apify's MCP server can expose this Actor to an AI agent. Configure the MCP endpoint with the Store path, then ask for a search with explicit filters and a row limit:

{
  "mcpServers": {
    "apify": {
      "url": "https://mcp.apify.com/?tools=axiomworks/french-companies-scraper"
    }
  }
}

Example prompts are “Find up to 20 active French software companies with a presence in department 75 and include their SIRENs and legal forms” and “Search for boulangerie in department 69, return the company names, head-office cities, and source URLs.” An agent can use the field descriptions in the input schema to pick filters. It should state maxItems when a bounded result set matters. It should also explain the distinction between the matching search location and the displayed head-office address when presenting local search results.

Because the output includes stable identifiers and provenance, an agent can deduplicate records, cite the sourceUrl, and join the company data with another approved workflow. Fields such as workforce band and legal form are codes, so an agent should avoid presenting them as exact employee counts or legal-form labels unless it has a separate authoritative mapping. null in a finance field means the API did not return a figure, not that the figure is zero.

Integrations and API

Use the Actor from an Apify workflow or call it through the Apify API. The examples below submit the same input and ask for dataset items. Supply your Apify credentials through your normal client environment; do not place them in a public script. The Actor itself reads the public registry endpoint without a target-site login.

Python with the Apify client:

from apify_client import ApifyClient
import os

client = ApifyClient(os.environ["APIFY_TOKEN"])
run = client.actor("axiomworks/french-companies-scraper").call(run_input={
    "departements": ["75"], "nafCodes": ["62.01Z"], "maxItems": 25
})
for item in client.dataset(run["defaultDatasetId"]).iterate_items():
    print(item["siren"], item["name"])

JavaScript with the Apify client:

import { ApifyClient } from 'apify-client';
const client = new ApifyClient({ token: process.env.APIFY_TOKEN });
const run = await client.actor('axiomworks/french-companies-scraper').call({
  departements: ['75'], nafCodes: ['62.01Z'], maxItems: 25,
});
const { items } = await client.dataset(run.defaultDatasetId).listItems();
for (const item of items) process.stdout.write(`${item.siren} ${item.name}\n`);

cURL using an environment token:

curl -X POST \
  'https://api.apify.com/v2/acts/axiomworks~french-companies-scraper/run-sync-get-dataset-items' \
  -H "Authorization: Bearer $APIFY_TOKEN" \
  -H 'Content-Type: application/json' \
  --data '{"departements":["75"],"nafCodes":["62.01Z"],"maxItems":25}'

Datasets export from Apify Console as JSON, CSV or Excel.

FAQ

Can I filter by revenue or by director? Yes. Set minRevenue/maxRevenue or minNetResult/maxNetResult in euros, or list names in directorNames. The registry applies them to the latest figures it holds, so companies with no published finances are left out of revenue filters. The finances and directors fields show what matched.

Can I search without a company name? Yes. Supply a department, postal code, NAF code, legal-form code, or another supported filter. At least one search or filter value is needed. Broad searches can be large, so set maxItems deliberately.

Can the same company appear twice? The Actor uses SIREN as the stable ID and returns each SIREN only once per run, even if different pages or filters find it.

What happens when a query finds nothing? The run succeeds with zero dataset items and a warning. It does not write an error object as if it were a company.

Does the Actor need a proxy? The default is direct access to this public API. The optional proxyConfiguration input lets users configure Apify Proxy. Blocked or rate-limited requests are retried with backoff and fresh proxy sessions when a proxy is configured.

Is it legal to scrape French companies?

The source is a public French government company search API based on open data. The Actor requests publicly available fields and does not access login-protected information. You remain responsible for your own use of the resulting data, including applicable privacy, attribution, and retention requirements. Public availability does not make every downstream use appropriate. Director information can concern identifiable people, so assess your purpose before storing or distributing it. This Actor is independent of the registry operator and does not claim official endorsement.

Feedback

If a search filter appears to return unexpected records, include the input JSON and a sourceUrl from an affected row when reporting it. If a field is missing, note the SIREN and which field you expected, while avoiding unnecessary personal details. Reports with a reproducible input help distinguish API coverage limits from Actor mapping issues. You can also suggest additional company-level fields that are present in the public response and useful for analysis. The Actor's output schema and README can then be updated to match any verified change.