Developer guides

BOAMP open data API: French public tenders and how to dedupe against TED

France publishes BOAMP, its national procurement bulletin, as open data on Opendatasoft with no key and the Licence Ouverte 2.0; the JOUE family in it is a French copy of a TED notice and must be skipped when you also read TED. Source: Arcadium Data, data as of 2026-10-25.

The Bulletin officiel des annonces des marchés publics (BOAMP) is run by DILA and carries notices of French public buyers: the national procedures between about EUR 90,000 and the EU thresholds, the "procedure adaptee" (MAPA) below that, concessions, and French copies of notices that also go to the EU's TED. Its open dataset has history from 2015.

The data is good and easy to reach. The catches are the row cap on one query, a notice body that contains personal data, and the double counting with TED.

The official route: Opendatasoft records API

The endpoint

GET https://boamp-datadila.opendatasoft.com/api/explore/v2.1/catalog/datasets/boamp/records is anonymous. Filters use the Opendatasoft query language: where, select, order_by, limit (100 a call) and offset. Offset plus limit cannot pass 10,000 on one query, so slice by publication day. A day holds a few hundred national notices, well under the cap. For a full historical copy use the dataset's exports endpoint instead of paging.

curl

curl -s -G "https://boamp-datadila.opendatasoft.com/api/explore/v2.1/catalog/datasets/boamp/records" \
  --data-urlencode "where=dateparution = date'2026-10-23' AND famille != 'JOUE'" \
  --data-urlencode "select=idweb,objet,famille,nature,dateparution,datelimitereponse,nomacheteur,code_departement,url_avis" \
  --data-urlencode "limit=100"

Python

import requests

URL = "https://boamp-datadila.opendatasoft.com/api/explore/v2.1/catalog/datasets/boamp/records"

def boamp_day(day: str):
    offset = 0
    while offset < 10_000:
        r = requests.get(URL, timeout=60, params={
            "where": f"dateparution = date'{day}' AND famille != 'JOUE'",
            "select": "idweb,objet,famille,nature,dateparution,nomacheteur,url_avis",
            "limit": 100, "offset": offset,
        })
        rows = r.json().get("results", [])
        yield from rows
        if len(rows) < 100:
            return
        offset += 100

for row in boamp_day("2026-10-23"):
    print(row["idweb"], row["nature"], row["nomacheteur"])

The fields worth keeping

Personal data in donnees

The donnees JSON holds contact persons, phone numbers and email addresses (keys such as correspondantPRM, nomContact and, in notices before 2024, CORRESPONDANT, TEL, MEL). Read only what you need: CPV codes, estimated and awarded amounts, and the winner's company name. Winners can also be natural persons (PersonnePhysique); skip those. Notices before 2024 use an older schema with upper-case keys and values like {"#text": ..., "@DEVISE": ...}, so write two parsers or start in 2024.

Deduplicating against TED

French buyers above the EU thresholds publish in the Official Journal, and BOAMP keeps a French copy under famille = JOUE. If you read TED as well, filter that family out and take above-threshold notices from TED, which has the richer eForms data. Below the thresholds, BOAMP is the only source. After that filter the two do not overlap in our data. Winners are often withheld in French notices (a large share of awards have no named winner), so do not read a missing name as an error.

Keeping the copy current

A notice is not final when it is published. Buyers issue rectifications (nature contains "rectificatif") and award notices that point back to the call through annonce_lie. A daily job should therefore query the last three or four publication days again, not only yesterday, and upsert by idweb. Store the raw record next to your parsed row; the schema changed once and may change again.

What you can build from it

• An opportunity feed by department (code_departement) and keyword in objet, with the deadline in datelimitereponse. Below the EU thresholds this is the only complete source, and small and mid-size suppliers live in that range.
• A winners list from award notices: take the company name from donnees, skip natural persons, and join the name to the French company register by name and place. The register is open and covered in our registers guide.
• A buyer profile: how often a buyer publishes, which procedure types it uses and in which months. Many French buyers publish in waves at the start of the budget year.

Two checks before you trust a count

First, reconcile with the source: pick one day, count the records the API returns for it with the family filter on and off, and compare with what the BOAMP site shows for that day. A mismatch usually means a time zone slip in the date filter or a family you forgot. Second, watch for the same procedure published twice by the same buyer, once as MAPA and once after a correction; group by buyer, object and date window before you report the number of tenders in a department. French buyers also publish "avis d'attribution" for contracts that were never advertised, so awards can outnumber calls in a quarter without any error. Finally, record the date of your last full reconciliation in your own pipeline notes; a quarterly check against the BOAMP website catches silent schema changes early. Use the department code to build regional views, and remember that overseas departments and some national buyers carry codes that are not metropolitan departments.

Licence: Licence Ouverte / Etalab 2.0. Attribute DILA and keep url_avis.

The fast route: French notices with TED merged

Arcadium Data reads BOAMP daily next to TED, skips the JOUE copies, parses both notice schemas, drops personal fields, and returns buyers, CPV codes, values in EUR and the department. The one-year market summary for France is open without a key:

Calls on this page work without a key from one IP (10 requests a minute, 100 a day, short answers, point lookups). A free key lifts the limits: sign up. Routes, parameters and fields: OpenAPI.

curl

curl -s "https://api.arcadiumdata.io/v1/tenders/markets?country=FR"

Python

import requests

data = requests.get("https://api.arcadiumdata.io/v1/tenders/markets?country=FR", timeout=30).json()
print(data["awards"], data["awardsWithWithheldWinner"], data["frameworks"])
for b in data["topBuyers"]:
    print(b["name"], b["awards"])

TypeScript

const res = await fetch('https://api.arcadiumdata.io/v1/tenders/markets?country=FR');
const data = (await res.json()) as { awards: number; frameworks: number; topBuyers: { name: string; awards: number }[] };
console.log(data.awards, data.frameworks);
for (const b of data.topBuyers) console.log(b.name, b.awards);

Response (abridged)

{
  "awards": 41799, "awardsWithWinner": 23497, "awardsWithWithheldWinner": 21128,
  "frameworks": 18157, "valueEur": 85179274320,
  "topWinners": [{ "name": "IDVERDE", "awards": 119 }, { "name": "OTIS", "domain": "otis.com", "awards": 118 }],
  "topBuyers": [{ "name": "EDF SA", "awards": 204 }, { "name": "Bordeaux Métropole", "awards": 184 }]
}

GET /v1/tenders/expiring?country=FR lists French contracts by end date and expected re-tender window, and a free key opens GET /v1/tenders with country=FR. See also the French company registers and tenders by sector.

Comparison

Related

• TED API guide
• Find public contracts that end soon
• CPV codes explained
• Company registers in Europe
• French tenders

Sources

• BOAMP dataset on Opendatasoft
• BOAMP (DILA)
• Licence Ouverte 2.0
• Opendatasoft Explore API v2.1

Last updated 2026-10-25. This guide is ours and independent of the organisations it describes. Found a mistake? Tell us and we correct it.

Cite as: Arcadium Data, "BOAMP open data API: French public tenders and how to dedupe against TED", https://arcadiumdata.io/guides/boamp-api-france-public-tenders/, data as of 2026-10-25.