Sign up

Use case · Public-buyer risk (beta)

Public-Buyer Risk API

Two calibrated scores about a public contracting body in Spain or France: how likely its next award is to draw a single bid, and how its supplier base has fared. Never a score for a supplier, never a score for one tender.

Published 19 September 2026 · Esta página en español · Cette page en français

The problem

A tender document tells a supplier what a contracting body wants to buy. It does not say whether that body's awards usually draw one bid or a dozen, whether the same handful of suppliers win every time, or whether the suppliers it has awarded to before tend to stay in business. A lender financing an invoice against a public contract has the same gap: the receivable is only as good as the buyer's history and the supplier's dependence on it. Today that reading comes from opening award filings one contracting body at a time.

GET /procurement/buyer/{id} and GET /procurement/buyers put a calibrated read of that history behind two REST calls. The subject is always the buyer. Beta; Spain and France.

The endpoints

GET /procurement/buyers lists scored contracting bodies ordered by one of two scores, highest first: model=competition (single-bid risk on the buyer's next award) or model=counterparty (insolvency risk in the buyer's supplier base). Filter with country (ES or FR), min_score (1-10), name or region, paginated with limit and a keyset cursor.

# Spanish contracting bodies scoring 8+ on single-bid risk
curl "https://api.prometiam.com/functions/v1/risk-api/procurement/buyers?country=ES&model=competition&min_score=8&limit=20" \
  -H "Authorization: Bearer rk_live_..."

GET /procurement/buyer/{id} returns the full profile for one buyer: both risk blocks, its coverage (awards, distinct suppliers, awarded value), behaviour ratios (single-bid rate, direct-award rate, framework rate, top-supplier value share), up to three evidence awards and its five largest supplier relationships. GET /procurement/relationship returns the detail behind one buyer-supplier pair: awards, awarded value, the supplier's share of the buyer's awarded value, first and last award, single-bid awards, average bids, and the supplier's insolvency date when it failed.

Use cases

  1. Bid or no-bid. Before a supplier spends on preparing a tender response, pull the buyer's competition score and single_bid_rate. A buyer whose awards near-always draw one bid, with a handful of names repeating in top_suppliers, is a different opportunity than one where bids are contested — worth knowing before the cost of bidding is sunk.
  2. Financing public receivables. A lender or factoring firm can see which authorities' supplier bases keep failing through the counterparty score, and how dependent one supplier is on a single buyer through GET /procurement/relationship — the top_supplier_value_share and the supplier's share of the buyer's awarded value both speak to concentration risk on the receivable.
  3. Audit and integrity. Screen a portfolio of buyers with min_score, then open the evidence rows on the ones that score high before drawing any conclusion.
# Open one buyer's profile, then the detail on one of its top suppliers
curl "https://api.prometiam.com/functions/v1/risk-api/procurement/buyer/LA0014114?country=ES" \
  -H "Authorization: Bearer rk_live_..."

curl "https://api.prometiam.com/functions/v1/risk-api/procurement/relationship?country=ES&buyer=EA0008424&supplier=B64892839" \
  -H "Authorization: Bearer rk_live_..."

What comes back

Each model — competition and counterparty — carries the same shape: score (1-10, a ladder on the log-odds of probability; 1 is the 1st percentile of the country's scored buyers, 10 is the 99th, the anchors are frozen), probability (the buyer's calibrated rate for the outcome on its next award — the observed count shrunk toward the model's expectation with a 30-award prior, calibrated out of time), base_rate, relative_risk (probability over base rate), percentile (of the residual between observed and expected — the peer comparison), observed_rate, expected_rate, awards_scored and confidence (high, medium or low, set by how many awards were scored). A buyer needs at least 20 awards in the period to be scored at all.

The counterparty probability is a weak signal: at the top of the ranking it runs at roughly twice the base rate, not a certainty for any one supplier. Read it with the evidence rows and top_suppliers, not on its own — a high score is a prompt to look, not a finding of wrongdoing.

Scope limits, stated plainly

  • only Spain and France, beta. country only accepts ES or FR.
  • Every score is about the buyer — the contracting body. Never a score for a supplier, never a score for one tender.
  • A buyer is scored only once it has at least 20 awards in the period; smaller or newer buyers are not returned.
  • France discloses a bid count on only part of its awards, which limits how complete the competition inputs are there.
  • The counterparty score is a weak signal, about twice the base rate at the top — read it with the evidence rows, not alone.
  • Coverage totals for a buyer (awards, awarded value) are an upper bound, not a certified count.
  • GET /procurement/relationship only holds a buyer's five largest supplier relationships by awarded value; the full award list for a supplier is on GET /procurement/awards, which covers Spain, France, the UK, Ireland, Poland and Norway.
  • No credit score and no supplier score, for any country. No financial statements from these endpoints.

FAQ

What does the public-buyer risk API cover?
Two calibrated scores about a contracting body in Spain or France: competition, the probability its next award draws a single bid, and counterparty, the probability a supplier it awards to enters insolvency within 24 months. Beta; a buyer needs at least 20 awards to be scored.

Is this a score for my company or my tender?
No. Both scores describe the buyer, a contracting body, and its pattern of awards over time. Neither score is ever about a specific supplier or a specific tender, and there is no supplier score anywhere in this API.

How reliable is the counterparty score?
It is a weak signal. Buyers at the top of the counterparty ranking see roughly twice the base insolvency rate among the suppliers they award to, not a guarantee for any one relationship. Read the score together with the evidence rows and the buyer's top suppliers, not alone.

Why is a buyer missing from the results?
A buyer is scored only once it has at least 20 awards in the period covered. France also publishes a bid count on only part of its awards, which limits the competition inputs there. Coverage totals for a buyer are an upper bound, not a certified count.

Does this replace a credit check or financial statements?
No. The API returns two behavioural scores about a public contracting body, not a credit score and not company financial statements (a separate endpoint serves those for GB, FR, DK, SE, NO and SEC filers in the United States, for Finland (statements filed digitally with PRH, about one filer in twenty, and listed groups) and for listed companies in ES, PL, HR and BE). There is no supplier score, and a high score is a prompt to look, not a finding of wrongdoing.

Related