The data behind the barometers, open.
Every number the barometers publish is computed from official open records, and these are the aggregates themselves, downloadable under CC BY 4.0 with the methodology beside them. Aggregates only, never the contract-level register, which stays at its official source. Cite them with a link to this page or the barometer that reads them.
The Portuguese tables by CPV code, buyer and direct award, and the single-bid shares for Europe and the UK, are also a versioned repository on GitHub, as CSV with a column dictionary and a citation file: github.com/trinta-ai/public-procurement-data
| barometro-indicadores.csv | The fourteen headline indicators, one per row |
| barometro-mensal.csv | Contract value and count, by month |
| barometro-setores.csv | Competition by CPV division: tenders, single-bid share, average bids |
| barometro-distribuicao-propostas.csv | How many tenders received 1, 2, 3... bids |
| barometro-compradores.csv | The ten largest public buyers, by tax number, with how many spellings of each the register holds |
| barometro-fornecedores.csv | The ten largest suppliers, same shape: the spelling count is why the aggregation key is the tax number |
| barometro-2026-01_2026-08.json | Every aggregate above in one file, top buyers and suppliers included |
| barometro-README.md | Methodology: the five rules, each matching a known way to publish a wrong number |
| cpv-market.json | The same register at code level, not division level: 4,770 CPV codes with contracts signed, value contracted and, above the bar, how many bids they drew |
Editorial reading of these numbers: trinta.ai/barometro
Every edition, at its own address with its own files, as published: edition 2, January to August 2026; edition 1, January to July 2026. Edition 2 was corrected on 12 October 2026: the percentages are now truncated to one decimal place, as the method requires; before, they were rounded. No count and no amount changed. One sector label moved to the current spelling: “Equipamento elétrico”.
| cpv-market-fr.json | 4,186 CPV codes with marchés notified, notified amounts and, above the bar, how many offers competed marchés drew, from the field offresRecues that the DECP schema makes mandatory |
104,329 marchés after deduplication, 84.6% declaring their offer count. Method and caveats: the French barometer, in French
Every EU country, one definition
The Portuguese and French registers above each stop at their own border. eForms puts the same count on every contract award notice in the Union, so the single-bid share of 34 countries is counted on 571,148 lots from 164,798 award notices, with every CPV division and code beneath it. Above the EU thresholds only, which is why each national register reads lower. The whole aggregate is one file: single-bid-europe.json.
| ted-cross-border-2026-01_2026-09.csv | One row per EU and EEA country: award notices, the share won by a company registered in another country, the five foreign winner countries, single-bid share, median bids and the bid-origin shares with their coverage |
| ted-cross-border-2026-01_2026-09.json | The same rows with the counts behind every share, the 27 and 30 country totals, the inputs with the day each was fetched, and the method |
Editorial reading: cross-border public procurement in Europe, by country
| barometro-latam-mensual.csv | Contracts and value by month, COP |
| barometro-latam-ofertas.csv | Competitive processes by number of offers received |
| barometro-latam-procedimientos.csv | Single-bid share by procedure type |
Editorial reading: trinta.ai/barometro-latam
| placsp-espana-2026-01_2026-09.csv | Single-offer share of the lots awarded on Spain's national platform, by threshold, procedure, contract type, buyer type, region, regional platform and CPV division |
Editorial reading, in Spanish: trinta.ai/barometro-espana
How to cite
TRINTA, Barómetro da Contratação Pública Portuguesa, edition 2, January to August 2026, computed from Portal BASE open data (IMPIC), CC BY 4.0, trinta.ai/barometro/2026-08. The Colombian set cites SECOP II via datos.gov.co the same way. The methodology file lists the five rules each number was computed under, including the trap each rule exists to avoid.