Reference · Data
Public procurement, in numbers.
Every figure on this page is one TRINTA publishes and stands behind, imported from the same data that renders its source page: the two barometers, built from Portal BASE and SECOP II records with the methodology printed beside them, and the platform registry behind the radar. Quote them with their sources; the links carry the full context.
Who buys and who supplies, measured in Portugal
Contracted value reported to Portal BASE, January to July 2026. Seven of the ten largest public buyers are health units; on the other side, pharmaceuticals lead the supply with energy behind it.
Largest buyers
Largest suppliers
Aggregated by tax number, never by name. The same entity is written up to 51 different ways in the source register and 764 entities have more than one spelling, so adding rows up by name would inflate every ranking and double-count EUR 206M in this edition alone. That is why the spelling count is printed beside each name: a league table of Portuguese public buyers built on names is wrong before it starts. Both tables are downloadable as CSV from the open datasets.
Where competition is thinnest, by CPV division
The same Portuguese register, cut by what is being bought. A supplier reading down this table is reading how hard each part of the market is to enter. Procedures for software package and information systems drew a single bid 12.9 times as often as procedures for agricultural, forestry, horticultural, aquacultural and apicultural services. 12,260 open procedures across 25 divisions, January to July 2026.
Portugal only, and open procedures only: direct awards are excluded because there the buyer chooses who bids, so a single offer is the design of the procedure rather than an absence of competition. Divisions below the reporting threshold do not appear. Method and sources on the Barómetro.
A division holds hundreds of codes, so its average covers procedures a single supplier would never all bid for. The same register answers this one level down: 4,624 individual CPV codes carry what Portugal signed beneath them, 261 of them with enough open procedures to publish a single-bid share of their own, and the whole table is one file at /cpv-market.json.
What another bidder is worth, in money
Every figure above counts how contested a market is. This one prices it. For each Portuguese open procedure that produced a single contract, the contracted price is compared against the buyer's own published base price, and the median discount is grouped by how many bids came in.
A procedure that draws one bid closes 0.6% below the ceiling the buyer set. One that draws more than ten closes 30.2% below it. The curve rises at every step, and the generator refuses to publish an edition where it does not.
This is an association and not a cause, and the difference matters: a buyer who attracts ten bids is usually buying something more standardised than one who attracts a single bid. Counted on 6,183 open procedures in janeiro a julho de 2026 that produced exactly one contract. Procedures split into lots are excluded, all 12,968 of them, because a ceiling belongs to the whole procedure while a price belongs to one contract, and dividing the two makes lotting look like price pressure. Also excluded: 3,562 without a declared bidder list, 39 above the ceiling and 39 below a tenth of it.
The same question in three countries, counted the same way
How often does a public tender draw a single bid. Three states publish enough to answer it, and each is read on its own open procedure rather than on everything it awards, because a rate that folds in direct awards measures how much a government chose to compete, not how much competition it met.
Same months, January to July 2026, and each figure counted from the state's own open register rather than estimated or surveyed. It is not a verdict on any of the three systems. What it is useful for is narrower and more practical: a supplier deciding which market to prepare a bid for can see where a prepared bid meets less company. Coverage differs and is published on each source page, because a rate without its coverage is half a number: Portugal reports a bidder list on 58.2% of open procedures, France fills offresRecues on 84.6% of marchés.
Why single-bid rates matter
A procedure that receives one bid is a procedure somebody won without competition, and the share of them is the most honest single measure of how contested a market really is. The two barometers measure it the same way on two continents, from the official records, with the caveats printed beside the number: equivalent procedure types, different thresholds, so it is an order of magnitude, not an exact ratio. For a supplier, every single-bid award is a tender someone else saw first.
Direct answers
The questions these figures answer
- Does more competition actually lower the price in public tenders?
- How often does a public tender receive only one bid?
- Which sectors of public procurement have the least competition?
- Can you trust a published single-bid rate?
- Which countries publish enough data to measure tender competition?
- How many days does a public tender give suppliers to respond?
- How many public tenders are published in Portugal, France and Colombia?
- Which month is the busiest for public procurement?