Market comparisons in procurement almost always use totals, which is the least informative number available: a total describes how large a country is, not what buying there looks like. The median notice is the useful figure, because it describes the transaction a supplier will actually meet. Put six European markets side by side on that measure and they turn out to be doing something close to different activities under the same name.
The spread is a publishing decision, not an economy
Finland’s median notice is €430,998 and Spain’s is €1,200. No plausible difference between two European economies produces a ratio of three hundred and fifty to one in the size of a typical public purchase. What produces it is where each country draws the line for publication. Spain publishes almost everything, including the very small contracts that most administrations settle without a notice at all, and its record is correspondingly dense — 999 of 1,000 sampled notices carry a value. Finland publishes the significant procedures and handles the small end elsewhere. Both records are honest; they are simply describing different slices of the same activity, and any comparison that treats them as equivalent is measuring publishing policy while believing it is measuring the market.
Plotted on one axis the spread stops looking like a difference between economies and starts looking like what it is — a difference between publishing rules.
| Market | Notices sampled | With a value | Median notice |
|---|---|---|---|
| Finland | 1,000 | 509 | €430,998 |
| Estonia | 730 | 327 | €140,000 |
| Lithuania | 1,000 | 413 | €60,000 |
| Latvia | 1,000 | 554 | €42,000 |
| Spain | 1,000 | 999 | €1,200 |
| Poland | 1,000 | 0 | not published |
What this does to a market-entry decision
The practical distortion runs in both directions. A supplier sizing Spain on median notice value concludes the market consists of trivial contracts and skips it, when in fact the same record contains a €930M framework a few rows above. A supplier sizing Finland the same way concludes the opposite and arrives expecting every opportunity to be substantial. The Baltic three sit between, and their ordering — Estonia at €140,000, Lithuania at €60,000, Latvia at €42,000 — says more about each portal’s threshold for publication than about the relative wealth of three neighbouring states. Poland resolves the problem by not participating in it: not one of a thousand sampled notices carried a value at all, which makes the median unanswerable rather than low.
Comparing markets honestly
The workable approach is to compare distributions rather than medians, and to state the publication threshold alongside any figure. A market where half the notices sit under €1,200 and a handful sit above €300M is not well described by any single number, and the median is actively misleading there. Where the aim is to compare countries, the only fair basis is a band — contracts above some common floor, present in every record being compared — which discards most of the Spanish data and almost none of the Finnish. That is unglamorous and it is the difference between a comparison and a ranking that rewards whoever publishes least.
No difference between two European economies produces a ratio of three hundred and fifty to one in the size of a typical purchase.
WHY IT MATTERS
Cross-market comparison is where procurement analysis fails most reliably, because the numbers arrive looking commensurable and are not. A median is a statement about what a country chooses to publish before it is a statement about what a country buys, and a market that publishes its smallest contracts is penalised by every ranking built on averages. Comparisons that state their floor, and that show a distribution rather than a single figure, survive contact with the data; the rest reward opacity. Poland’s blank column is the clearest case — a market cannot look expensive or cheap if it never says what anything costs.
Data: Otnox — live procurement intelligence across 56 markets. Basis: euro-denominated notices published in the last thirty days, sampled per market, 2026-06-26..2026-07-26.
