← Research

Research

What nobody has measured

The most confidently-asserted claims in procurement have never been tested. Here’s what we intend to do about it.

Spend a day reading about how to win public tenders and you will encounter a great deal of confidence.

Bids fail on compliance. Presentation matters. Evaluators respond to clarity. A well-structured proposal beats a well-written one. Relationships help, even when they legally shouldn't. Get the formatting right, or you'll be marked down.

Some of this is probably true. That is not the same as it being known.

We went looking for the empirical basis of the claims that dominate practitioner conversation in European public procurement — the things bid consultants charge for, the things training courses teach, the things everyone in the room nods along to. And we found something we didn't expect.

They have not been measured. Not disproven — unmeasured. The mechanisms are documented in law and asserted in guidance. The frequencies and effect sizes, in most cases, simply do not exist in the literature.

This piece is about the holes. And about the fact that they are fillable.

Hole one: how often do bids die administratively?

This is the big one, and it is the one that should bother you most.

Everyone in procurement knows that a bid can be excluded before anyone reads it. A late submission is a non-submission — courts across Europe have upheld this without sympathy, and five minutes is enough. A missing or incomplete ESPD/UEA creates a compliance problem that may or may not be curable. An unsigned declaration, a price schedule on the wrong template, an attachment uploaded to the wrong section: any of these can end a bid on formal grounds, without a single evaluator ever reading a word of it.

The legal literature on this is clear and reasonably deep. The consequences are severe and well-documented.

The frequency is unknown.

We could not find a study that quantifies, across European tenders, what proportion of bids are rejected for non-substantive reasons rather than on merit. Not by defect type. Not by country. Not by sector. The legal scholarship analyses whether a missing ESPD may be cured; it does not tell you how often one goes missing.

The academic literature on procurement is heavily concentrated on the award phase — who wins among evaluated bids, and why. Pre-qualification and verification are comparatively under-studied. Which means the field is far better at explaining who wins the game than at explaining who never got to play.

So when you read — including in our own writing — that bids are commonly lost on compliance rather than content, understand what that claim rests on. It rests on mechanism, not measurement. It is entirely plausible. It is not established.

Hole two: does any of the bid-writing advice work?

Here is the finding that should give the entire bid-consulting industry pause.

We could not find a single empirical study, in the European procurement literature, that measures an independent effect of proposal polish, formatting, visual design, prose quality, or team-CV presentation on tender scores, controlling for substantive compliance and criterion-fit.

Not one that finds a positive effect. Not one that finds a null effect. The question has not been asked.

This is genuinely strange. These are the things practitioners talk about most. They are what training courses cover, what consultants are hired to improve, what internal bid reviews argue about. An entire professional discipline rests on the premise that how you say it changes the score.

And it may. We are not claiming otherwise. We are pointing out that the claim has never been put to a test that could have failed.

What has been measured, repeatedly, is something duller and more useful: procedural capability. Knowing how the process works. Handling the documents correctly. Managing the mechanics. Surveys of thousands of SMEs find this associated with both more bidding activity and a better win ratio.

Not creativity. Not prose. Process.

The uncomfortable possibility is that the profession has spent decades optimising the thing that was never measured, while the thing that was measured is the unglamorous one that nobody sells a course in.

Hole three: do the SME reforms work?

The 2014 EU procurement directives were designed, in significant part, to open public markets to smaller firms. Turnover requirements were capped at twice the contract value. Contracts were to be split into lots where practicable. The ESPD was introduced to reduce documentation burden.

Ten years on, some of this has been evaluated and some of it has not.

We know, from cross-country TED analysis, that open procedures increase SME participation but not SME win rates conditional on competing — the policy gets small firms into the room without making them likelier to leave with the contract. We know that lots only improved SME win probability below roughly €23,469, which is a startlingly small number.

But the turnover cap — arguably the single most concrete SME-protective rule in the directives — has, as far as we can establish, never had its causal effect isolated. Did capping turnover requirements at 2× contract value measurably increase SME participation? Nobody appears to have answered that with a clean identification strategy.

Nor has anyone measured how often buyers actually comply with it, or how often they invoke the exception, or how often a bidder challenges a disproportionate requirement and gets it changed.

The rule exists. Its effect is unknown. That is a strange place to be, a decade in.

Hole four: how are tenders actually scored?

We know from Dutch research that relative price scoring produces rank reversal — that in roughly one in five tenders studied, adding a bid that could not win changed which of the others did. That is a hard, replicated, alarming finding.

We know from the scoring-rule literature that some formulas compress or expand price ranges, so that a nominal 40% price weight may discriminate almost nobody, or a nominal 30% may behave like 50%.

What we do not know is how common any of this is.

What proportion of European tenders use relative rather than absolute price scoring? Does it vary by country, by sector, by buyer type, by contract size? How often is there a minimum quality threshold — the single feature that most reduces rank-reversal risk?

These questions are answerable. Every one of them is disclosed in the tender documents. Nobody has counted.

What we're going to do about it

We hold a corpus of over ten thousand open EU tender notices, and — for a substantial and growing share of Dutch tenders — their actual procurement documents. The Aanbestedingsleidraad. The selection criteria. The award criteria and their weights. The scoring formulas, written out.

We built this to help companies bid. But it is also, incidentally, a dataset that could answer several questions the literature has left open.

Specifically, we think the following are measurable, and we intend to measure them:

  1. The prevalence of relative versus absolute price scoring. By country, sector and contract size. This is stated in the documents. It has never been counted, and it directly determines how exposed a bidder is to rank reversal.
  2. How often a minimum quality threshold is set. The single design feature that most protects a tender from rank reversal. Present in some tenders, absent in others. The ratio is unknown.
  3. Whether the nominal and effective weights diverge. Where a formula is stated explicitly, its shape can be examined. How often does a stated price weight behave, mechanically, like a substantially different one?
  4. How often turnover requirements exceed the 2× cap. This requires the estimated contract value and the stated turnover requirement — both disclosed. If disproportionate requirements are common, that is a finding with policy consequences. If they are rare, that is also worth knowing.
  5. What selection criteria actually demand, in aggregate. Which certifications, how many references, of what recency, per what competency. The barrier that excludes SMEs has never been systematically described.
  6. Where award criteria are published at all. In our corpus, fewer than two in five open tenders carry usable award criteria in the TED notice — and when they do, the data is frequently duplicated per lot or filled with placeholders. The real criteria sit in the documents. How large is that gap, and does it vary by country?

An open offer

We will publish what we find, including the results that are inconvenient for us.

If we discover that administrative disqualification is rare, we will say so — even though a chunk of our own argument leans on it being consequential. If we find that award criteria are usually in the notice after all, we will say that too, even though "we read the documents because the notice doesn't have them" is a large part of why our product exists.

That is the deal, and it is not complicated: you cannot ask an industry to stop asserting unmeasured things and then assert unmeasured things.

We would also rather not do this alone. If you are a researcher working on European procurement, a buyer with data on rejection reasons, or a bid team with a record of why your bids were excluded, we would like to hear from you. The most valuable dataset here — the actual reasons bids get rejected — sits with contracting authorities, and it is the one nobody has assembled.

The mechanisms are documented. The frequencies are not. It is not, in the end, a hard problem. It is just one that nobody has bothered to do.


FindWell reads the tender's own documents — the Aanbestedingsleidraad and its annexes — and shows companies the stated requirements and the award criteria with their weights, each quoted with the section it came from. The corpus described here is the same one that powers it. We will publish findings as we produce them, with methodology, and we will say plainly what the data cannot support.

Sources and the gaps in them

  • Schotanus, F., Van den Engh, G., Nijenhuis, Y. & Telgen, J. (2021). Supplier selection with rank reversal in public tenders. JPSM. — Establishes rank reversal in 303 Dutch tenders. Does not establish how common the enabling conditions are across Europe.
  • Hoekman, B. & Tas, B. (2020). Procurement policy and SME participation in public purchasing. Small Business Economics. — Large-scale, but participation and win rates only; not rejection reasons.
  • Ballesteros-Pérez, P. et al. (2015). Scoring rules and abnormally low bids criteria in construction tenders. CME. — Shows formulas can reweight price. Does not measure how often they do.
  • Flynn, A. & Davis, P. (2017). Investigating the effect of tendering capabilities on SME activity and performance. ISBJ. — The strongest evidence for procedural capability. Self-reported, single country.
  • Van Garsse, S., Verhoeven, S. & Wouters, E. (2022). An Incomplete or Missing European Single Procurement Document? EPPPL. — Legal analysis of consequences. No frequency data.
  • Anelli, D. et al. (2025). Structuring Multi-Criteria Decision Approaches for Public Procurement. — Notes that procurement research is concentrated on the award phase; prequalification is under-studied.