Every award rule makes you publish the weighting. None makes you defend it.

Somebody decided the tender would be scored 70 on quality and 30 on price.

It is worth asking where that number came from, because it is the most consequential number in the procurement and it is almost never argued about. It arrives in a template, survives review because nobody has a better one, and is published because the rules say it must be published.

Which they do. All of them.

What every instrument requires

Three instruments, three jurisdictions of very different reach, one requirement. The weighting is disclosed in advance, and the evaluation is bound to it afterwards.

That is a good rule and it is worth saying so. It removes the failure mode where a panel reads the proposals, notices which one it prefers, and then chooses the weights that produce that answer. Publish first and you cannot do that.

What none of them requires

None of them says what the weighting should be.

That sentence is the Model Law's own commentary, which is the closest thing procurement law has to a basis for conclusions, and it is explicit. The weights are the buyer's business. The rule is about disclosure, not about the number.

The nearest anything comes to constraining it is Article 67(4) of the EU directive, which says award criteria "shall not have the effect of conferring an unrestricted freedom of choice on the contracting authority" and must be accompanied by specifications that allow what the tenderers say to be effectively verified. That governs the criteria. It does not govern their relative size.

The arithmetic is specified once, and it is for the machine

Here is the part that is exhibitable rather than arguable. Put two clauses of the same treaty side by side.

For a normal tender, GPA Article X:7(c) asks for "the relative importance" of the criteria. Not the formula. Not the weights, even. The importance, relatively.

For an electronic auction, GPA Article XIV(a) requires the entity to give each participant, before the auction starts, "the automatic evaluation method, including the mathematical formula, that is based on the evaluation criteria set out in the tender documentation and that will be used in the automatic ranking or re-ranking during the auction".

The Model Law does the same thing in its own vocabulary. Article 11(4) says that "To the extent practicable, all non-price evaluation criteria shall be objective, quantifiable and expressed in monetary terms", which is a shall carrying its own escape hatch. But Article 31(1)(c) makes criteria that "are quantifiable and can be expressed in monetary terms" a precondition for using an electronic reverse auction at all. No monetisation, no auction.

So both instruments know exactly how to demand a complete, unambiguous, fully specified evaluation. They demand it when a computer performs the ranking, and they stop short when a person does.

What 70/30 actually says

So work it out.

Take the price formula the World Bank publishes, which is the one most widely used and one of the few written down anywhere: the lowest offered price takes the full price allocation, and every other price scores inversely proportional to it. Quality is scored out of 70, price out of 30.

A bidder who is one quality point ahead of the lowest bid can be 3.4% more expensive and still tie. Ten points ahead, 50% more expensive. Fifteen points, twice the price.

The price premium a quality lead buys over the lowest bid

Read the 70/30 curve at ten points: a bidder ten quality points ahead of the lowest bid can charge 50% more and still tie, which on a 100m contract is 50m. Each dashed line marks a lead equal to the whole price allocation, 20 points under 80/20 and 30 under 70/30. Past it no price wins the tender, because price can only ever move a bid by its own allocation. The 50/50 wall is at 50 points, beyond the frame and beyond any real spread of quality scores.

Check one point on it. On a 100m contract with a 70/30 split, a bid at 100m scoring 60 on quality gets the full 30 price points and totals 90. A bid at 150m scoring 70 on quality gets 30 multiplied by 100 over 150, which is 20 price points, and totals 90 as well. The two are tied. The tender has declared, without anybody writing it down, that ten quality points are worth 50m.

The gap no price can close

The second consequence is sharper, and it is the finding two European Parliament procurement officers put in print in 2015.

Price can only ever move a bid by the whole price allocation. Under 70/30 that is 30 points. So a bidder more than 30 quality points behind, which is 43% of the quality range, cannot win at any price. Not a lower price. Not a free bid.

Under 80/20 the uncatchable gap is 20 points of 80, a quarter of the quality range. Under 90/10 it is one ninth. Kiiver and Kodym describe the effect as an implicit minimum quality threshold that the tenderers, and perhaps the buyers, are not necessarily aware of, and they draw the structural conclusion: where price and quality scores are added, the price weight should not fall below 50%.

Their paper also shows the price formula punishing the middle. With bids of 50, 75 and 100, the inverse-proportional method scores them 100, 67 and 50. A linear distribution between the same endpoints would give the middle bid 75. It loses about 8 points, roughly 10% of the score it would otherwise hold, for no reason other than sitting in the middle of the range.

The same instrument, one stage earlier, where the Green Book advises against it

, read: Weighting and scoring is the method the guidance warns againstWeighting and scoring is the method the guidance warns against

One end of the scale, and the other

The World Bank publishes a weighting range, a minimum technical score, a merit point range and the shape of the price formula. Quality and cost weightings of 90/10 for highly complex assignments, through 70 to 80 against 30 to 20 for moderate complexity, down to 60 to 50 against 40 to 50 for routine work. A minimum technical score normally between 70 and 85%.

The Saudi Executive Regulations publish none of the four. Articles 28 and 29 require the criteria to be clear, objective and not designed to favour a particular bidder, permit a minimum pass mark with the lowest price winning where the works do not need high technical capability, and say that technical criteria carry the higher weight for consulting services that do. The formula, the weighting range and the merit point range are all delegated.

Both are award rules for public money. They are two very different quantities of published constraint on the same judgement, and the comparison is drawn from the two documents rather than inferred.

What the score the weighting multiplies rests on, and how it can be reviewed

, read: The score that is not a numberThe score that is not a number

Where the number can be argued with

There is a published method that puts the figure where somebody can disagree with it, and it is not subtle about what it is doing.

Two levels per parameter, a threshold that is the least the buyer will accept and an objective that is the most it can use, and a stated price for movement between them. That is the same information a weighting already contains. The difference is that it is in money, in the open, and in a form a bidder can respond to and an auditor can question.

It is also what UNCITRAL Article 11(4) asks for in the first place, and what EU Article 68 disciplines when the costs run over a life cycle: the authority must state both the data tenderers provide and the method it will use, and an externalities method must rest on objectively verifiable and non-discriminatory criteria, be accessible to all interested parties, and need only data a diligent bidder can supply with reasonable effort.

None of them supplies the number. That is not a gap in the guidance. A threshold and an objective are statements of what the buyer wants, and no document about procurement can know that.

A note on the two envelopes

One procedural control does remove a bias rather than relocating it, and it is usually attributed to a document that costs 150 euros.

The Model Law has it, free, in Article 47. Proposals arrive in two envelopes, technical and financial. The technical envelopes are examined and evaluated before any financial envelope is opened. Proposals that fail the minimum requirements are rejected and their prices returned unopened. Every responsive bidder is told its own technical score, invited to the opening, and hears each score and each price read out together.

Then Article 47(10) stops: the successful proposal is the one with "the best combined evaluation" of the non-price criteria and the price, and the Model Law does not say how to combine them.

Which is where this piece started.

Why a total over judgements you supplied returns your own beliefs with a number attached

, read: A dashboard computed from your own beliefs is not informationA dashboard computed from your own beliefs is not information

What to do on Monday

Take the weighting in the tender document on your desk. Divide the price allocation by the price allocation minus one, subtract one, and multiply by the contract value. On a 30 point price allocation that is 30 divided by 29, less one, which is 3.4%. That is what one quality point is worth in money on this tender.

If the answer is a number you would have defended in a meeting, the weighting is doing what you meant. If it is not, the weighting has been making a commitment on your behalf, published in the tender documentation exactly as the rules require, and read by every bidder who bothered to do the division.

Or put your own split, contract value and bids into the weighting calculator, which does the same division and shows the lead no price can close.

The other number in the same tender that nobody prices

, read: The payment chain is a loan, and the contractor is the lenderThe payment chain is a loan, and the contractor is the lender

---

Sources. WTO Agreement on Government Procurement, as amended 30 March 2012, Articles X:7(c), X:11, XIV(a) and XV:5, read at the WTO legal text. UNCITRAL Model Law on Public Procurement, 2011, Articles 11, 31(1)(c) and 47, read in full at the United Nations. UNCITRAL Guide to Enactment, part two, on the procuring entity's discretion. Directive 2014/24/EU, Articles 67 and 68, read in the Official Journal text at EUR-Lex. World Bank Procurement Regulations for IPF Borrowers, seventh edition, September 2025, Annex X paragraphs 4.5 and 4.8 and Table 2. US Department of Defense Source Selection Procedures, USA000740-22-DPC, paragraphs 1.3.1.4, 2.3.2.5 and 3.12. Niels Kiiver and Jakub Kodym, Price-Quality Ratios in Value-for-Money Awards, Journal of Public Procurement 15(3), 2015, pages 275 to 290, free in full at ippa.org, for the worked price table, the penalty on a mid-range bid and the minimum price weight. Saudi Executive Regulations of the Government Tenders and Procurement Law, Articles 28 and 29; the 2019 law and its regulations are replaced by the law gazetted on 4 September 2026 when it takes effect. Every premium and threshold in this piece is computed from the price formula in Annex X paragraph 4.8, and the worked tie at 100m and 150m is checked arithmetically rather than asserted, in the engine behind the weighting calculator as well as by hand.