One missing test certificate, four points of difference: the Council of State upholds the award of the Integrated Police’s public-order protective equipment
In the award of lot 1 of the framework agreement for visibly worn GBOR protective equipment for the Belgian Integrated Police, the winning tender finished on 57.51 points against 56.20 for the applicant, and the Council of State rejected all three of its pleas as not serious — on the word ‘sufficient’ in the evaluation texts, on the formula for the ‘warranty’ award criterion, and on the absence of an announced ‘operational assessment’ for stab resistance — with a procedural indemnity of 770 euros as the result.
What happened?
On 28 May 2025 the Federal Police launched a public supply contract for a multi-annual framework agreement for the purchase of visibly worn GBOR protective equipment — GBOR standing for the negotiated management of public space — for the Integrated Police and the police academies. In short, this is the gear officers visibly wear during public-order operations: a protective vest with upper-arm protection, forearm, shin and thigh protection, groin protection, a transport bag, visual-identity elements, and holders for the collective spray, the short and long baton and the 56 mm tear-gas grenade. The procedure was an open one; tender documents ‘Procurement no. 2025 R3 024’ split the contract into two lots, of which lot 1 is at issue here. The award criteria for lot 1 were finely divided: price 25 points, a small-quantity surcharge 5, operational aspects 43 — comprising freedom of movement and ergonomics 13, comfort and thermal sensation 9, ease of donning and adjustment 6, protection 4, resistance to Molotov cocktails 4, transport bag 3, resistance to corrosive products 2 and use of holders 2 — mechanical characteristics 16, split into VPAM impact resistance 12 and stab resistance 4, submitted permitted options 4, standard sizes 3 and warranty 4. An information session was held on 18 June 2025, minuted, and until 25 August 2025 candidates could put further questions on a digital forum. Three candidates submitted a tender for lot 1; two of them were found regular. On 18 June 2026 the minister awarded lot 1: 57.51 points for the party that later intervened, 56.20 for the applicant. A gap of 1.31 points. The applicant applied on 3 July 2026 for suspension under extreme urgency; the beneficiary intervened on 13 July; the hearing was held on 23 July 2026. A pleading note lodged by the applicant the day before the hearing was excluded from the debate at the other parties’ request — the rules of procedure make no provision for such a document — but was taken into account as mere information in so far as it reflected the oral argument. The first plea turned on a single word. The contracting authority worked with a rating scale: a descriptive evaluation leads to a qualification, and that qualification corresponds to a mark out of ten — ‘acceptable’ 0, ‘satisfactory’ 2, ‘fairly good’ 4, ‘good’ 6, ‘very good’ 8 and ‘excellent’ 10 — which is then converted to the weight of the criterion. The applicant noted that the evaluations of the winning tender for ‘use of holders’ and ‘sizing system’ contained the word ‘sufficient’ — the holders are made of a material that retains its shape ‘sufficiently’, all holders can be fastened ‘sufficiently’ firmly, the size range is ‘sufficiently’ broad — while 6 out of 10 was awarded each time. It calculated that those two scores were together two points too high (1.2 instead of 0.4 out of 2 for the holders; 1.8 instead of 0.6 out of 3 for the sizing system), so that the winner would end at 55.51 and the applicant, at 56.20, would rank first. The Council did not follow. Both evaluations end expressly with the sentence ‘The tender is assessed as good. The tender receives 6/10.’ The tender was therefore assessed not as ‘sufficient’ but as ‘good’, and the word ‘sufficient’ in the descriptive text makes no reference to the rating scale. The Council added a sharp observation: the evaluation of the applicant’s own tender for ‘sizing system’ also states that the size range is ‘sufficiently’ broad, while that tender was found merely ‘acceptable’ on that point and thus scored zero — an assessment the applicant did not contest. The plea proceeded from a false premise and lacked any factual basis. The second plea had two limbs, both about the ‘warranty’ criterion. The applicant first argued that the authority should also have assessed the warranty on the protective properties of the material, not only that on the protective elements themselves. The Council read the tender documents differently: only the latter item specifies that longer periods count, and a forum answer confirmed it — ‘As regards point 16b, a better score will be given to the tender offering a guarantee of more than 2 years.’ The other parts of the warranty scheme therefore play no role in the comparison, and no separate reasoning was required on them. The second limb was arithmetical. The warranty score is calculated by a formula in which ‘Wx’ is the warranty period of the tender being assessed, ‘WMin’ the minimum period set in the tender documents and ‘WMax’ the maximum period offered. The entire dispute lay in ‘WMax’. The authority used the weighted average of the periods offered across the five relevant items — six years; the applicant took the longest period for a single item — ten years, arriving at 2 out of 4 for the winner instead of 4 out of 4. The Council pointed out that the tender documents expressly spread the warranty over the five items at 20 per cent each, and that the applicant used that weighted average for ‘Wx’ but a single item for ‘WMax’: on that basis any comparison collapses. The authority applied the average consistently on both sides of the fraction, which is not unreasonable. The applicant’s further remark at the hearing — that the formula itself does not appear in the tender documents — was in any event out of time, since it could have been made on reading the contested decision. The third plea concerned the sub-criterion ‘stab resistance’, worth 4 points. The decision is terse on that point: a tender that supplies a test certificate to a recognised standard gets all the points, one that does not gets none. The winner supplied a VPAM certificate — resistance level K1 — and received 4 points; the applicant supplied no certificate and received 0. It relied on point 23 of the minutes of the information session, where the question whether certificates had to be provided in relation to perforation was answered: ‘No, certificates are not required for this point. It is an operational assessment.’ The Council untangled that confusion. The question concerned point 5.4.4 of the technical specifications — ‘The protective elements offer the best possible protection against perforation by sharp objects’ — which sits under heading 5 ‘Operational aspects’ and relates to the sub-criterion ‘protection’. Stab resistance sits elsewhere, under heading 6.2 ‘Mechanical characteristics’, where point 6.2.5 expressly provides that the tenderer shall attach the test certificate of a recognised laboratory if the vest also resists knife attacks to a recognised standard, and that this is an award criterion. On the forum the point had moreover been pressed expressly: would a stricter standard score higher than a looser one? The authority’s answer was a single word: ‘No.’ It follows that supplying a valid certificate suffices for the maximum score and that only two possibilities exist — certificate or no certificate. The applicant did not contest that all-or-nothing method as such; it only found its application problematic in the light of the announced operational assessment, and it failed to persuade on that. None of the three pleas proved serious. On 24 July 2026 the Council admitted the intervention, dismissed the application and ordered the applicant to pay the costs: 200 euros in roll fees, 26 euros in contributions and a 770-euro procedural indemnity payable to the Belgian State. The intervening party bears the 150-euro roll fee for its intervention. The Council did not need to rule on the defendant’s objection to admissibility.
Why does this matter?
This judgment is about something that sinks far more contracts in practice than any grand legal principle: reading an evaluation report, a set of tender documents and the answers on a question forum with precision. Three times the tenderer thought it had spotted an error, and three times the error lay in the reading. The first lesson is the most surprising. An evaluation text mixes ordinary words with technical terms. ‘Sufficiently’ in ‘the holders can be fastened sufficiently firmly’ is an ordinary adverb; it is not a qualification on a scale which does not even contain a level called ‘sufficient’ — the scale runs from ‘acceptable’ through ‘satisfactory’, ‘fairly good’ and ‘good’ to ‘very good’ and ‘excellent’. The Council disposes of the confusion by pointing to the closing sentence of each evaluation, where the qualification and the mark appear in so many words. And it turns the argument around: exactly the same ‘sufficientlys’ appear in the evaluation of the applicant’s own tender, where they led to zero points. Anyone wishing to build an argument on a word from an evaluation text must first check how that same word operates in their own assessment. The second lesson is methodological and broadly applicable. In any relative scoring formula — and tender documents are full of them, for price, lead time, warranty or delivery — the numerator and the denominator must refer to the same quantity. The applicant used a weighted average across five items for its own offered period, but the longest period of a single item for the reference value. The result of such a mixed calculation means nothing, and the Council says as much: on that basis any comparison collapses. It is also useful that the Council refused to deny the applicant an interest in the plea: the point that with two tenderers one will always score zero and the other the maximum goes to the merits, not to admissibility. The third lesson concerns the status of information sessions and question forums. They are genuinely relevant to interpreting the tender documents — the Council itself uses two forum answers here to determine what the ‘warranty’ and ‘stab resistance’ criteria mean. But an answer governs only the question asked. ‘No certificates for perforation, that is an operational assessment’ related to point 5.4.4 under ‘Operational aspects’ and had nothing to do with point 6.2.5 under ‘Mechanical characteristics’, where the certificate was expressly requested. Anyone relying on an answer must be able to point to the clause it hangs on. Finally, note what the figures themselves say. The gap was 1.31 points; on ‘stab resistance’ alone 4 points were at stake, with the winner scoring 4 and the applicant 0. The certificate that would have earned those points was set out in point 6.2.5 of the technical specifications, with the words ‘This is an award criterion’ attached. The Council draws no such conclusion — it reviews legality, not commercial strategy — but for anyone drafting tenders it is inescapable: a non-mandatory document that earns points is mandatory in practice.
The lesson
For tenderers: comb the technical specifications for every sentence ending in ‘this is an award criterion’. Optional certificates, attestations and test reports are rarely genuinely optional — here the difference between 0 and 4 points on ‘stab resistance’ was considerably larger than the 1.31 points that cost the contract. If you rely on an answer from an information session or a question forum, check which clause the question hung on; an answer about ‘protection’ does not cover a criterion about ‘mechanical characteristics’. When reading an evaluation report, compare the closing sentences containing the qualification and the mark rather than the adjectives, and place the winner’s evaluation alongside your own before alleging unequal treatment. If you want to challenge a scoring formula, do so in your application: criticism first raised at the hearing when it could have been raised on reading the decision is out of time. And do not lodge a pleading note the rules of procedure do not recognise. For contracting authorities: this file shows how an award decision should be built. The rating scale was fixed in advance and every evaluation closed with an explicit qualification and a mark, leaving nothing to interpret afterwards. The formula was applied consistently, on both sides of the fraction. And the forum answers were precise enough — ‘a better score will be given to the tender offering a guarantee of more than 2 years’ and, to the question whether stricter standards score higher, ‘No’ — to serve later as instruments of interpretation. If you score all-or-nothing on a certificate, confirm that in advance as clearly as was done here.
Ask yourself
Have you traced every document that the technical specifications say counts as an award criterion, even where it is not mandatory? For every answer from the information session or the question forum you rely on, do you know which clause it hangs on? When reading an evaluation report, did you compare the qualifications and marks rather than the adjectives — and place the winner’s alongside your own? Does your challenge to a scoring formula use the same quantity in the numerator and the denominator? And as an authority: does every descriptive evaluation you write close with an explicit qualification and the corresponding mark?
About this database
The Council of State (Raad van State / Conseil d'État) is Belgium's supreme administrative court. In disputes over public procurement — from contract awards to tenderer exclusions — the Council of State is the final arbiter. The rulings in this database are summarised by TenderWolf in plain language, with practical lessons for tenderers and contracting authorities. View all rulings →