Fifteen minutes of supervision a day and 0.16 points: the VRT loses the award of the cleaning of the Amerikaans Theater because it rewrote its own award criterion
The VRT's specifications stated that the first award criterion (35 points) would measure the hours offered for both operational and supervisory staff, but in the evaluation the VRT counted only the cleaning crew's hours and left out the hours of the contractor's supervisory representative; since Multiple Immo Services scored better than ISS precisely on those hours and the points gap was a mere 0.16, the Council of State annulled the award to ISS: a contracting authority may not narrow the meaning of its award criterion along the way.
What happened?
In December 2009 the VRT, the Flemish public broadcaster, launched a general call for tenders for the cleaning of the Amerikaans Theater in Brussels (specifications TIG 0949). The first of five award criteria, worth 35 points, read: ‘the number of hours offered for operational staff and for supervisory staff and the resulting wage cost’. The hours offered would be compared with a pre-calculation, by a consultancy, of the production hours needed for an acceptable quality level; that benchmark turned out to be 18 hours a day. The specifications also provided (article II.9, c) that the contractor appoints an ‘authorised representative’ who is ‘non-working’ and ‘not to be counted in the minimum working hours guaranteed by the firm’, and who is responsible for the smooth running and supervision of the works. Seven bidders submitted offers. In the evaluation report of 17 February 2010 ISS scored 91.73 points and Multiple Immo Services (MIS) 91.57 — a gap of 0.16 points. On the first criterion MIS scored 33.21 points with 16.36 hours a day, ISS 32.87 with 16 hours. On 17 February 2010 the VRT awarded the contract to ISS. MIS found that, for the first criterion, the VRT had counted only the operational staff hours and not those of the supervisory representative: MIS had offered one hour of supervision a day, ISS a quarter of an hour less (0.75 hours). With those hours included MIS would have offered 17.36 hours a day against 16.75 for ISS, coming closer to the 18-hour benchmark — enough to bridge the minimal points gap. In a letter of 17 March 2010 the VRT's counsel admitted that supervisory staff hours had not been counted for any bidder under the first criterion, because nobody had quoted an hourly rate for supervisory staff; those hours had, according to the VRT, been taken into account under the third criterion (quality of the composition of the daily cleaning team), where both bidders were rated ‘very good’ and thus received 15 points, because a quarter-hour difference was not enough to rate ISS one step lower. In the proceedings the VRT defended a second reading: ‘supervisory staff’ in the first criterion meant the working team leader, not the representative; only actual cleaning hours are ‘production hours’, and the specifications themselves said the non-working representative's hours do not count towards the minimum guaranteed working hours. The Council of State rejected both explanations. The first failed because the bill of quantities asked for global prices per room and no separate price for supervision at all; that nobody had quoted an hourly rate for supervision was therefore no reason to ignore those hours. The second foundered on the text of the specifications: the first criterion expressly referred to the hours of both operational and supervisory staff, and article II.9, c) expressly stated that the representative is responsible for supervising the works. Moreover, in the wage-cost part of that same criterion the VRT had counted all wage costs, including the representative's — it was thus itself inconsistent. The provision on minimum guaranteed working hours was inconclusive: the attached form provided separate boxes for operational staff (team leader included), window cleaning and the representative, which rather implied that the latter's hours do count as working hours. The Council held ‘as a matter of law’ that the representative's hours had to be counted under the first award criterion, which the VRT had failed to do. It was therefore no longer established that the contract had been awarded to the bidder with the most advantageous regular offer. The first part of the first plea was well-founded; the Council annulled the award to ISS and the implicit refusal to award to MIS, and ordered the VRT to pay the costs of 175 euros.
Why does this matter?
Award criteria are a promise. Whoever describes them in the specifications fixes what the offers will be compared on, and bidders tailor their offers accordingly. This judgment shows what happens when the authority quietly narrows that promise at the evaluation stage. The VRT had announced a criterion on the hours of both operational and supervisory staff, and then assessed only the first category. That seemed innocuous — a quarter-hour of supervision a day, compensated through another criterion where both parties scored ‘very good’ anyway — but with a gap of 0.16 points out of almost 92, nothing is innocuous. The Council sets the bar where it belongs: the text of the specifications determines the content of the criterion, and the authority may not reinterpret that content afterwards to justify an evaluation choice. Striking is how the Council weighs the VRT's lines of defence one by one against its own documents: the bill of quantities asked for no separate supervision price, the specifications expressly called the representative a supervisor, the form had a box for his hours, and in the wage cost the VRT had counted his cost. Counting something in one half of a criterion and not in the other is inconsistent — and inconsistency in applying award criteria is a classic ground for annulment. The judgment also recalls that a reference to another criterion (‘we took it into account elsewhere’) does not cure a shortfall in the first: each criterion must be applied as described. For bidders the lesson is equally concrete: read the description of each award criterion literally, offer what it says, and check in the evaluation report whether every component was actually assessed.
The lesson
For contracting authorities: write your award criteria so that you can apply them exactly as they stand, and then do so. If a criterion has several components (here: operational staff hours and supervisory staff hours), each must appear in the evaluation; leaving out a component because the offers contain ‘no separate price’ for it will not hold if your own bill of quantities did not ask for that price. Be consistent within a criterion: whoever counts the representative's wage cost must also count his hours. Never compensate an omitted element through another criterion. For bidders: offer against the literal description of the criterion and make visible in your offer how you score on each component — here one hour of supervision against a quarter less made the difference. After the award, request the evaluation report and check, criterion by criterion, whether all announced components were assessed. With a minimal points gap, every omitted component is a potential ground for annulment, even years after the award.
Ask yourself
Does every component of every award criterion in your specifications explicitly reappear in your evaluation report, with the same meaning as in the specifications? For each term in a criterion (‘supervisory staff’, ‘production hours’), can you point to where the specifications define it, and are your provisions on it mutually consistent? Did you leave out an element of a criterion because the offers gave no separate price for it, while your bill of quantities did not ask for that price either? And as a bidder: after the award, did you check whether your strengths on each component of the criteria were actually weighed?
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 →