Suspended a second time: INASTI cannot show that Infobase and Coface were put through the same test in awarding the company-database contract
After the Council of State had already suspended, in August 2011, a first award of INASTI’s company-database contract to Coface Services Belgium, the institute withdrew that decision, redid the assessment and on 21 December 2011 awarded the contract to Coface again (51,000 euros a year, against 152,000 euros for Infobase); the Council suspended that second award too, because the file did not show that both databases had been subjected to the same queries for the ‘speed’ criterion, and because INASTI could not support with relevant reasons its finding that Coface was ‘clearly superior’ — what was missing was not the ‘reasons for the reasons’ but simply a pertinent justification.
What happened?
On 14 January 2011 the National Institute for the Social Security of the Self-employed (INASTI/RSVZ) published an open call for tenders for access to a computerised database on companies active in Belgium: articles of association and publications, internal, structural and financial data, and — by querying the name or address of a natural person — the existence of self-employed activities or company directorships. The contract, previously held by Coface Services Belgium, comprised a fixed and a conditional tranche; only the fixed tranche was at issue here. The specification set four award criteria: price (30%), completeness and accuracy of the information (25%), speed of data updates (25%) and quality of the technical infrastructure (20%). The three qualitative criteria were to be assessed through a test by some fifteen INASTI users, lasting at most one month and involving at least 1,500 queries; each tenderer had to provide a test environment with access codes. Three tenders came in: Coface (51,000 euros a year), Infobase Europe (152,000 euros) and Company Web (28,000 euros). In the background lay a long-running dispute: Infobase, which had taken over its database from the company HELP in 1998, had successfully sued Coface (formerly Euro DB, a subsidiary of the French OR Télématique) for infringing its rights under the Act of 31 August 1998 on the legal protection of databases, with judgments of the Brussels Court of Appeal in 2008 and 2009. A first award to Coface, notified on 13 July 2011, was suspended by the Council of State in judgment no. 214.780 of 10 August 2011. INASTI withdrew that decision on 7 September 2011 in order to redo it; Infobase brought no action for annulment and the suspension was lifted on 7 December 2011. In the new award decision of 21 December 2011 INASTI first expressly examined the exclusion grounds in respect of Coface: the optional exclusion for grave professional misconduct (article 69(2)(4) of the Royal Decree of 8 January 1996) was not applied, because the dispute with Infobase was almost twenty years old, cast no doubt on Coface’s professional integrity and gave it no competitive advantage within the meaning of article 11 of the Act of 24 December 1993. The three databases were then compared on sixteen sub-criteria. On speed (sub-criteria 1 and 2) Coface was found ‘clearly superior’ on sub-criterion 1 — the information shortly after publication of the deed of incorporation was said to be more complete — and ‘slightly inferior’ on sub-criterion 2, because a copy of an old deed was more often immediately available at Infobase. On completeness Coface won slightly, on infrastructure Infobase won slightly; unweighted, Coface scored 1,355 against 1,319 for Infobase, a difference of 36 points. After proportional conversion Coface obtained 69.41 out of 70 on the three qualitative criteria and Infobase 68.13; on price Coface received 16.47 out of 30 and Infobase 5.53. Final totals: Coface 85.88, Company Web 81.22, Infobase 73.65. Infobase received the notification of 19 January 2012 on 23 January and on 3 February 2012 — within the fifteen days of article 65/23 — again sought suspension under extreme urgency. Coface intervened at the hearing of 17 February. Acting president Paul Lewalle confined himself to the fourth plea, which attacked the assessment of the sub-criteria. In its first two branches Infobase showed, using the query logs, that for sub-criterion 1 (‘do recent deeds appear quickly in the database?’) INASTI had looked up exactly three 2011 incorporations, respectively 6, 8 and 55 days after publication, with only one deed of incorporation actually consulted, and that the six 2011 deeds consulted were between 6 and 88 days old — hardly a test of ‘recent’ deeds, and the only recent document was simply available at Infobase. For sub-criterion 2 (‘is an old deed obtained immediately?’) INASTI had been able to view 39 pre-1990 deeds directly on screen at Infobase, whereas at least 22 of them could only be ordered by e-mail from Coface, with no delivery time stated: Coface thus offered fewer than half (43.58%) of those old deeds immediately online, which made the label ‘slightly inferior’ hard to justify. INASTI replied that Infobase was demanding ‘the reasons for the reasons’, that the queries in the three databases did not have to be identical because the testers searched according to the needs of their departments, that failed queries did not appear in Infobase’s logs, and proposed in the alternative an expert report or a hearing of the testers. Coface said nothing about the plea. The Council found that neither INASTI nor Coface disputed the facts set out in the first two branches, so that they were prima facie established — even though Infobase relied on listings of figures and abbreviations it did not see fit to explain. It recalled the preparatory works of the Act of 23 December 2009: the extreme-urgency suspension is meant for manifest irregularities that cannot bear a longer procedure, so that a plea is serious only if it reveals a manifest or at least apparent irregularity. That was the case here. Despite the obscurities in the documents, nothing showed that Infobase and Coface had been subjected to the same tests for the speed criterion, and INASTI did not justify that difference in treatment. Moreover, the documents attached to INASTI’s note — the intermediate results table and a screenshot of the number of queries — did not reveal what founded the conclusion that Coface was ‘clearly superior’ on sub-criterion 1 and only ‘slightly inferior’ on sub-criterion 2. INASTI had failed, in the Council’s words, not to state the reasons for its reasons, but more simply to justify its decision in pertinent terms. The plea was serious on the ground of breach of the principle that every act must rest on accurate, relevant, legally admissible and non-contradictory reasons, and of the equal treatment of tenderers. As INASTI did not invoke a balance of interests, the award of 21 December 2011 was suspended, with immediate enforcement and notification by fax; costs were reserved.
Why does this matter?
A user test is a tempting way to assess software tenders: let fifteen staff members search for a month and count the results. This judgment exposes the weak spot of that method. As soon as the testers run different queries in each database — because they search for what their department needs, as INASTI itself admitted — one is no longer comparing products but incidental user experiences. The Council of State then demands two things: that the authority can show the tenderers were subjected to the same test, and that it can support its conclusions with the test results. An intermediate table and a screenshot of the number of queries do not suffice. The judgment is also a lesson on redoing a decision after a suspension. After the first suspension of August 2011 INASTI had duly examined the exclusion grounds and redone the award, but the assessment itself still rested on the same test with the same flaws. Whoever redoes a suspended decision must fix the problem, not merely thicken the reasoning. Procedurally the judgment confirms two markers from the 2009 legal-protection rules: an applicant who applies within the fifteen days of article 65/23 need not further justify extreme urgency, and a plea is serious only if it brings to light a manifest or apparent irregularity. That is a high bar, and Infobase cleared it — with listings the Council itself called ‘obscure’ — mainly because INASTI and Coface did not dispute the facts. Finally, the case shows that an applicant whose tender is three times more expensive (152,000 against 51,000 euros a year) and who trails by more than twelve points in the final score can still obtain a suspension: the interest lies not in the chance of winning after all, but in the right to a correct and equal assessment. The dispute over database rights between the two parties, which INASTI regarded as too old to justify exclusion, played no part in the ruling.
The lesson
If you are a contracting authority assessing through a user test, design that test as an experiment: a fixed list of queries that every tester runs identically in every database, logged per tenderer and per sub-criterion, with a definition of ‘recent’ and ‘old’ that fits the sub-criterion. Make sure you can show afterwards, per sub-criterion, which queries led to which score; ‘clearly superior’ without that underpinning will not hold. If you redo a suspended award, do not repeat the same assessment with a longer statement of reasons, but cure the defect that led to the suspension. If you are a tenderer and lose after a test, request the query logs and reconstruct exactly what was tested: the number of searches, the age of the documents consulted, and whether the same documents were requested from the competitor. Show concretely where the test diverges, and apply within fifteen days — the extreme-urgency procedure is then open without further justification. But let your exhibits speak for themselves: the Council read Infobase’s listings charitably here but called them obscure; a short reading guide with each exhibit would have made that charity unnecessary.
Ask yourself
As a contracting authority: can you prove for each sub-criterion that all tenderers were subjected to the same queries, and can you trace the labels ‘clearly superior’ or ‘slightly inferior’ back to concrete test results? After a suspension, did you cure the underlying defect or merely expand the reasoning? As a tenderer: have you requested the query logs and checked whether the test measures what the sub-criterion describes — were the ‘recent’ deeds recent, and were the same old deeds requested from the competitor? Have you annotated your exhibits so the Council does not have to decipher them itself? And do you know that an application within the fifteen days of article 65/23 suffices to justify extreme urgency, but that your plea must demonstrate a manifest or apparent irregularity to be serious?
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 →