CISGIL v IBM concerned the failed implementation of an IT system that was core to CISGIL’s continuing ability to do business, with sums claimed by CISGIL in excess £128 million. By way of background, CISGIL was a wholly owned subsidiary of the Co-operative Group Limited (“Co-op Group”) and was in the business of underwriting and distribution of general insurance products. Following a re-organisation within the Co-op Group, the life insurance division was divested and there was a need to separate CISGIL’s network infrastructure, data centre, hardware, software and applications from that which it shared with the Co-operative Bank plc. Therefore, CISGIL decided to purchase and implement a new IT solution for its business. Following a tendering process CISGIL decided to appoint IBM with Innovation Group as IBM’s main subcontractor providing the core insurance platform Insurer Suite.
Although the project was obviously not a public sector project, its failure and the subsequent legal case bear strong parallels to similar failed public sector transformation projects (in particular those involving system integrators and software solutions requiring localisation for the UK market).
In this article we identify some of the key legal considerations for the public sector. In a following article we will consider why the project failed and what steps might have been taken to avoid failure.
1. The “lottery” of termination
The contract was first entered into in June 2015 and during the first eighteen months, performance was plagued by a series of delays on both sides. Things came to a head in early 2017, when IBM asserted that it had accrued, pursuant to the terms of the agreement, a right to terminate the contract for non-payment of undisputed charges, following failure by CISGIL to pay an invoice raised by IBM for around £3,000,000 in relation to the completion of milestone AG5. IBM eventually relied upon this termination right, terminating the contract for CISGIL’s non-payment in July 2017.
The Court concluded (see point two below) that CISGIL had no valid grounds to dispute the invoice. Instead, the invoice was rejected by CISGIL on the grounds it did not quote a correct purchase order number. A valid purchase order number had not been raised by CISGIL or provided to IBM because CISGIL did not consider that payment for milestone AG5 was due, in light of the failure of the project to achieve earlier milestones; CISGIL did not communicate this as being the actual reason behind the non-issue of the purchase order.
Despite this, the Court also concluded that, as a matter of fact, CISGIL had disputed the invoice by rejecting it and that in its own internal correspondence disclosed in the proceedings IBM had accepted that the invoice was rejected and was in dispute, even crediting the invoice in its internal accounting system and identifying it as disputed.
Thus, since the invoice was in fact disputed (even if for an invalid or trivial reason within the control of CISGIL), the Court determined that IBM had not validly terminated the contract.
Lessons for public sector projects and contracting authorities
The legal risk with termination for cause will almost always be very high. Where a project is failing, the exercise of a right to terminate for breach is very much a double-edged sword. This is because an unjustified termination of a contract amounts to a repudiatory breach. This can be gratefully accepted by the other party who is not only released from its obligation but also earns the right to claim damages arising from the wrongful termination itself.
In the CISGIL vs IBM case this is exactly what happened to IBM once the Court concluded, surprisingly perhaps, that IBM’s purported termination was invalid.
For public sector contracts the dilemma of termination more usually faces a contracting authority which will typically have an armoury of Supplier Termination Events allowing it to bring a contract with a failing supplier to an end. As was the case in the CISGIL/IBM contract, a supplier will typically have no grounds to terminate a contract for breach by its public sector customer except for non-payment of an invoice, usually with safeguards around materiality, time and notification.
The problem in practice is that the factual and legal matrix do not always support a clear and unequivocal termination right for a contracting authority. Often there is contributory cause by the contracting authority and/or delay in exercise of rights which may result in a waiver of those rights. It will seldom be the case that a contracting authority can be even 60% confident that it has a valid right to terminate (and/or receives legal advice to that effect). A first lesson from this case is that getting out of failing contracts can be very difficult for both parties and that however aggrieved a party may feel, often it is better to wait for the other party to pull the trigger.
It’s OK to withhold a PO. A second lesson for contracting authorities is that disputing an invoice (regardless of the merits but so long as there is some belief that payment can be withheld whether based on administrative defects or otherwise) is a valid way of stopping a supplier terminating a government contract for breach. This sounds somewhat cynical but seems an inescapable conclusion.
2. The whole contract and not just the Ts and Cs require legal review.
As indicated above, one of the core issues was whether Milestone AG5 had been achieved and whether payment of the associated milestone payment had become due.
CISGIL’s view was that milestone AG5 had not been achieved, because of delays and failures in the project generally and specifically because preceding milestones had not been achieved. However, when examining the agreement in relation to milestone AG5, the Court ‘s view was that applying the applicable rules of contract interpretation: (i) the only criteria for completion of milestone AG5 was delivery of certain pre-agreed software licences; (ii) milestone AG5 was not subject to any formal acceptance process, unlike certain of the other milestones; (iii) milestone AG5 stood on its own and had no direct predecessor milestones, with the Court specifically noting that the parties could have agreed this in the contract, but they did not do so; (iv) the implementation plan did not include criteria or a process for the application gate milestones; and (v) the only pre-condition to the milestone payment was timing, so CISGIL did not have any discretion to withhold such approval where the stated delivery criteria were satisfied.
As such, the AG5 milestone became due and payable in January 2017; there was no ground on which CISGIL could properly withhold approval. The AG5 invoice was correctly prepared and properly submitted as required by the contract and, as it was set out in the contract to be standalone, the AG5 milestone was not affected by the other delays in the project.
Lessons for public sector projects and contracting authorities
It’s sometimes a temptation not to ask ‘the lawyers’ to look at the technical schedules or to involve them only in a cursory way. This case is a salutary warning as to the need to ensure that schedules 6.1/6.2 and 7.1 ( and indeed all technical schedules) have a thorough legal review. The fact the Court specifically stated that the parties could have agreed a form of sequential dependency, but did not do so, is an important one to note, as it means that, unless otherwise expressly stated, later project milestones can standalone and trigger payment, even if earlier milestones are failed.
3. Due Diligence Warranties might not cover subcontracts
As is often the way with complex disputes, CISGIL formulated a variety of different ways of trying to affix blame to IBM, beyond just pure poor performance of the services. One such approach related to a warranty in the contract that IBM “had satisfied itself as to all risk, contingencies and circumstances regarding performance of the MSA and the Statements of Work”. CISGIL claimed that IBM was in breach of this warranty by reason of its alleged failure to properly assess the fitness of the Insurer Suite software product for the UK insurance market (i.e., such that instead of being a largely ‘out of the box’ implementation, it in fact required a substantial amount of bespoke amendment).
IBM tried – and failed – to run an argument that this language was not a form of additional comfort to CISGIL and should not provide CISGIL with any additional kind of claim or remedy. The Court instead concluded that it was in fact intended to give CISGIL comfort that IBM was not aware of circumstances at the time of executing the MSA that might have a significant adverse impact upon IBM’s ability to perform its obligations, and which could then support a breach claim, if this obligation were to be breached.
However, the Court looked at the wording of the MSA and noted that the warranty as stated above was subject to a qualifier that IBM would be taking ‘reasonable steps’ to satisfy itself. The question then was what would constitute such ‘reasonable steps’. In practice, IBM had undertaken both intensive workshops with CISGIL as to their requirements, and also gap analysis sessions with the licensor of the Insurer Suite product vis-a-vis its suitability for them, which in turn led to negotiations between IBM and the licensor as to time and cost to undertake any necessary redevelopment work. Accordingly, whilst such activities might not in fact have come up with the ‘right’ conclusions, the court concluded that IBM had done all that could reasonably have been expected of it as part of its due diligence activities, and that accordingly it could not be said to have been in breach of the relevant contract warranty.
Lessons for public sector projects and contracting authorities
The warranty given by IBM in this contract has no direct comparator in the Model Services Agreement. The nearest is Clause 2.1 which addresses a different topic, namely the extent to which the supplier has had the opportunity to carry out adequate due diligence on the authority. At stake in the CISGIL case was the extent to which IBM had carried out adequate due diligence on its own supply chain.
There is probably no need to modify the Model Services Agreement as such, but the issue does flag a perennial issue in public sector procurement, which is routes to market, packaging and supply chain management. We will address in our second related article.
4. A relief event mechanism is effective
As milestones were not met, a debate arose as to who was at fault for the delays. Under the Agreement, IBM would “not be liable for any delay or failure to perform .. if and to the extent that such delay … is caused directly by a failure by the Customer to perform the Customer responsibilities that is material or has a material impact … and the Supplier has promptly notified the Customer when it reasonably believes that any failure by the Customer to perform the Customer Responsibilities … has or is likely to have a detrimental effect on the Supplier’s ability to perform its obligations”. No notification was ever made by IBM, pursuant to this drafting, and therefore IBM could not rely on this mechanism to excuse its performance.
Nonetheless IBM argued that a term should be implied in to the Agreement so that IBM should not be liable where the failure was caused by CISGIL or alternatively that a concept known as the ‘prevention principle’ would prevent CISGIL from insisting on performance. Having rejected the argument that a term should be implied (it not satisfying the test for implying a term given it was not required to provide “commercial or practical coherence” and would contradict the express terms), the Court considered the application of the prevention principle. The prevention principle operates so as to afford relief to a supplier where the client’s actions prevent performance; but the prevention must be such that it “rendered it impossible or impracticable for IBM to meet the relevant milestones”; the act must “actually prevent” the carrying on the contractually required activities. Evidentially, CISGIL did contribute to “the collapse of the sprints” that were being used to progress the configuration and implementation activities but they did not prevent the performance by IBM and in fact, IBM was responsible for the delays. As such, the prevention principle was inapplicable.
Lessons for the public sector and contracting authorities
The first lesson is that relief event clauses can work – and so too, it must follow, will comparable obligations upon the contracting authority. This leads one to the age-old lesson, not always applied by the public sector, that parties must operate the contract as written if they later want to avail themselves of its protections.
A second lesson relates to the contribution by contracting authorities to supplier failure. Virtually all transformation projects have material dependencies upon the authority. Indeed, in some cases, they are almost akin to a “ joint venture”. Neither standard procurement processes (which seeks a supplier of services) nor the Model Services Agreement (which entails carefully identified dependencies) fully recognise this and we wonder if the time is ripe to look again at some of the fundamentals as to public sector technology procurement.
5. Claim for wasted expenditure on a failed project
Now to the most interesting aspect of the Court’s judgment from a public sector perspective.
CISGIL claimed for the costs it had incurred in relation to the programme, and which it claimed were now wasted as a result of non-delivery of the solution, arising from IBM’s wrongful termination of the agreement.
The agreement contained an exclusion of indirect and consequential losses, and also of loss of profit, revenue, savings (including anticipated savings), data (with one exception), goodwill and reputation, whether these were direct or indirect losses.
IBM argued that the main component of the loss claimed by CISGIL, the costs incurred in relation to the implementation of the IT system itself, although categorised by CISGIL as ‘wasted expenditure’ was in fact a loss of profit, revenue or savings, and therefore excluded from recovery by the exclusion clause wording. The reasoning for this was that CISGIL’s profit, revenue or savings were, essentially, impacted by the expenditure and so its loss was the inability to make the profits it would otherwise have been able to. CISGIL argued that, if it was put in the position it would have been if the contract was performed, it would be in a break-even position on the basis that the benefit of the delivery of the services/solution would have been worth at least as much as the expenditure; as the services were not delivered, it was not able to achieve this position and therefore its loss was its whole expenditure.
The Court reaffirmed the long established principle that the purpose of damages is to place the innocent party in the same situation as if the contract had been performed. It then assessed that the usual basis for assessing damages is to recompense for the amount of money required to effect the desired contract benefit (i.e. to ‘make good’ the loss) – this would usually be the restatement costs, or to make up lost profits or a diminution in value. This is known as ‘expectation loss’. As an alternative, a party could claim for reliance losses – expenditure incurred in reliance on the defendant’s promises, such as sums paid to the defendant; losses calculated on this basis are said to be calculated on a reliance basis.
Reviewing a line of precedents, the court concluded that – ultimately – contractual damages operate so as to compensate for loss of the contractual benefit but must never put the innocent party in a better position than it would have been if the contract had, in fact, been performed. In this respect, expectation loss is, by definition, loss incurred in expectation of proper performance and therefore it would be proper to look at the (expected) position of the party as if the contract had been performed. This would mean assuming that money spent in expectation of performance would be recouped had the contract been performed, and so the relevant test is to examine the loss of benefit, or loss of bargain, resulting from the breach.
In its judgment, the court agreed with IBM, concluding that wasted expenditure is another means of quantifying the loss of the bargain; and that the loss of bargain/benefit suffered by CISGIL comprised the savings, revenue and profit that it would have made if the IT solution had been properly implemented. In essence, the Court seems to have agreed with the assessment that the wasted expenditure would have been recouped. Because it was not recouped, it meant that CISGIL did not make the profit it expected; therefore the loss was the profit. As indicated above loss of profit was wholly excluded from recovery.
Lessons for the public sector and contracting authorities
To the layman the conclusion of the Court, and the leaps of logic involved, are quite puzzling. A ‘backwards-facing claim’ for wasted expenditure, and a forwards-facing claim for loss of profits seem quite different types of losses and yet the Court seems to have concluded that CISGIL’s claim for wasted expenditure was a claim for loss of profits by the backdoor.
What therefore does this mean for the public sector?
In a similar scenario would a contracting authority finds its claim for wasted expenditure at risk?
The answer would appear to be no for two reasons.
The first and most obvious point is that the public sector is not there to make profit and so an exclusion of claims for ‘loss of profit’ is hardly troubling. That has certainly always been our view. It is true that some trading funds create surpluses and that there are some contracting authorities which are in fact limited companies – but by and large profit is not what the public sector is about.
This public sector point was actually considered by the Court in the CISGIL case.
The Court referred to a previous case where the supplier sought to exclude a claim by an NHS Trust for wasted expenditure arising from a failed electronic medical records project on the grounds that it amounted to a claim for loss of profits, revenue or savings,which were excluded. The Royal Devon and Exeter NHS Foundation Trust v ATOS IT Services UK Ltd  EWHC 2197 (TCC).
In the Royal Devon case the Court (same judge) rejected the supplier’s assertion stating that the claim for wasted expenditure was “based on a rebuttable presumption, not that the EMR system would produce revenues to defray such expenditure, but rather that the use of the EMR system was worth at least the expenditure incurred”. As a result, the Trust was able to recover its own wasted costs and also the charges it had paid ATOS for a nugatory project.
CISGIL might well wonder why it was not afforded a similar rebuttable presumption, something that will perhaps be considered in the event of an appeal, as it seems to lead to contradictory outcomes depending on whether the customer is in the private or public sector.
A second point is that this issue is directly addressed by the Model Services Agreement which provides:
“ 26.7 Neither Party shall be liable to the other Party for:
(a) any indirect, special or consequential Loss; or
(b) any loss of profits, turnover, business opportunities or damage to goodwill (in each case whether direct or indirect).
26.8 Notwithstanding Clause 26.7 but subject to Clause 26.4, the Supplier acknowledges that the Authority may, amongst other things, recover from the Supplier the following Losses incurred by the Authority to the extent that they arise as a result of a Default by the Supplier:
(a) any additional operational and/or administrative costs and expenses incurred by the Authority, including costs relating to time spent by or on behalf of the Authority in dealing with the consequences of the Default;
(b) any wasted expenditure or charges”.
6. A repudiatory breach does not necessarily amount to wilful default
CISGIL sought to argue that IBM’s termination was a wilful default. The reason this was an important point was because the contract provided for an increased liability for IBM if it was in wilful default.
The basis for CISGIL’s argument was that IBM knew that there were various reasons why its termination, based as it was on an unpaid invoice, could not be justified. Those included that the preceding milestones had not been achieved, and therefore the invoice was not due, that IBM did not have a valid purchase order number for the invoice, and that CISGIL had in fact disputed the invoice. Therefore, IBM’s purported termination was a deliberate breach of the contract and IBM was reckless as to the financial consequences for CISGIL.
IBM’s countered that it genuinely believed the invoice to have been payable despite the absence of a purchase order number, and in any event it did not consider that CISGIL had validly rejected or disputed the invoice. IBM’s position was that it was entitled to exercise its termination rights, and that proper consideration of both its internal discussions and external communication with CISGIL would bear this out.
The Court preferred the argument of IBM, considering that IBM terminated in the genuine but mistaken belief that they were entitled to do so. This brought about the interesting consequence that, though the Court found that that IBM’s termination was a repudiatory breach, this in itself did not automatically make the breach into a wilful default. In reaching this decision, it is relevant to note that the contract contained a definition of wilful default, which was “an intentional breach of the Agreement with either an intent to cause harm or recklessness with regard to the consequences of the breach.”
Lessons for the public sector and contracting authorities
The Model Services Contract also provides for an uplift in limits of liability “as a result of the Supplier’s abandonment of this Agreement or the Supplier’s wilful default, wilful breach of a fundamental term of this Agreement or wilful repudiatory breach of this Agreement”.
Unlike the IBM MSA there is no definition of ‘wilful’ but there is a specific reference to repudiatory breach. Despite the additional reference to repudiatory breach in the Model Services Agreement, we believe that a similar fact pattern would most likely lead to a similar outcome to that in the CISGIL case, given the qualification of ‘wilful’.
The Court’s decision is likely to have pleased no one. If there in any overarching lesson at all it is perhaps that disputes such as this are best avoided. As for the public sector, the Model Services Contract (which should be used for comparable projects at least by central Government) is helpful but other central Government or wider public sector contracts (such as the mid-tier contract) or frameworks should be reviewed to ensure that the lessons from this case are reflected.
In our second related article ( CISGIL Vs IBM- the Practical Lessons Part Two) we will consider the broader lessons the public sector can take from this project’s failure.
For more information please contact Richard Bonnar or Simon Kenyon. They would like to thank the Duncan Pithouse, Chloe Forster, Kit Burden and John McKinlay who were the original authors of this article.
 See for example Schedule 1 (v 1.09) Cabinet Office Model Services Agreement
 Of the Model Services Agreement (Implementation Plan, Testing Procedures, Charges and Invoicing).
 Of the Model Services Agreement, version 1.07.
 In Schedule 3 Model Services Agreement (Authority Responsibilities)
 As to which we have put forward some detailed ideas in our response to the Cabinet Office Green Paper on Public Procurement Reform
 Clause 26.7 and 26.8.