The multi-million pound Centrica v Accenture litigation was another major adverse finding against an IT supplier that cut across clauses intended to limit liability. The judgment contains lessons beyond the narrow point that exclusions for indirect and consequential loss are of limited practical use. While the drafting of the exclusion and limitation clauses had clearly taken into account previous case law, other pitfalls were revealed by the Court’s analysis which need to be borne in mind for the future.
A brief reminder of the facts
The case is reported as GB Gas Holdings Limited v Accenture (UK) Limited & Others [2010] EWCA Civ 912.
Accenture agreed to design, supply, install and maintain a new IT system for Centrica (the ‘Jupiter System’). The parties envisaged that the Jupiter System would be released in five stages, the third of which was to be an automated billing system that would, amongst other things, identify incorrect meter readings, calculate bills based on actual or estimated meter readings, send invoices to be printed and receive payments.
Work commenced on the Jupiter System in 2002 and shortly afterwards there were disputes between the parties relating to releases 1 & 2. By the time release 3 (the automated billing system) was rolled out, the relationship had deteriorated to such an extent that Centrica had asked Accenture to leave the site and had taken over migration and stabilisation of the system itself. Releases 4 & 5 were cancelled.
When the automated billing system was eventually implemented, considerable problems emerged, resulting in millions of customers being undercharged, overcharged, or simply not charged at all. As a result, Centrica had to raise thousands of invoices manually and a large backlog ensued. Accenture accepted that there were some errors in the billing system but the parties disagreed as to whether Accenture’s breaches constituted a Fundamental Defect within the meaning given to that phrase in the contract. Centrica sued for damages.
The losses
The Court of Appeal answered various preliminary issues. Amongst other things it was asked whether in principle the following losses were recoverable from Accenture:
1 £18,700,000 in charges overpaid by Centrica to wholesale gas distributors because of over-estimates of gas usage due to lack of automated meter readings;
2 £8,000,000 in ex gratia payments made to customers in an attempt to limit reputational damage;
3 £2,000,000 in additional borrowing charges caused by a drop in cash flow as a result of late or non-billing of customers;
4 £387,287 in costs wasted trying to enforce debts that were not actually due; and
5 £107,120 in stationery and correspondence costs.
The recoverable losses
The contract excluded liability for any indirect or consequential losses but the Court held that each of the losses set out above constituted direct losses and were therefore recoverable under the first limb of Hadley v Baxendale. This is a result which will not in principle surprise IT lawyers, but the Court of Appeal’s acceptance of the recoverability of some of the individual heads of loss may cause an eyebrow or two to be raised, particularly in connection with the ex gratia payments to customers.
Even someone familiar with the English courts’ interpretation of Hadley v Baxendale might find it strange that voluntary payments following a strategic decision by Centrica to preserve goodwill from its customers should be held to arise ‘naturally’ (ie according to the usual course of things) from Accenture’s breach of contract. A broad interpretation of ‘direct loss’ is nevertheless consistent with a long list of previous authorities.
It has also long been good law that claimants may recover as damages for breach of contract the reasonable costs incurred in mitigating losses caused by a breach or in otherwise dealing with the consequences of a breach. An early example is Richard Holden Ltd v Bostock & Co Ltd (1902) 18 TLR 317 where the claimant recovered the cost of issuing an advertisement notifying the public that it had changed brewing materials when it discovered that the defendant had previously supplied it with sugar containing arsenic!
It would also seem fairly straightforward that payments that a claimant is contractually obliged to make to third parties as a result of the defendant’s breach can be recoverable. An example from the cases is compensation paid by a buyer of goods to a sub-buyer where the seller’s cargo was late (Contigroup Companies Inc. v Glencore AG [2005] 1 Lloyd’s Rep 241).
It appears from the Centrica case that the courts are willing to extend that principle to ex gratia payments. It cannot be the case that contracting parties are to be allowed in effect a blank cheque book to use at their will at the expense of the party in breach in order to encourage their own customers not to leave them. It nevertheless appears that, so long as they have acted reasonably (ie met the normal threshold for mitigation), such losses will be recoverable. Suppliers wishing to alter that default position would be well advised to seek to do so explicitly in the contract by means of a specific exclusion of such payments.
The other ‘standard’ exclusions
Accenture had not merely excluded the generic categories of indirect and consequential loss but, in line with current practice, had listed specific categories of loss (such as loss of profits, contracts, business and revenues) that would be excluded whether they were held to be direct or indirect. Nevertheless, Accenture failed to convince the Court that the most significant losses claimed for by Centrica were captured by this fairly standard list of exclusions.
Accenture had, for example, argued that the claim for ex gratia payments would be excluded as it was incurred by Centrica in an attempt to mitigate customer ‘churn’ and was therefore covered by Accenture’s exclusion for ‘loss of business’. The Court pointed out that any claimant has a choice how it pleads its loss and this element of the claim was framed in reputational loss. On that basis, the Court held that, in principle, such loss did not fall within the exclusion for loss of business and was therefore recoverable.
The real issue was that the list of exclusions set out above was not tailored to the deal and therefore did not cover categories of loss that Accenture had assumed were excluded. Overpayments to wholesale gas distributors and compensation payments to customers are two perfectly foreseeable (and the Court held, natural) consequences that a gas supplier might suffer if its billing system fails to function properly. The contract did not specifically deal with them so Accenture was left attempting to argue that the losses fell within the loss of business exclusion.
A more tailored approach to drafting will often serve suppliers better than a standard list of exclusions. Parties in other words may wish to spend more time considering the potential categories of loss that could be suffered by a customer at the outset and list each one as either included or excluded from the supplier’s potential liability. A longer and more ‘deal-specific’ list of exclusions and ‘inclusions’ will leave the parties and the courts in no doubt as to what is and is not to be covered if things go wrong.
Aggregation of breaches
Another key question put to the Court of Appeal (perhaps one of less general application) was whether the consequences of individual breaches could be aggregated to produce a ‘Fundamental Defect’. The Court held that, in principle, they could be.
The contract provided for what would happen in the case of a breach of warranty depending on whether it was a ‘Material Defect’ (defined, in summary, as a breach which had or was likely to have a material adverse effect on Centrica’s business) or a ‘Fundamental Defect’ (defined, in summary, as a fundamental breach which causes a severe adverse effect on Centrica’s business).
The Court was not persuaded by Accenture’s contention that the parties would have expressly provided for the concept of aggregation if they intended it to apply. They had, indeed, done so elsewhere in the contract in the context of termination for material breach where the relevant clause stated that material breaches would include ‘the commission of a series of related or unrelated breaches of this Agreement which, taken together, constitute a material breach’. This argument was, perhaps surprisingly in the context of an agreement between two sophisticated parties, said not to take the debate very far.
Deal with aggregation explicitly
Those involved in drafting and interpreting contracts will be familiar with contractual wording that introduces the concept of aggregation, typically in the context of termination for material breach (where persistent minor breaches could, when taken together, constitute material breaches).
It is more unusual to see drafting that carves out aggregation but this case demonstrates that, where parties intend that two concepts (eg non-material and material breaches) should be mutually exclusive, this should be explicitly stated in the drafting. In this case, the Court’s findings were largely based on the particular drafting used in the contract but it demonstrates the risks of being silent on the issue.
Variations in language
This tip will be ‘old hat’ to many but the final reminder comes out of the Court’s finding that the words ‘…which causes [an effect]…’ and ‘…which has or is likely to have [an effect]…’ have different meanings. Where wording used in two clauses is intended to have an identical meaning, identical language should be used otherwise the courts are likely to assume, quite reasonably, that the parties intended the different use of language to create a different effect.
Andrew Horrocks leads the IT disputes practice at Barlow Lyde & Gilbert LLP and Simon Jones is an associate in the firm’s commercial and technology team.
