Standard Life Assurance Ltd v Gleeds (UK) (a firm) & Ors

Extrapolation from representative sampling is permissible in principle in professional negligence and construction claims. The claimant must still prove, on the balance of probabilities, that the sample supports the inference sought. The sample must be sufficiently representative, and the inference may be defeated by rebuttal evidence or by features showing that the alleged breaches or losses are not comparable. A global claim, without strict attribution of every loss to a specific breach, is also permissible in principle, subject to proof and the evidential difficulties of the particular case. Strike out or summary judgment is appropriate only where the necessary inference is one that no reasonable trial judge could draw.

Standard Life brought a professional negligence claim concerning a major building development against members of the design team. It claimed approximately £38.1 million for additional variations and contractor’s loss and expense. Approximately £25.2 million of that sum was based on extrapolation from a sample of about 122 variations and selected delay notices.

The fourth, fifth and sixth defendants applied under the Civil Procedure Rules 1998 to strike out the extrapolated claim or obtain summary judgment. They argued that the sample was unrepresentative, that the pleading did not identify the breaches or causation sufficiently, and that the claim reversed the burden of proof. The central issues were whether extrapolation and a global claim were legally permissible, and whether the pleaded inferences were so untenable that the claim should be disposed of before trial.

1. Applications refused. The defendants failed to show that the whole extrapolated claim disclosed no reasonable grounds, was an abuse of process, or had no real prospect of success. The pleading was difficult and untidy, but it was comprehensible and identified the case to be met.
2. The court rejected the submission that the claimant had reversed the burden of proof. An invitation to draw inferences from primary facts may place an evidential burden on the defendant, but the legal burden remains with the claimant. The claimant must prove breach and causation on the balance of probabilities.
3. Extrapolation from sampling is permissible in principle. It is not confined to cases involving a formally “systemic” breach. The decisive question is whether the sample and the surrounding evidence justify the inference that the same type of negligence caused loss outside the sample. A case involving repeated failures to read, report or control costs may in principle support extrapolation even if there was no operating system that reproduced an identical error.
4. Extrapolation is unsuitable where the alleged breach concerns matters outside a defendant’s remit, or where the differences between the sampled and extrapolated work make the proposed inference manifestly impermissible. Some pleaded examples attributing responsibility to SGA or Cundall for matters outside their responsibility were therefore unsustainable. The court could not, however, conclude at the interlocutory stage that every extrapolated item was bad.
5. A global claim is permissible in principle, even without strict attribution of every item of loss to a specific breach. The issue is whether the claimant can prove the resulting loss. The loss and expense claim was therefore not demurrable merely because some amounts were not directly linked to individual breaches.
6. Proportionality was a material consideration. A trial of thousands of individual variations would be impracticable and disproportionate. The court proposed further sampling, including samples selected by the defendants, fuller tabulation of breach, causation, quantum and apportionment, and a fresh composite pleading. The claimant was warned that new, unpleaded allegations could not be introduced at trial.

First instance decision. No appellate history was stated in the judgment.