The case of Ted Baker Plc & Anor v Axa Insurance UK Plc & Ors [2017] EWCA Civ 4097 could serve as a parable of modern litigation. The claimant won the first trial on this matter, establishing the defendant insurers were providing coverage (“the defendant insurers… lost on every issue”. However the claimants then lost the subsequent trial when they failed to establish that that had suffered any loss that the insurers were bound to insure.  The trial judge held that the claimants’ claim for damages was based on speculation and not evidence. The Court of Appeal upheld that finding.

This was an assumption of fundamental importance to the establishment and calculation of the loss allegedly suffered by TB. But it was only an assumption and the judge considered that it was too speculative to provide a viable foundation for TB’s case. At [154] he said that it was “not based on any real evidence at all – or at least, it is based on evidence which is extremely tenuous.”


The claimants brought an action against their insurers seeking payment following thefts of goods by one of the claimant’s employees. In order to establish liability under the policy the claimants had to establish that it had suffered loss of more than £5,000 for each and every loss

    1. The judge rejected TB’s claim in its entirety on the grounds that it failed to establish any loss recoverable under the policies. To reiterate, TB’s claims were for the loss of profits resulting from those of JON’s thefts of stock under the business interruption cover. It abandoned before trial claims for the value of the stolen stock. There was an excess of £5,000 under the policies for each and every loss. The insurers speculate that TB withdrew its claim for the loss of stock because it knew it could not show that stock with a value of more than £5,000 had been stolen on any occasion. Instead its claim was based on showing that each theft resulted in a loss of profits in excess of £5,000.
    2. It was an issue before the judge as to whether the excess fell to be treated as an exception to cover, so that the burden of proving that the loss was less than £5,000 fell on the insurers, or as part of the cover under the policy, so that the burden of proving loss in excess of £5,000 lay on TB. The judge decided in favour of the insurers on this issue and TB has not appealed against that ruling.
    3. TB sought to establish its case by expert accountancy evidence provided by Mrs Katherine Britten, to which the insurers responded with expert accountancy evidence from Mrs Catherine Rawlin and expert retail stock control and systems evidence from Mr Richard Emery.
    4. A huge volume of expert evidence was filed and relied on at trial. Each expert provided four reports and there were two joint statements. These reports extended to well over 1,000 pages. In addition, each expert was the subject of searching cross examination. The judge referred at [31] to “a vast mass of expert evidence which depends on certain assumptions and very complicated analyses which, in turn, depend on matters which are somewhat speculative”. At [141], he was driven to describe it “as something of a swamp” and commented that the difficulty in handling such evidence and identifying relevant issues was exacerbated by the fact that a large amount of this material “was a moving feast (if that is the right word) right up to – and even during – the trial”. Coming, as they do, from a judge with great experience of trials with highly complex issues and evidence, these are very serious criticisms of the way this evidence was presented to the court, and made a task that would be challenging in the best of circumstances significantly more difficult.
    5. These observations and descriptions of the evidence are material, too, to the application of the principles that govern the approach of appellate courts to challenges to findings of fact made by the trial judge. These principles and the authorities establishing them are too well-known to need lengthy discussion. Referring to the decisions of the Supreme Court in McGraddie v McGraddie [2013] UKSC 58[2013] 1 WLR 2477 and Henderson v Foxworth Investments Ltd [2014] UKSC 41[2014] 1 WLR 2600, Henderson LJ recently commented in Hamid v Khalid [2017] EWCA Civ 201 that what emerges with clarity is that an appellate court should not interfere with a trial judge’s conclusions on primary facts unless satisfied that the judge was “plainly wrong”. In Henderson, Lord Reed said [62] that “[w]hat matters is whether the decision under appeal is one that no reasonable judge could have reached. At [67] Lord Reed said:
“It follows that, in the absence of some other identifiable error, such as (without attempting an exhaustive account) a material error of law, or the making of a critical finding of fact which has no basis in the evidence, or a demonstrable misunderstanding of relevant evidence, or a demonstrable failure to consider relevant evidence, an appellate court will interfere with the findings of fact made by a trial judge only if it is satisfied that his decision cannot reasonably be explained or justified.”
    1. On the issue of quantum, TB relies on two grounds of appeal. The first (Ground 3) is that the judge was wrong to conclude that TB had failed to demonstrate that on each occasion of theft, stock with a value of more than £5,000 was stolen, so that it could not demonstrate that its loss of profit resulting from each theft exceeded the policy excess of £5,000. The second (Ground 4) is that the judge was wrong to reject TB’s case as to the gross profits lost as a result of the thefts or, alternatively, should have accepted the figure advanced by the insurers’ expert witness.
    2. Ground 4 was added only after the hearing of TB’s renewed application for permission to appeal. Underhill LJ considered that the judge had rejected TB’s case on the loss of gross profits in its totality and had not accepted that TB had suffered any loss of gross profits, and that therefore Ground 3 on its own would not assist TB. Plainly, if TB could not establish that it had sustained any loss of gross profits, it could not show losses of gross profits in excess of £5,000 as a result of any theft.
    3. TB did not accept that this was a correct reading of the judgment, and so Ground 4 is prefaced by a statement that it is without prejudice to TB’s contention that the judge did not make any finding that no loss of gross profit had been established. In my view, TB is right about this. Underhill LJ identified paragraphs [159] – [164] as containing the finding that TB had failed to prove that it had suffered any loss of gross profit at all. The judge states his conclusion arising from those paragraphs at [165] where he referred to his “overall conclusion that TB has failed to establish even on the basis of the broad approach any lost profit above the excess” (my emphasis). That this accurately states his conclusion and that he did not find that TB had failed to prove any loss of gross profits at all is made clear in [175] where he referred to “the facts that….(ii) there is no doubt that TB suffered substantial losses arising out of the thefts…(iii) such losses amounted, even on the defendants’ experts’ evidence, to some £2.16 million”.
    4. Properly understood, the judge found that TB had not proved any loss of profits in excess of £5,000 as a result of any of JON’s thefts, and he had two discrete grounds for this finding. The first is that covered by Ground 3, that the judge did not accept the assumptions made by TB’s expert as to the number of boxes and items stolen on each occasion. This was referred to by the parties and the judge as JON’s modus operandi. The second is that he rejected the assumptions as to the rate of lost sales that underlay Mrs Britten’s model. This is covered in his judgment at [159] – [172] and I will proceed on the basis that this is covered by Ground 4.
    5. It is accepted by TB that if it fails on Ground 4, it cannot succeed on this appeal. I propose therefore to deal first with Ground 4.
    6. At [135] the judge noted Mr Cogley’s submission on behalf of TB that the fact that an insured peril and consequential loss may be difficult to demonstrate by direct evidence does not preclude the Court from finding that both have occurred. Mr Cogley relied in particular on Equitas v R & QReinsurance Co (UK) [2009] EWHC 2787 (Comm); [2010] 1 Lloyd’s IRLR 600 and Municipal Mutual Insurance Ltd v SEA Insurance Co Ltd [1998] Lloyds Rep IR 421 to establish, and the judge accepted, that modelling and drawing inferences can be used both to establish that the insured event occurred and the quantum of loss. This is not disputed by the insurers. At [138] the judge, while acknowledging the role of this type of evidence, noted that “ultimately, in order for a claimant to succeed, the Court must be satisfied on a balance of probabilities that one or more events have occurred as a result of one or more insured perils; “nothing less will do”. This is not disputed by TB.
    7. At [140] the judge held that once one or more events caused by an insured peril is shown and an actionable head of loss established, the Court will generally assess damages as best it can by reference to the materials available to it. Eder J accepted that (i) the balance of probability test is not an appropriate yardstick to measure loss; and (ii) lack of precision in relation to quantum is not a bar to recovery. The judge referred to this approach as the broad approach.
    8. The basic but critical aspect of TB’s claim for the loss of gross profits resulting from the theft of an item is that TB loses no profit until the time at which the item would have been sold, and its loss of profit is calculated by reference to the price at which the item would then have been sold.
    9. At [159] the judge identified correctly that the biggest issue concerning loss of gross profit is what has been referred to throughout the proceedings as the “timing of unfulfilled demand”. The “timing of unfulfilled demand” describes the point at which a sale of an item is lost as a result of the theft of the item. As Eder J set out:
“Thus if an item were stolen on (say) 1 January 2008, the potential profit lost on that particular item would have to be calculated, in theory at least, by reference to some hypothetical potential sale not on that date (1 January 2008) but on some future date. Inevitably, such an exercise would need to seek to take account of matters which would (or would not) have occurred but for the theft including (i) how long that item would have remained at the TBDC; (ii) when it would have been sent out to a particular retail location and, at that time, to which location; (iii) if and when it would have been sold; and (iv) if sold (as opposed to disposed of by way of scrappage) at what price”.
    1. TB’s case was based on the modelling exercise carried out by their expert witness, Mrs Kathryn Britten. At [162] the judge said:
“As to the validity and general reliability of such exercise, the central issue was the methodology adopted by Mrs Britten in assessing the timing of unfulfilled demand and, in particular, her use of what was described as a “blended gross margin”. In calculating lost profit, Mr Cogley accepted, indeed emphasised, that the starting point was that a lost sale on a particular item which had been stolen might occur at any time and he also readily accepted that the calculation of the lost profit on that item and he also readily accepted that the calculation of the lost profit on that item can present what he described as a “problem”. However, he submitted that the claim in the present case was not for the lost profit of any particular item stolen but for the lost profit on a vast number of stolen items; and that in that context, the use of a “blended gross margin” was both justified and indeed more reliable than what he described as the more “granular” approach adopted by Mrs Rawlin and Mr Emery.”
    1. The blended gross margin approach started by taking the average sales figures for TB, which produced an average margin. Certain adjustments were then made to those average sales figures to reflect that, compared to actual sales, there was a lower probability that the stolen items would have been sold in the early weeks of the sales period for any season. At [163] the judge noted how this adjustment was made by Mrs Britten:
“(a) to exclude any sales made prior to the start of the season -because there are unlikely to be lost sales before the start of the main selling season/de minimis; (b) to include only 35.65% of the sales made in the next four weeks; (c) to include only 87.5% of the sales made in the next four weeks; and (d) to include all sales from that point….Mr Cogley (and Mrs Britten) recognised that this only produced a blended gross margin for a single season; and that, since the seasons and the policy years did not run together, in order to establish the appropriate loss of gross profit in any given policy year, it was necessary to take into account the figures for more than one season and make further adjustments”.
    1. At [164] the judge said in relation to the modelling exercise of Mrs Britten:
“…even applying the broad approach stated above, I remain unpersuaded that it provides a sufficiently reliable or reasoned method for assessing the appropriate loss of profit in the present case. In particular, as submitted by Mr Nicholson, it seems to me that (apart from some “dilution” in the early part of each season), it assumes, wrongly in my view, that the (hypothetical) sales profile of the stolen items would have followed the profile of actual sales. By way of a number of homely examples, Mr Cogley vigorously advocated that such criticism was fundamentally flawed; and Mrs Britten herself emphasised that she had applied the “full range of discounts”. However, as submitted by Mr Nicholson, although the latter was true in a sense, I remain unpersuaded that the stolen items would have followed the path of average items. In response, Mrs Britten sought to justify the approach that she had adopted on various different bases; and the various adjustments she had made. However, at the end of the day, I remained unpersuaded (even applying the broad approach) that the explanations proffered by her satisfactorily addressed what seems to me the main flaw in her modelling exercise.”
    1. Eder J then went on at [165] to state that:
“…I do not necessarily accept the accuracy or reliability of the exercises carried out by Mr Emery and Mrs Rawlin. In that regard, Mr Cogley advanced a number of criticisms with regard to what he described as their “granular” approach. I recognise the potential force of at least certain of those criticisms. However, it is unnecessary for me to consider them in detail because, even if certain of those criticisms are justified in whole or in part, they do not affect my overall conclusion viz that TB has failed to establish even on the basis of the broad approach any lost profit above the excess”
    1. Before Eder J there was an argument advanced by the insurers that the model employed by Mrs Britten was not contractually compliant with the adjustment provision for calculating loss. Mr Cogley in his written submissions advanced arguments that Mrs Britten’s methodology is contractually compliant. That aspect of the insurers’ case is not pursued by them before us.
    2. Mr Cogley submitted before us that the judge’s rejection of Mrs Britten’s evidence is flawed because it does not recognise the similarity between her evidence and the evidence of the insurers’ expert Mrs Rawlin. Mr Cogley submitted that the reason for the discrepancy in the figures which each expert produced is that Mrs Rawlin’s sale of the lost stock would have occurred later in the season (being the timing of unfulfilled demand).
    3. Mr Nicholson for the insurers pointed out that in her first report Mrs Britten did not consider the unfulfilled demand threshold. It was only after one of the insurers’ experts, Mr Emery, said that there would be a lag before a depletion in the warehouse resulted in a lost sale that Mrs Britten made adjustments. Her adjustments took Mr Emery’s rate at which a particular item ran out in the last few weeks of a season and applied it to the first few weeks of the season. Mr Nicholson referred to Mr Emery’s criticism of this approach in his fourth report dated 7th July 2014 at 5.3.5-6:
“The purpose of my analysis was to establish the rate of diminishing SKU availability during the last 8 weeks of the Retail season. Mrs Britten has used it to establish the rate of Diminishing SKU Availability for the first 8 weeks of the season. It is clear to me that this is not an appropriate use of my analysis because the rate of sale products in the first few weeks of the season is much slower than in the last few weeks”
    1. This was put to Mrs Britten in cross examination:

“Q. Do you agree with that as well?

A. Yes I do.

Q. Well, that is a nonsense, isn’t it, Mrs Britten, that you have done that, because you’re looking at the wrong end of the season? The reason why that doesn’t work is as Mr Emery explains at 5.3.6?

A. It would be a nonsense if the result was significantly different, but in fact I tested the percentage that I’d used by carrying out the same analysis, but only looking at last dates following the TBDC going out of stock in the first two SUN periods, and the percentages that I arrived at were very, very similar. So I decided that, having adopted Mr Emery’s numbers before I saw his criticism, I had no need to change them.

Q. Do we find any reference to that testing in your 25 June report?

A. No, because I’d written the report before I saw this.

Q. Do we find your test or analysis in any material put before the court?

A. No, because I haven’t changed my calculations.

Q. Mrs Britten, the point is it’s just fundamentally inappropriate to use an analysis that one expert has carried out for one purpose and then apply it backwards the wrong way round for an entirely different purpose; that’s correct, isn’t it?

A. Put like that, yes.”

  1. At [164] of the judgment Eder J made findings of fact when he said that he was not persuaded that Mrs Britten’s methodology “provides a sufficiently reliable or reasoned method for assessing the appropriate loss of profit in the present case” and that “it assumes wrongly in my view, that the hypothetical sales profile of the stolen items would have followed the profile of actual sales”.
  2. Mr Cogley’s submissions on Ground 4 are essentially an attempt to re-argue TB’s case. In my judgment, he cannot establish, as he must, that the judge was plainly wrong in his assessment and rejection of Mrs Britten’s modelling exercise. The rejection was not because of any objection to modelling per se as a means of establishing loss but because the judge did not accept Mrs Britten’s approach to the timing of unfulfilled demand, which Mr Cogley accepted before us was the nub of this part of the case.
  3. As earlier noted, Eder J was provided with voluminous expert evidence, heard extensive cross examination of the expert witnesses and received very full written and oral submissions. All of this he examined in detail in the preparation of his careful judgment. In reviewing his decision, it is not for this court to embark on the same exercise and it is precisely because of the nature of the trial judge’s fact-finding exercise that this court will interfere only if he has neglected evidence, which is not suggested here, or has gone plainly wrong.
  4. As I earlier observed and he made clear, Eder J was not well served by the way in which the expert evidence was presented to him. Nor were we. In giving permission to appeal, Underhill LJ underscored the necessity of a clear presentation of the issues on quantum and the relevant expert evidence. In his decision granting permission to appeal, he stated that “the quantum issues will require the Court to be given a thorough understanding of the approaches taken by the parties and their experts” and he went to emphasise “that it is particularly important that the skeleton arguments give a clear, succinct [his emphasis] and well-analysed guide to the quantum evidence and to how their respective cases on the quantum issues were put to Eder J, supported by full cross-referencing”. On this basis, Underhill LJ did not think it necessary to increase the estimate of two days, of which one was available for the quantum issues. I am grateful to Mr Nicholson and his team for the analysis annexed to their skeleton argument for the appeal. But the greater onus was on TB which needed to explain the issues and clearly identify the ways in which the judge was plainly wrong. The benefit of an oral hearing is that the essential issues can be identified and examined, albeit with only limited available time. Having done so, I am for the reasons given above satisfied that TB cannot succeed on its appeal under Ground 4.
  5. Ground 3 relates to the other basis on which the judge found that TB had not established any loss of profits in excess of £5,000 as a result of each and every theft.
  6. In order to establish recoverable loss, TB assumed that the amount of stock stolen by JON on each occasion was 3.5 boxes, with each box containing 68.25 items, meaning an average of 238 items per incident. The calculation of the profits lost as a result of each theft was calculated on this basis. The figures chosen were an extrapolation of the quantity of stock stolen by JON on the last five or six thefts before his arrest, of which there was some evidence.
  7. This was an assumption of fundamental importance to the establishment and calculation of the loss allegedly suffered by TB. But it was only an assumption and the judge considered that it was too speculative to provide a viable foundation for TB’s case. At [154] he said that it was “not based on any real evidence at all – or at least, it is based on evidence which is extremely tenuous.” While agreeing that it was “a possible average” (the judge’s emphasis), “it seems to me difficult, if not impossible, to say that other figures are not equally possible and, indeed, reasonable and realistic”.
  8. The judge observed that even slight changes to the input assumptions produce quite significant changes to the number of incidents and, ultimately, the amounts recoverable after applying the £5,000 excess to each and every incident. He proceeded to illustrate this by some examples, “which would dramatically reduce the total amount recoverable and in certain policy periods extinguish any claim completely” and some other examples that would have the opposite effect. He concluded that “[o]n the evidence available and within a relatively broad range, it seems to me quite impossible to say that one set of assumptions is more or less reasonable than any other set of assumptions even adopting the broad approach” [i.e. to establishing the quantum of a claim].
  9. The judge considered that figures for the last few thefts carried out by JON did not provide a sound basis for extending a similar average backwards over a period of four years. Nor did it help to assume, as the judge accepted was likely, that JON started small but grew greedy. If anything, it suggests that JON would steal a smaller number of items on each occasion. While the judge was also prepared to accept that JON would try to minimise the risk of detection, “whether he would do this by minimising the number of incidents of theft or the number of boxes/items stolen on each incident of theft is, in my view, entirely speculative”.
  10. For these reasons, the judge at [158] concluded that “the evidence fails to show that the claimed losses as calculated by TB fall above the excess of £5,000 each and every loss”.
  11. In his submissions, Mr Cogley focussed on the last year or so of JON’s thefts. He emphasised, correctly, that JON could not probably have executed more than one theft per day and that, on that basis, some of the judge’s hypothetical examples were not feasible for that period. Mr Cogley accepted that there would have been variations, with the amount of stock stolen on some occasions being incapable of resulting in lost profits in excess of £5,000. But, he submitted, that was the whole point of the modelling of an average loss. It was not speculation, but the only way realistically to account for the known facts of the case.
  12. In my judgment, these submissions do not provide a sound basis for overturning the judge’s finding that TB had not, and indeed probably could not, establish the loss it claimed. The assumed volume of stolen items for each theft was crucial for establishing the quantum of TB’s claim but it was highly speculative. For understandable reasons, Mr Cogley focussed his attention on the last year or so, because the scale of the thefts had grown to a point that, when combined with the maximum number of probable incidents of theft (one per each of JON’s working days), enough items may well have been stolen on a substantial number of occasions to create the possibility of lost profits in excess of £5,000. But that is very far from being enough to establish a basis for TB’s claim as pleaded and presented to the judge.
  13. I conclude that the judge was entitled to form the view that this fundamental element in the establishment and calculation of TB’s claim lacked the substance needed to prove its case on the balance of probabilities.
  14. For these reasons, I reject the challenge to the judge’s decision that TB had failed to establish the loss it claimed on both of Grounds 3 and 4. I would therefore dismiss this appeal.