Mulsanne Insurance Company Limited v Marshmallow Financial Services Limited

This is a judgment given after trial of liability in relation to a dispute between the claimants (an insurance company - “Mulsanne”) and its former broker/intermediary (the first defendant - “Marshmallow”) and the latter’s related insurance company (the second defendant - “Marshmallow Insurance”) about the alleged misuse of confidential information said to have been supplied to Marshmallow by Mulsanne during the course of their relationship, and the circumstances in which the relationship ended. An outline of the circumstances and nature of the dispute is as follows - an outline will help to navigate the complex facts which will have to be recited later.

In what follows it will occasionally be necessary to refer to the two defendants at the same time. Where that happens I shall, for the sake of convenience, call them compositely “Marshmallow”. It will be obvious where that expression is used in that way as opposed to its being used to describe Marshmallow the first defendant.

In 2017 Dr Tim Holliday and others formed Marshmallow as a broker for motor insurance policies, intending to deploy technology to enhance the customer’s experience and enhance the profitability of the insurance product it sold. It developed a ratings engine (a series of tables in a spreadsheet coupled with some underwriting rules) which it considered would achieve those objectives (when properly implemented) and hoped to engage an insurer called Hiscox UK as the underwriter who would provide the “capacity” - underwriting for the policies which Marshmallow would sell. Unfortunately Hiscox pulled out at a relatively late stage and Marshmallow had to find another underwriter to provide capacity. Contact was made with Mulsanne and they both agreed that they could work together to target in particular the migrant market, a sector of the market which they considered to be under-served by the rest of the market and which would yield, or contribute to, a profitable enterprise for both of them. Hitherto Mulsanne had specialised in the higher premium section of the market - for example, drivers with disqualifications or expensive cars. The parties started business together in summer 2018, when Mulsanne policies were first sold by Marshmallow (as intermediary with authority to enter into policies). The ratings engine they used was based on the engine prepared for Hiscox (“the Hiscox engine” or “the Hiscox tables”), but with some modifications, some of which were contributed in whole or in part by Mulsanne. The parties continued to develop and refine the ratings engine over time.

The relationship between the parties was governed by a written Terms of Business Agreement signed on 26th April 2018 but with an effective operating date of 11th May 2018 (“TOBA”).

During 2019 Dr Holliday and others in Marshmallow decided that they would like to form their own insurance company to sell motor policies, and they set about forming one. This was not disclosed to Mulsanne at that time or for some considerable period thereafter. It was not envisaged that that the new insurance company would simply replace Mulsanne, but it does not seem that clear views had been formed as to how they would co-exist commercially.

Marshmallow was getting ready to launch in autumn 2020, and in September a meeting took place at which Mulsanne was informed of its intentions. Mulsanne’s initial reaction was not hostile, but very soon its attitude changed and Mulsanne expressed its hostility to the idea. Mulsanne decided that it wished to terminate the TOBA and sought to do so in February 2021. A dispute developed between Marshmallow and Mulsanne as to whether and when the TOBA was terminated, and its consequences, and in due course a further dispute developed about the alleged misuse of confidential information by Marshmallow in starting up, and trading with, Marshmallow Insurance. The confidential information averred by Mulsanne is said to be material information provided by Mulsanne from time to time for feeding into the ratings engine used for the purposes of entering into Mulsanne policies, or underwriting rules. In relation to some of that information it is said that it was actually used directly (i.e. in Marshmallow Insurance’s own engine); in relation to other parts it is no longer said it is being used, but it is said that it was used as a “springboard” to enable Marshmallow Insurance’s insurance business to get under way quicker and/or more effectively than would otherwise have been the case. Some of the information is also said to have been used in the course of getting investor buy-in, regulatory approval and re-insurance. There is also a claim for passing off, claiming that in persuading customers to renew their expiring Mulsanne policies into Marshmallow Insurance as they expired, Marshmallow passed off Marshmallow Insurance’s policy as Mulsanne’s.

Before and then during the course of the trial various parts of the confidence claim were abandoned by Mulsanne. This judgment deals only with what was left by the time of final submissions, and is based on the assumption that the remaining claims are those identified in the written final submissions of Mulsanne. That is the only way of dealing with what would otherwise be a slightly bewildering series of abandonments.

Mr Martin Howe QC led for the claimant on confidentiality and passing off matters; Mr Aidan Christie QC led for the claimant on insurance matters; Mr Jonathan Hough QC led for the defendants on all matters and I also received submissions from Mr Jonathan Moss for the defendants on passing off.

It is necessary to set out a certain amount of procedural history because it is a necessary background to how parts of the case were run and to alleged limits on the case that it is open to Mulsanne to run now.

These proceedings were started by Mulsanne on 11th March 2021. The claim includes injunctive relief restraining the use of confidential information, passing off, the carrying out of an audit, the delivery up of papers and other relief. It also claims damages. When the proceedings were commenced Mulsanne applied for extensive interim relief restraining the use of the allegedly confidential information. That substantive interim dispute was ordered by Marcus Smith J to be heard as an application by order on 23rd-24th June, and directions were made for the service of evidence. However, on 21st May, when the claimants were due to serve evidence in reply, Mulsanne withdrew its application for an interim injunction.

Meanwhile Marcus Smith J had also ordered a split trial of liability and quantum, and expedition of the trial of the former so as to take place as soon as possible after 1st October 2021 (order dated 22nd April 2021). Directions were given to that end, including a direction that disclosure be given by reference to requests from each side rather than the implementation of the disclosure pilot scheme. In due course the trial was set to start on 17th November 2021 (and thus it came before me). The witness statement regime which was ordered was one which allowed the interim application statements to stand and to be supplemented by further witness statements; and the same technique seems to have been adopted in relation to experts’ reports filed in the interim application. The result is a string of evidence which has at times been hard to navigate, though it doubtless seemed like (and was) a good idea at the time.

The original claim claimed confidentiality in, and infringement of, various pieces of “Confidential Information” (a defined term) identified in a confidential schedule to the Particulars of Claim. That schedule listed 13 pieces of information by reference to identified documents and the dates on which they were provided. The documents were annexed to the Particulars of Claim in further schedules. On 23rd June 2021 Mulsanne applied for further disclosure before Marcus Smith J, most of the documents in question relating to information which was put before regulators, investors and reinsurers. Marshmallow resisted that application on the basis (inter alia) that the disclosure did not relate to a pleaded issue. Marcus Smith J in his judgment ([2021] EWHC 1838 (Ch)) indicated that he thought that that was probably right, though not certainly right and he refused the application on the basis that it was too wide and ill-focused and involved “lifting the lid on the entire business plan of the Defendants from essentially when it began to when it got onto its feet as a business operation”. He did, however, require the defendants to file a “business development timeline” of key business development milestones, because he considered that that was (or might be) relevant to the issue of whether the Confidential Information relied on was put to investors, regulators or reinsurers in some way. His judgment indicates that he suggested that documents would have to be produced in relation to that, but his order (and his judgment) did not specifically provide for that.

On 6th September 2021 Marcus Smith J refused an application by Mulsanne for permission to amend the Particulars of Claim to add a new confidentiality claim based on the loss ratio and claims information relating to the Mulsanne policies sold by Marshmallow ([2021] EWHC 2478 (Ch)). This application was one which sought to add a new and (as the judge held) very significant category of information and infringement to the claim. He refused the application because it was an extensive one which required very significant new evidence and disclosure (principally from the defendants) which could not be achieved by the trial date. He attributed considerable importance to the fact that this sort of claim was articulated in pre-action correspondence from Mulsanne’s solicitors (letter of 22nd October 2020) but then not pursued, and not included in the originally pleaded claim when, if it was a good one, it could have been.

That was not an end of the amendment applications. At the pre-trial review (conducted by me) on 29th October 2021 Mulsanne made a further application to amend the schedule of Confidential Information in the Particulars of Claim to add a reference to the state of the ratings engine as it stood at November 2021. Considering such a claim would have involved going into alterations to the engine which were not encapsulated within the already pleaded pieces of Confidential Information, investigating how they came about, who introduced them, whether they were confidential and whether they were used thereafter in Marshmallow’s business. I refused that application on the footing (in short) that it was too late to add such a significant amendment, requiring very significant work, bearing in mind the proximity of the trial and the already complex and extensive nature of the claim.

I heard oral evidence from the following witnesses. All the witnesses that I heard had undergone a process of (proper, not improper) witness training. That was more obvious in the case of some witnesses than others, but nothing in their respective credibilities turns on that. It probably contributed to the care and consideration which they gave to their answers in cross-examination.

Mr McCauley is the chief executive of Mulsanne. He has a degree in economics (1992) and has had his professional life in various insurance companies in various capacities. He joined Mulsanne as CEO in July 2017. He has considerable experience in underwriting, and no-one challenged his great expertise in the area. He was Mulsanne’s only non-expert witness and gave evidence about Mulsanne’s business activities as a motor insurer, the nature of the confidential information involved in this case and dealings with Marshmallow (with which he was intimately involved). His credibility was not significantly challenged. He was a very careful witness (as is best demonstrated by his indication one morning that he had read the previous day’s transcript and had a correction to make to his previous day’s evidence), and gave carefully measured answers. He was, in the most part, not evasive, but he did become unjustifiably defensive when cross-examined on the dealings between the parties (through their lawyers) when the dispute first blew up when there were discussions about Mulsanne taking back dealing with Mulsanne policyholders.

Mr Rourke is a director of Willis Towers Watson, a consultancy firm with (inter alia) insurance expertise. He has worked within the insurance industry (predominantly pricing and underwriting) for over 18 years, and clearly has significant expertise in the field. He gave evidence going to alleged copying of parts or the whole of ratings engines in the case and the likely dealings between a would-be insurer and investors, regulators and reinsurers. He was plainly a careful witness who could not be accused of being a hired hand for those who engaged him as an expert witness.

Mr Alexander Kent-Braham is the co-founder and co-CEO of Marshmallow Technology Ltd which is the ultimate holding company of the two defendants. He has a twin brother Oliver who is also a co-founder and co-CEO, and he also gave evidence. In order to distinguish between them I will call them AKB and OKB respectively.

AKB and OKB decided to found Marshmallow and were jointly responsible for the initial steps to that end. They approached Mulsanne at the outset of their relationship, and thereafter AKB was responsible for various business aspects of Marshmallow and was involved in the arrangements to set up Marshmallow Insurance as an alternative capacity provider. He was not involved in the detail of the development of ratings table. Again, he was a careful witness who obviously thought about his answers. He was a witness of truth.

As appears above, he is the twin brother of AKB and has a history of involvement with him in business projects. His evidence in chief was about the “milestones” of getting the Marshmallow Insurance project to fruition in terms of engagement with investors, on which he was cross-examined. He gave clear and careful answers and I consider I can accept their accuracy.

Dr Holliday is the CEO of Marshmallow Insurance and until the beginning of December 2020 was the non-executive chairman and interim Head of Insurance of Marshmallow. He was a driving force behind the proposals of Marshmallow on which Mulsanne engaged. He has over 20 years of experience as an insurance underwriter with a variety of substantial insurance organisations, and 10 years experience in the personal lines motor business. He was very significantly involved in underwriting and business decisions leading to the Mulsanne/Marshmallow relationship, and a directing hand behind various steps taken in setting up Marshmallow Insurance’s business. He was plainly (and contrary to suggestions made by Mulsanne) an experienced underwriter capable of making reasonable judgments about underwriting risk and a reliable witness.

Ms Coulthard is now the Insurance Director of Marshmallow Insurance, and has been since 1st January 2021. From 1st October 2020 to 31st December 2020 she acted as a consultant to Marshmallow Insurance pending that company’s getting a licence to operate from its Gibraltar regulator, the Gibraltar Financial Services Commission (“GFSC”). In that latter capacity she was instrumental in what is said by the defendants to the creation of Marshmallow Insurance’s underwriting rules and ratings tables. She was a careful witness on whose evidence I can rely.

Mr Hemming acted as a consultant to Marshmallow on various occasions, on each occasion acting through his consultancy company Hanoverian Interim Management Ltd, in relation to the creation of a ratings engine or parts of an engine such as postcode tables for a ratings engine (an insured’s postcode being a factor in assessing risk). He gave short unchallenged evidence mainly about aspects of that activity. His skill (for present purposes) lay in building components for underwriting pricing structures, and he created sophisticated models to that end. His accepted skill in doing so is demonstrated by the fact that from time to time both Mulsanne and Marshmallow engaged him for their own purposes. He was a reliable witness whose credibility was not challenged.

Mr Hillon is a director of KPMG and an actuary with over 23 years of experience in the general insurance market. He has experience in the development of pricing models for motor and home insurance, and gave evidence for the defendants on the same areas as Mr Rourke. I consider that he discharged his functions conscientiously.

The facts found in this section are found on the basis of various parts of the evidence given to me. They are uncontroversial, and are necessary to understand the nature of the confidential information which is said to have been misused, and various aspects of the insurance market.

Motor insurance policies, like all policies, are priced according to risk. Any insurer setting out in the market considers the impact of various risk factors on premium. The confidentiality aspect of this case are mostly about the alleged misuse of some of the work product used to assess risks and price policies.

The techniques relevant to this case (and common across the industry) for the acceptance and calculation of risk and for pricing premiums involve setting out underwriting rules and then developing a rating model comprising tables of information which appropriate risk to relevant factors. In order to assess risk an insurer will consider various factors which go to make up a risk picture. In the case of motor insurance those factors include (to take basic examples) the make of car, the value of a car, the age of the driver and that driver’s claims experience. A more sophisticated example is postcode areas – some apparently reflect more risk than others. Other matters will appear in the narrative of this case appearing below.

In order to work out a relevant premium for any given insurance applicant an insurer will start from a base premium, assessed by the insurer. To that premium the insurer will apply a series of successive multipliers (“relativities”, a central feature of this case), each of which will reflect the insurer’s views of the risk involved in that particular factor. Thus, for example, an insurer will have a multiplier for each of a given make, type and characteristic of the car involved. There will be a multiplier for the age of the driver. There is usually (as referred to above) a multiplier for the postcode of the applicant. And there are likely to be many more. By applying the various multipliers (relativities) successively to the base premium an insurer can arrive at a premium which is to be charged. If that is a slight simplification of the process it will nonetheless do for the purposes of this judgment.

There are also various underwriting rules. Some of them may be reflected in relativities, but some are simple cut-outs. For example, an insurer is likely to have an underwriting rule as to a driver’s age below which insurance will not be given.

All these materials are combined into a ratings model which reflects the mathematical calculations and underwriting rules which are to be applied. It is made up of various tables (ratings tables). The tables are usually created in a form of programming language called R. They are also presented in spreadsheets for people to consider, because spreadsheets are more intelligible to the human mind. It is in that form that information (including the relevant alleged confidential information) was exchanged between the parties and presented to me. Underwriting rules are either considered in spreadsheet form or in a more standard document form.

When the pattern of underwriting rules and relativities has been established they are further coded into a computer programme which is capable of taking the input from an applicant and working out and presenting a premium, or (if appropriate) a rejection of cover. In the modern world an applicant for insurance is, or can be, presented with an on-screen website in which an electronic form is filled in. The underlying software then takes that information, applies the rules of the engine and presents a price which, if accepted, begins the process for obtaining that policy of insurance. When the Marshmallow/Mulsanne enterprise commenced the website was Marshmallow’s own, but after a few months Marshmallow joined an “aggregator” or “price comparison website” (“PCW”). Those are websites at which an applicant can fill in his/her details, which are fed into the ratings engines of all the subscribing insurers, and each presents the applicant with a price if that insurer wishes to offer cover (if an insurer’s engine declines cover for that applicant then its price is not displayed). Thus an applicant can get a number of quotes online.

The appearance on the scene of the aggregators has introduced a substantial measure of competition and transparency into the market. It has also provided a source of market information to insurers which they can use to price their products. The aggregators provide the insurance market with various details (anonymised) of applicants and the quotations provided (where they are provided) from which an insurer can assess the competitiveness of its own rules and product and draw certain conclusions as to the risk appetite and risk assessment of other insurers. Its own tables can be adjusted if it wishes to do so, to reflect another view of risk and/or to present a competitive price. The details of that do not matter for present purposes. What is relevant is the fact that this market data exists and it can assist insurers in determining their own ratings, and therefore prices and profitability assessments. This form of data plays a part in the story of this case.

It is some ratings tables and underwriting rules passing between the parties during the course of their relationship that are said to be the Confidential Information which has been misused by Marshmallow Insurance once it was trading.

In all this narrative, as in all subsequent narratives in this judgment, any narration of fact should be taken as a finding by me unless the contrary appears. This section sets out the overall course of the relationship between the parties which led to the present dispute, but it does not contain the details of the Confidential Information which is said to have been abused, or the other claims that are made. It is intended to act as a factual framework into which those details can be notionally inserted as and when I come to deal with them.

Mulsanne is an insurance company established in 2010. It is regulated by the GFSC in Gibraltar and is authorised to underwrite, inter alia, policies of motor insurance. In 2020 it wrote approximately 72,000 “vehicle years”, that is to say policies insuring a vehicle/driver for one year, with a premium income of around £70.9m. In financial terms it is not among the largest players in the industry. It operates at the high premium (and therefore unusual risk) end of the market, offering insurance to policy holders whom “standard” underwriters may not be interested in insuring - for example, drivers with significant claims or conviction histories. Like many insurers it does not offer its product directly, but operates through brokers (intermediaries) to whom it delegates authority to issue policies and who therefore acts as its agents. Marshmallow became one of those intermediaries. If a claim is made under one of its motor policies Mulsanne delegates its handling to another third party, namely Key Claims Ltd, another company in Mulsanne’s group. It has its own intermediary in the group, Complete Cover Group Ltd (“CCG”).

AKB and OKB acquired an interest in technology and finance and by 2017 had acquired an interest in insurance. They realised that the migrant market could offer a useful niche motor insurance market which was not well catered for by traditional insurers. In February 2017 OKB contacted Dr Holliday and shortly afterwards AKB and OKB pitched their idea to him. Dr Holliday agreed that the migrant sector was an untapped one and they took their idea forward. They decided to form a broker rather than an insurer because that required less capital, though their ultimate aim was to set up an insurer. An existing company of the Kent-Braham brothers had its name changed to Marshmallow Technology Ltd, which ultimately became the holding company for both Marshmallow (the broker) and Marshmallow Insurance. Dr Holliday provided the underwriting and insurance expertise required for these purposes.

In order to be able to approach an insurer with the implementation of their idea the then participants engaged Mr Hemming to produce a ratings engine. This (as described above) contained a number of tables containing the factors to be used when pricing a policy, and relativities to be applied reflecting the risk variables within each table. They developed their own ratings engine because they believed that there was value in owning as much of their pricing as possible. Various substantial insurers were approached with the idea of creating a niche insurance product, with the proposal that Marshmallow would act as broker, but they were rejected. They then approached Hiscox UK, who agreed in principle to take the matter forward. The matter was to be taken forward on the basis of the adoption of the newly developed ratings table, (the Hiscox model or Hiscox engine). If the deal had gone ahead then this table, without modification, would have provided the basis on which Marshmallow acted as broker for Hiscox and wrote policies under a delegated authority.

The Hiscox model is important because it formed the basis of the approach to Mulsanne and because it was, with some modifications, the basis on which the business with Mulsanne was commenced. It is also said that parts of its content were carried over into the later model used by Marshmallow Insurance so that those parts cannot be confidential information of Mulsanne. It is important to note that in its original form it was the product of Marshmallow and had no Mulsanne input. Its contents, as they stood, form no part of the present Confidential Information. It was derived from using market data and was essentially the work of Mr Hemming and Dr Holliday, the latter providing underwriting input and judgments. In forming the tables making up the model Dr Holliday had in mind the principal target audience (migrants) which required adjustments to the applicable market datasets. Thus, for example, the model had a question asking where the applicant had first passed his/her driving test. That would not appear in the standard model. The exercise of the model and its constituent tables and rules was a complex and specialised one. The nature of the content of the tables will become clearer when I consider parts of tables which are said to have been misused Confidential Information in this case.

The model (or engine) was ready by November 2017 when it was presented to Hiscox. Hiscox was initially minded to enter into a broking agreement with Marshmallow, which would be based on the model without amendment, but in February 2018 it pulled out. Dr Holliday then looked for an alternative insurer and he and AKB met Mr McCauley of Mulsanne on 23rd April to explain their ideas. Dr Holliday already knew Mr McCauley because Dr Holliday had formerly been Mr McCauley’s superior when they both worked for Zurich insurance company. Mr McCauley was interested in the proposal (which still included migrants as a target market) and various matters were discussed.

After the meeting Mr McCauley emailed AKB on 23rd April 2018 encapsulating various things that were discussed. The email reads (so far as relevant):

The scheme outlined (and implemented) was that Marshmallow would act as brokers with delegated authority to enter into policies on behalf of Mulsanne. Marshmallow would promote it and account to Mulsanne for the premiums. Claims handling was to be done by Mulsanne’s preferred claims handlers (Key Claims), so Marshmallow would have no part in that, though it did provide the telephone number for claimants to ring initially in respect of claims. Marshmallow would not receive a commission from Mulsanne but would charge the customer a fee on top of the premium. Any ratings and rating changes, which would ultimately be reflected in the premium charged, would need to be agreed by Mulsanne (which was carrying the insurance risk) in advance.

Just 3 days later, on 26th April 2018, the parties signed the TOBA, with an effective date of 11th May 2018. I shall devote a separate section of this judgment to setting out relevant terms of the TOBA. Some of them are relied on by Mulsanne in the separate part of this action as giving rise to claims in damages arising out of what happened when the relationship turned sour.

Over the course of the next few weeks the two sides discussed and altered various aspects of the ratings tables. Mr McCauley, AKB, Mr Hemming and Dr Holliday were variously involved in this. The postcode and occupations tables were the subject of discussion and alterations were made, though whether or not a new postcode file was actually incorporated is a question which arose at the trial. Since particular claims are made in relation to those tables it will be convenient to describe those tables separately when I come to consider the detail of the claims made in this action.

A base premium was agreed in July 2018 between Dr Holliday and Mr McCauley. It was determined at a point which would make the anticipated loss ratio an acceptable one. The loss ratio is a ratio of gross premiums to claims, and there are variations of its method of calculation which do not matter for present purposes. Nor does the way in which the premium was set. The parties (and particularly Dr Holliday) were trying to come up with a base and models for a niche in which there was little specific market information so underwriting judgment was required in relation to various parts of the overall exercise.

Marshmallow started to sell Mulsanne policies in late July 2018. Over the course of the next year or so various additions to and amendments of the ratings engine were proposed and implemented. The detail of these matters, so far as relevant to the currently constituted claims in this action, will appear when I consider the details of each claim of misuse of Confidential Information.

In so acting Marshmallow was acting as agent of Mulsanne pursuant to the TOBA (which did not fully set out the agency in this respect, but the existence of the agency was common ground). Marshmallow gathered in the premiums and accounted to Mulsanne for them. No issue arises out of premium gathering and transmission. Marshmallow provided a daily report to Mulsanne of new policies entered into, with details of the policy, the policyholder, his/her contact details, the amount of the premium and other relevant matters. For its part Mulsanne from time to time provided Marshmallow with details of claims made, even though they were actually handled by Key Claims Ltd (in the Mulsanne group).

The alleged Confidential Information on which this claim is based can be broadly described as being Mulsanne’s input into some (but by no means all) the tables and rules in the ratings engine. It is said that this information was misused when that material was used (in various ways) in or for the purposes of the new Marshmallow-only business.

At first, policies were sold on what was effectively a Marshmallow website. However, after a short period Marshmallow felt sufficiently confident about its product that the Marshmallow/Mulsanne product was sold via one or more aggregators. Various amendments were made to the engine to make it more suitable for selling policies in that way.

In the spring of 2020 Marshmallow decided to try to broaden its target market from migrants to others who might find getting cover more difficult or expensive from standard insurers. The Kent-Brahams had the idea of using “open banking” details to enable an assessment of risk (and therefore premiums) in a way which seems not to have been used in the insurance industry before (though it was used elsewhere). The idea was to seek permission from applicants to review their banking details so that risk could be assessed by reference to (for example) income levels, certain excluded occupations (eg delivery drivers) and the existence or otherwise of expensive (payday) loans. This, when implemented, would allow Marshmallow to cater for a different section of the market in a way in which (it was thought) other insurers did not and obtain a commercial advantage. The idea was approved by Mulsanne after discussion and input, and detailed rules and tables drawn up by or for Marshmallow for the ratings engine to implement this new idea. Parts of the relevant rules are the basis of the confidentiality claim, and a refusal to hand over data from this exercise is part of the claim based on non-compliance with the TOBA.

Stepping back in time, in the summer of 2019 the Marshmallow team began their deliberations and arrangements to set up their own underwriting insurance company operating under the Marshmallow brand for which Marshmallow would again be the broker. This required regulatory approval (in Gibraltar, where it was proposed to base the insurance company), funding from investors and the making of re-insurance arrangements. Over the course of the next year or so all these matters were dealt with. The evidence of the defendants was that it was not intended simply to take over the Mulsanne business. It was intended to operate side by side with the Mulsanne business, perhaps sharing it in a way to be defined in due course.

In the course of negotiations with the regulator, investors and re-insurers, various disclosures were made as to the intended business. It is said by Mulsanne that that involved disclosure of ratings models, which in turn meant there was disclosure (and therefore use) of confidential information insofar as those rating models contained confidential material originating from Mulsanne. It is also said that there was misuse of the Confidential Information when sample quotes were run through the ratings engine, because the ratings engine contained confidential information of Mulsanne. The detail of this is considered below.

Marshmallow did not tell Mulsanne of its plans and intention to set up a new related insurer for about a year, and Mulsanne, or at least those conducting the relationship, did not come to hear of those plans until they were disclosed in general terms at a meeting between Mr McCauley and Mr Caspar Warre of Mulsanne, and AKB and Dr Holliday of Marshmallow, on 29th September 2020. Mr Warre attended on behalf of Sun Capital, a company that had recently taken over Mulsanne.

In the course of getting regulatory approval in Gibraltar Marshmallow had engaged a company named Artex Risk Solutions to provide assistance. Artex had two members on the board of Mulsanne and they will have known of the activity in Gibraltar, but it does not seem that the information got to Mr McCauley. It may be that that was attributable to Chinese walls; the point was not fully investigated. Although it seemed to be briefly suggested that this demonstrated that Marshmallow was not keeping the matter secret from Mulsanne, it seems that Dr Holliday knew that the new project would not be fed back to Mulsanne at least for a time, and he was careful not to tell Mr McCauley before the meeting. Because of the way this case was run I do not find it necessary to make any finding as to whether the knowledge of these two board members falls to be attributed to Mulsanne. Mr McCauley told me, and I accept, that he thought that something was going on because the atmosphere had changed recently, but he did not know of the plan to introduce a new Marshmallow insurer. He was aware that another of Mulsanne’s brokers had done the same sort of thing recently, but that broker had told Mulsanne in advance.

At all events, the meeting was the first time that Mr McCauley had heard of the proposal. His reaction was in no way hostile. I accept the evidence of AKB that Mr McCauley took the news well and said he had expected it to happen some day. AKB’s evidence, which again I accept (it was not challenged) is that that sort of thing was not uncommon in the industry. The result of the meeting was that each side would go away and consider how the relationship would work for the future, on the unstated common assumption that the broker/insurer relationship would continue.

For a short time thereafter the parties worked together as normal, and Marshmallow continued to prepare for its new venture on the footing that it would maintain its relationship with Mulsanne, but on 15th October there was a call between Dr Holliday and AKB on the one side and Mr Warre and a Mr Shaun Hooper on the other (Mulsanne) side, which changed things. Mr Hooper was a senior officer in CCG (another Mulsanne group company – see above). Mr Hooper was present because Mr McCauley had contracted Covid-19 and was about to go off sick. He was much less friendly towards the idea and said darkly that Mulsanne knew its rights. The mood had turned hostile, but no mention was made at that stage of misuse of confidential information or other intellectual property. Thereafter solicitors became involved (Hogan Lovells, “HL”, for Mulsanne and first HFW and then Norton Rose Fulbright, “NRF” for Marshmallow) and the correspondence reflected disputes about termination and termination rights. Once more, the detail of this appears below.

By the end of 2020 all Marshmallow’s arrangements for the new company were in place. Marshmallow Insurance had been incorporated in Gibraltar on 21st October 2020 and it got its authorisation on 7th December. Once the mood had turned hostile and the first solicitors’ letter was received on 20th October those involved in Marshmallow came to realise that the relationships with both the new insurance company and Mulsanne could not co-exist and they pursued the implementation of their new venture on the assumption that there would be a parting with Mulsanne (though it did not occur at that time). The opening letter from Hogan Lovells referred in fairly general terms to Marshmallow’s infringing Mulsanne’s “trade secrets” and sought bars on the use of such parts of the ratings engine that were “jointly owned”.

Dr Holliday decided that he did not want anything of Mulsanne’s to remain in Marshmallow’s engine (for the purposes of the new business - the engine used for Mulsanne business would remain for the purposes of Mulsanne policies while they remained relevant). Much if not most of the then existing engine had still come from the original Hiscox model. On 24th November Dr Holliday identified two things which had to be replaced - a “vehicle segmentation table” and a “claims/conviction table” (both explained below). He gave instructions for the drawing of new tables under those heads. In addition, he asked two employees, Mr Gray and Mr Cumiskey, to analyse each of the other components of the engine and try to identify parts that might arguably be Mulsanne’s so that they could either be re-written by Marshmallow or by-passed by taking decisions which made them irrelevant (eg by simply declining cover to an individual whose application might trigger an offending table or part of the table). Ms Coulthard was engaged by Marshmallow as a consultant to create other items for the new business such as underwriting rules.

The fruits of those efforts were combined with other parts of the Marshmallow/Mulsanne engine to create what Marshmallow considered to be a new engine which was free of any elements which might be said to have come from Mulsanne. It is the defendants’ case that they have succeeded in that enterprise. It is Mulsanne’s case that that they have not. It is also Mulsanne’s case that in setting up the new enterprise Marshmallow has used Mulsanne confidential elements of the then existing engine as a “springboard” to advance its preparations (for example, in putting reinsurance in place) even though those elements did not themselves emerge in the final new Marshmallow engine. It is also Mulsanne’s case that limited parts of its contributions to the ratings engine used in the business had found their way into the new Marshmallow engine despite Marshmallow’s efforts to prevent that.

Having obtained authorisation from the Gibraltar regulator on 7th December 2020, on 4th January 2021 Marshmallow Insurance wrote its first motor insurance policy via Marshmallow, its agent for these purposes. The process of sending out renewal notices to policyholders where expiry of cover was coming up (inviting the policyholder to allow automatic renewal to take place) continued. However, on 8th January 2021 Marshmallow stopped sending renewal notices inviting renewal into a Mulsanne policy and instead they sent out notices which effectively invited renewal into a Marshmallow policy. It is those activities which have attracted the passing off claim.

In the following weeks each side claimed to have terminated the TOBA and disputed the other side’s right to do so. This led to a dispute about the handover of information to Mulsanne so that it could (as Mulsanne says it wished to do) service its outstanding policies at least in terms of renewals and mid-term adjustments (“MTAs”). This dispute led to another aspect of this action, with Mulsanne claiming a breach of the TOBA in respect of a failure to hand over properly. In due course this action was commenced, and the correspondence and demands about this ran in parallel with this action.

The defendants claim that when they started the new Marshmallow venture in January 2021, selling Marshmallow policies, they ran 2 versions of the ratings engine. The first was the one formerly used for Mulsanne policies which they continued to run for the purposes of servicing subsisting Mulsanne policies where necessary. In the end, and after the resolution of some confusion as to what parts of that engine were used for, nothing turned on this. The other engine was one which Marshmallow claims was created using only Marshmallow material and not using any confidential material contributed by Mulsanne. This was used for the new Marshmallow business. Two versions of this model were theoretically in issue in this case - one used from January 2021 to April 2021, (“the January 2021” model), and another used from April onwards (“the April 2021 model”). For the most part, so far as relevant to the issues in this case, those models are the same and it is unnecessary to distinguish between them, and for ease of reference I shall call them together “the Marshmallow 2021 model”. On the odd occasion on which it is necessary to do so I shall distinguish between the two versions by using the descriptions just given. From about September 2021 Marshmallow used yet another model. That has not featured in these proceedings at all and is not the subject of the present claim.

It is necessary to understand the material in this section in order to understand the claims based on breach of TOBA. I shall set out the background here.

The business of Marshmallow was not selling Mulsanne policies simpliciter. Alongside those policies it also sold or provided different services. None of these were anything to do with Mulsanne save that they were occasioned by the selling of Mulsanne policies. Mulsanne had no objection to these. At the time they were entered into Mulsanne was not provided with any of their details, and at the time it did not need them, though it claims that on the termination of the TOBA it came to require details.

The additional services and facilities were as follows:

Although not an activity of the same nature, it is also useful to record here the servicing activities which Marshmallow also carried on as agent for Mulsanne. These are basically the activities around mid-term adjustments referred to above. If a policyholder wished to alter or cancel a policy during its term he/she would contact Marshmallow and arrange for the policy alteration. I have given examples above - the addition of another driver to the policy, or a change of car. The application would be considered and (if appropriate) implemented by Marshmallow on behalf of Mulsanne and the varied policy carried forward thereafter. It also, of course, dealt with renewals at the end of each policy year.

Those background matters thus give rise to three substantial areas of claim in this case, to each of which I have devoted a separate Part of this judgment. They have each been outlined above, and I summarise them again here:

Some further explanation is necessary of the development of the confidential information case. When the case was started the Confidential Information in issue was described by reference to specific information described by reference to the documents containing it and set out in a confidential Annex to the Particulars of Claim where the various items are given labels of DM-2.2 to DM-2.13 (with item 2.3 being divided into an A and B). There was a general allegation of ongoing misuse and a general allegation of springboard use with no particulars “pending disclosure and/or further information”. There was little formal particularisation thereafter, and that causes difficulties as will appear. As the case, and particularly the trial, progressed the case became more limited, with most allegations of continuing misuse being abandoned and springboard complaints becoming more significant. At an early stage of the trial I required Mr Howe to give some particulars of his springboard case which he did at the beginning of Day 2 in an informal document, not a formal pleading (he was not required to put in a formal pleading) - the “Springboard Note”. That indicated the items on which he still relied on as springboards, and in general terms indicated the nature of the springboard use. By the time of final submissions various items of that springboard use had also been abandoned. The items dealt with below are what is left of the original case after the various abandonments that occurred over time. Rather less than half the original substantial case was left by the end.

I shall deal with the overall claims in the order in which they appear above.

There was little dispute as to the underlying law on this topic. In what follows I deal only with the general law of confidence. The TOBA contains an obligation not to misuse confidential information, but for these purposes it adds nothing to the general law. There was also a pleaded reliance on the Trade Secrets (Enforcement etc) Regulations 2018, but both sides acknowledged that those regulations added nothing to the general law, so I shall not refer to them again.

Because of the absence of many material disputes as to the law, I can set out the relevant law relatively briefly.

With those legal points as background I can now turn to the detail of the surviving heads of alleged infringement.

Before turning to the detailed claims, certain generally applicable points have to be borne in mind.

First, the case has been all about the use of information provided by Mulsanne. Where Marshmallow has contributed the information to the Mulsanne/Marshmallow venture there is no allegation that it is prevented from continuing to use it in the new Marshmallow venture. Thus the Hiscox model was treated as being Marshmallow’s which it is free to use so far as it wishes to do so, and where it was alleged that during the venture Marshmallow contributed further material it was not alleged that that material (so far as genuinely contributed by Marshmallow) was restricted either. This case is all about material contributed by Mulsanne.

Second, as already pointed out, some of the surviving items of claim are relied on as having actually been used in the new Marshmallow model (an “ongoing use” claim), and some are relied on as having provided a shortcut or “springboard” for Marshmallow, without actual ongoing use thereafter. Some are relied on as having both elements. The way in which the claim was originally presented, developed and then partially abandoned makes the journey through the claim confusing and difficult to work out at times. In order to try to keep the matter as clear as possible I shall first go through the items said to be Confidential Information, deciding (where there is a dispute) whether it was confidential material at all, and then deciding whether there was ongoing use if such use is pursued by the claimant. Where springboard use is alleged I shall defer a consideration of that use to later in this judgment.

In what follows the heading to each section reflects the wording in the Annex and therefore the pleaded description of the Confidential Information in the confidential annex to the Particulars of Claim. The annex should never have been treated as confidential because the generalised description, as opposed to the contents of the documents described, cannot be said to be confidential, so I am free to use that wording here.

This item is the subject of both an ongoing use claim and a springboard claim. It arises out of the following circumstances.

Part of all relevant ratings engines in this case was a table relating to occupations. Risk is said to be related to occupations (some occupations presenting a greater risk than others) and the purpose of this section of the ratings engines was to allocate each occupation with a loading (a relativity) which reflected the perceived risk associated with that occupation. Some occupations would simply be declined, depending on the attitude of the insurer. The ratings engine would reflect that too, so that in the event of an applicant listing one of those occupations in an application it would simply be declined.

The Hiscox model had such a ratings section. It was built by Mr Hemming and Dr Holliday then had an input. It was the result of considering market data and underwriting judgment. At that stage it did not have anything of Mulsanne’s information in it. It contained approximately 2000 occupations. Then in June 2018, in the course of negotiations and preparations towards launching the Marshmallow/Mulsanne enterprise, Mr McCauley commented that “there is a lot of stuff that is just in the wrong place” and proposed amendments. The amendments proposed are those in item DM-2.3A in the heading to this section. They were duly accepted and incorporated into a working file, which is DM-2.3B in the heading to this section. Thus the information relating to changes is the starting point for this head of claim.

The amendments were in relation to two aspects of this part of the model. First, there were amendments to 50 of the loadings to reflect the views of Mr McCauley. He specified that some had to go up a bit and some down a bit, without specifying by how much, and Mr Hemming used his judgment as to the amount of the modifications required, between 10% and 20%. Then there were 5 alterations (generated by Mr Hemming) to render descriptions gender-neutral. As well as those altered loadings Mr McCauley proposed (and Marshmallow accepted) 30 changes to the decline rules. Mr Hemming had assigned 103 occupations to the “decline” category (ie applicants for cover falling within those occupations would be declined). Mr McCauley altered the status of 32 occupations. He did so by making 8 of the Mr Hemming’s declines acceptable, and adding another 24 to the cohort of declines. In doing so he used his experienced view as an underwriter. Those changes were specified in item DM-2.3A, and were incorporated into the overall table in DM-2.3B. All those matters then found their way into the ratings engine as it was initially operated by Marshmallow for Mulsanne’s business.

In April 2020 Dr Holliday embarked on the creation of a new occupations table, to replace the table just referred to. This was triggered by concerns expressed by AKB about recent adverse changes in loss ratios. He described how he went through each of the 2000 or so occupations and divided them into 50 groups, gave each group a base premium and then further individually calibrated a base premium for each occupation by reference to his “gut feel”. Mr McCauley accepted in his evidence that this was a different approach to the approach adopted in the original file. The tables were then submitted to Mulsanne for approval, and Mulsanne proposed only one modification, turning one category of an unemployed person from a “decline” to an “accept”. The decline rules in this new table were such that 30 of the 32 declines from 2018 had a match in this table. This table (with the one amendment proposed by Mulsanne) was accepted and passed into use in the Mulsanne business. It became part of the then existing model.

When Marshmallow started its new business in 2021 it used the ratings table thus created (with its declines).

Those facts are the background to the claim made under this head. The claim is that Mr McCauley’s contributions to the ratings and decline elements in 2018 are confidential information, and the current claim is that both heads were misused in springboard activities when the then current model was used in Marshmallow’s preparation activities for its new venture in 2020 in a manner to which I will come. There was originally a claim for ongoing use in relation to both elements, in the sense that both were said to be wrongfully used in Marshmallow’s 2021 models, but the allegation of ongoing use was abandoned in relation to the ratings input (the 50 adjusted rates), and is maintained only in relation to the declines proposed by Mr McCauley in 2018.

Mulsanne’s case involves identifying Mr McCauley’s contribution to the occupation loadings and decline rules as confidential information and following them through into something that amounts to use of them by Marshmallow for the purposes of its own business. The next step is to consider whether the contributions of Mr McCauley amounted to the contribution of confidential information which is capable of attracting the law’s protection. I can deal with this point shortly. I do not consider that it can. The suggestion that a few rates (out of 2000) should go up or down a bit is too vague and trivial, and the actual figures used in this respect were not Mr McCauleys they were chosen by Mr Hemming. The contribution of some ideas about declines is, in the context of the overall file, again too trivial and insubstantial to attract protection as confidential information. That is an end of this point.

However, for the sake of completeness, I should point out that a large part of the misuse case fails anyway. Mr McCauley’s contributions were not used once Dr Holliday’s 2020 model was adopted. If Dr Holliday had simply carried the information over into his 2020 model without thought, and particularly if he carried it over because he was relying on its having been provided by Mr McCauley, then that would have been capable of being misuse. However, I find that that is not what happened. I find that when Dr Holliday created his 2020 model he did so as a result of his own creative efforts, using his own judgment and not using any information provided by Mr McCauley. His evidence paints a clear picture of his starting with, in effect, a blank canvas and using his own ideas of how to go about the exercise. His evidence is to the effect that he formed his own views as to the rating to be applied to the various occupations and the declines and he described how he went about the operation. His manner of doing so does not involve reproducing Mr McCauley’s contribution to the loading to be applied, and I find he did not make use of it. He did his own exercise, even to the extent of choosing a different set of occupations - his new file contained 2,039 occupations as against 1,962 occupations in the previous file. Dr Holliday therefore added a number of occupations. Crucially, Dr Holliday was not challenged in cross-examination on his assertions that he performed an independent exercise.

I therefore accept his evidence that he did. His evidence in chief did not expressly deal with the correspondence of the decline changes originally proposed by Mr McCauley in 2018 as between the two versions (30 out of 32 changes proposed) but unless I decide he was lazy and merely took over the status of those 30 (out of Mr McCauley’s 32) then his unchallenged evidence of independent consideration extended to them too. There is evidence that he was aware of the matching declines, because in an email of 14th May 2020 to AKB and others in which he set out his methodology in some detail he ended by saying:

Though this document was relied on in Mr Howe’s final submissions as showing use of Mr McCauley’s work product but it was not put to Dr Holliday in cross-examination so I did not have his explanation. In those circumstances it would be wrong to hold it against him as indicating that he copied all of Mr McCauley’s declines because they were there. He may have considered them all separately in their original listing (it would seem that he at least carried out a comparison) but if he did he probably formed his own view about them on their merits and formed his own judgment. That does not amount to misuse of Mr McCauley’s confidential information (assuming it had that quality to start with). It amounts to applying his own underwriting judgment.

The experts were not in total agreement on the extent to which an underwriter would be likely to have come up with precisely the same 30 declines as Dr Holliday did. Mr Hillon said that it was unlikely that any 10 given underwriters could come up with exactly the same 30 decline rules as survived into the 2020 table, but that does not determine the point. What Dr Holliday says he did would be of prime significance if I accept it (which I do). It was not said that any one or more of them was so odd that copying that one, and therefore the others, was a more likely explanation than independent judgment.

It must be remembered that by now the result of Mr McCauley’s decline proposals back in 2018 were buried in a long list of occupations. The evidence is confusing on the extent of the burial. Mr Rourke counted 163 declined occupations in the 2021 Marshmallow model (in which the occupations table was the same as that created by Dr Holliday in 2020), as against 117 declines in total in DM-2.3B. It is not clear to me how this squares with what Dr Holliday said in his email about declines remaining the same, but in any event the McCauley changes are a quarter or less of a longer list, which is what is important here. It is not as though by now they were a separate list which can be seen, as such, to have been reproduced so as to give rise to inference of copying and therefore misuse (assuming confidentiality). They were part of a longer list, and I find that Dr Holliday considered them each on their merits and applied his own judgment in that context. There is no inference of copying misuse to be drawn from a correspondence of the particular set of declined occupations.

It follows that the limited parts of these two items DM-2.3A and DM-2.3B which are said to have been misused were not used at all after April 2020 because the ratings engine from then on did not reflect Mr McCauley’s changes. It uses the fruit of Dr Holliday’s independent evaluation. It follows, therefore, that both the direct use and the springboard claims which are made for the period after April 2020 fail.

There is one springboard event, in February 2020, in which this material would have been used because it was still part the business’s ratings engine then. However, since I have declined to hold the material to be confidential the question of misuse on that occasion does not arise.

This is a table which governs the acceptability of applicants with a prior claims or convictions history. Instead of having a rule which simply declined cover to such people, the tables contain various criteria reflecting such matters as the frequency, age, number and nature of claims and/or convictions and provides loading and excesses for some of them, and refer rules (requiring the matter to be referred to a human being) or decline rules for others, depending on the seriousness of the history. In the absence of these rules Marshmallow would (according to the evidence of AKB) be likely have a straight “decline” rule for levels of claims or convictions which crossed a given threshold.

It is no longer suggested that this table was the subject of actual ongoing use in Marshmallow’s 2021 model, but it is again relied on as part of the springboard claim. For this purpose it first has to be established that it was actually incorporated into an overall model used by Marshmallow/Mulsanne in their business. An unnecessary air of mystery was injected into the trial in relation to whether it was actually used. In his final submissions Mr Hough submitted that this particular table was not put into use by Marshmallow. He first injected an air of uncertainty into the trial by suggesting to Mr McCauley that he could not say that this table, while admittedly provided by Mr McCauley, was actually put into use, to which Mr McCauley responded that he could not himself say it was but that he expected that tables provided by him would be implemented. That, presumably, is what underpins Mr Hough’s submissions. The point was not actually dealt with by Mr Howe in his final submissions, though his submissions proceeded on the assumption that the table was put into use.

A bit of archaeology has led me to a finding that it was incorporated. A first version of the claims and conviction table was sent by Mr McCauley to AKB in a short email of 22nd June 2018. A revised version was sent by Mr Graham Preedy of Mulsanne, along with other tables, as part of an overall proposed engine on 16th October 2018. The email opens:

One of the items below is “Revised conviction rating tables - which also drives any additional conviction excess.” So at this stage the table was to be included with other specified tables as part of the new model, and it was to be encoded by an appropriate developer.

In his main witness statement (in a confidential part) at para 8.14(h) Mr McCauley said that the table was put into use. I suppose that that has to be qualified by his evidence that he did not know that it had actually happened, but it was not actually challenged in terms by any positive averment, let alone evidence, to the contrary. AKB’s evidence in cross-examination provides a clear indication that the table was adopted. The thrust of his evidence at Day 6 pages 118 to 120 is that the table sent to him in June 2018 was utilised “at a later stage” and that Marshmallow used a table provided by Mr McCauley.

I therefore find that the claims and conviction table was incorporated in the model used by Marshmallow for Mulsanne business from the end of 2018 onwards.

I also find that the table had the quality of confidential information. Mr Rourke’s evidence categorised this table and its contents as being “red” in his categorisation, which meant that the contents were difficult to derive from market data or other sources and would require experience and expert underwriter input for its creation. Mr Hillon did not dispute that. That is what Mr McCauley put into it. It falls within the description of confidential information referred to above. It was a table created by the expertise of Mr McCauley and was conveyed for the purposes, and only for the purposes, of the new venture on which Marshmallow and Mulsanne were embarking. As a special creation for that purpose it was capable of attracting confidentiality, and I consider that there was an implicit obligation of confidentiality about it - Marshmallow would not have been free to show it to anyone else, or to use it for its own purposes. Dr Holliday was probably implicitly acknowledging that when, as part of his attempt to purge the Hiscox table and other Marshmallow-used material from Mulsanne contributions, he gave instructions to Mr Hemming to create a new claims and conviction table (in an email dated 18th November 2020).

That being the case, it enters upon the territory of Mr Howe’s springboard arguments, which are dealt with separately below because they cover several items of confidential information in the same way. His arguments turn on the use of this material as existing in the ratings engine used by Marshmallow for Mulsanne business, so that when that engine is used otherwise than for the purpose of that business (in a manner that will appear) that amounts to misuse of the confidential information.

This is an elaborate file which divides individual vehicles into various categories and subcategories to which various loadings, excesses and declines are attached. It is a file of some complexity, derived from the application of underwriting skill and experience even though it starts from some listings and categorisations made available by an organisation known as Thatcham Research for the Association of British Insurers. I do not need to go into the details of this file. It is sufficient to record that the defendants did not challenge the confidential nature of it. It falls to be treated as Mulsanne’s confidential information for the same reasons as the claims and convictions table in Item DM-2.4. It seems to have been modified after 14th September 2018

No claim is now made of ongoing use of this material in the new Marshmallow Insurance business, no doubt because the evidence made it clear that Ms Coulthard created her own segmentation file on the instructions of Dr Holliday, and it is that new file, not derived from Item DM-2.5 or its variations, that is used in the Marshmallow Insurance business. However, there is yet again a springboard claim made out of the subsequent deployment of the engine of which it formed part for what Mulsanne claims were improper purposes connected with the start of the Marshmallow Insurance business. Yet again I deal with this point below where I gather together claims of a similar nature.

Reliance on this item is somewhat complex. It is not relied on as a model which was Mulsanne’s confidential model as a whole which was misused as a whole, but as a model which contains some elements of allegedly Confidential Information which were subsequently misused in three ways. First, it is said that items were misused when subsequently the overall ratings engine was used for a purpose other than the Mulsanne/Marshmallow business. The ratings engine is said to have been used so as to generate a series of sample quotes for various purposes concerned with setting up the new business, in a manner which will appear, the first of which was in February 2020. Second, it is said that a table of exclusions was misused when Ms Coulthard used it to formulate some underwriting guidelines in preparation for the new Marshmallow business. This is another springboard use. Third, it is said that the exclusions were used as such in the new Marshmallow business because Ms Coulthard essentially copied them. This is an ongoing or direct use claim.

I will first consider what was in this overall file, and the consider which elements are relevant to this part of the claim and whether a misuse claim has been made out in respect of them.

On 22nd June 2018 Mr Preedy of Mulsanne sent AKB a new ratings model. This was, according to the evidence of Mr McCauley, to improve on the Hiscox model so as to make the model fit to go on to aggregator websites (which in due course it did). What Mulsanne did was substitute its own version of some of the elements of the Hiscox model, or to add elements, so that the model was made up of mainly Hiscox elements but now also some Mulsanne elements. This was the model that went forward into the next year (if there were alterations they do not matter for present purposes). The matters which were said to be added were set out in the email, in some cases with cross-references to the relevant tables within the enclosed model.

Mr Howe does not rely on all elements of the DM-2.6 model, because he accepts that many parts were contributed by Marshmallow alone and he does not base any part of his claim on those. The trouble is that it is not at all easy to ascertain which parts he does now rely on, because of a lack of pleading and a lack of clarity in his final submissions as to how he puts his case. His Particulars of Claim merely sets out the wording as it appears in the heading to this section of this judgment without any particularisation of which parts of a complex set of tables are relied on or how they are relied on. In the course of the trial, and at my instigation, as part of his informal indication of how by then he relied on that Annex Mr Howe said he relied on elements of the exhibit for continuing use and parts as springboard. He then sought to summarise his case in the Springboard Note referred to above, which he then further amended during the course of the trial by making further excisions.

In relation to this Item DM-2.6 document he said that the “factual background” was said to be set out in paragraphs 274 to 277 of Mulsanne’s opening skeleton argument and said that “Marshmallow used this Confidential Information as a springboard by referencing to/cross checking it, and using it as the basis for price comparisons, when creating the January [2021] Model and April [2021] model”. The factual background in those paragraphs of the opening skeleton merely refers to additions. It refers to a number of matters (some of which post-date the creation of DM-2.6) without identifying them as matters actually relied on as misused confidential information and some of which were never pursued under this head (or at all) - the matters referred to are indeed set out as factual background and no more.

Mr Howe’s written final submissions do not fully clarify which parts of DM-2.6 as it was provided on the October date specified are relied on for springboard purposes. The first springboard event happened in 2020, and by then various parts of DM-2.6 had been amended or substituted. As a matter of pleading and case control that is important because the pleading in relation to DM-2.6 is in effect a pleading in relation to that exhibit, and its elements, as it stood in October 2018. But that turns out to be not entirely Mr Howe’s case as it was developed, because of subsequent alterations by the time of the springboard events relied on. It is apparent that Mr Howe from time to time relied on later versions of the tables within the overall engine.

I will have to decide whether that really matters for present purposes, because a case based on those alterations is a case which was not articulated in the pleadings, or in either form of the Springboard Document, and is not easy to tease out of the final submissions. It seems to me to be fair that the exigencies of a speedy trial may sometimes mean that a slightly more relaxed view of non-pleaded material should be taken provided that is not unfair to the other side. However, maintaining a proper control of the facts means I should do the exercise of considering “pleading points” where necessary, and I have to do it without the assistance from final submissions that I might otherwise have expected (again, probably the consequence of a speedy and condensed trial).

It is paragraph 536 of Mr Howe’s submissions that comes closest to identifying all the items in DM-2.6 that he relies on as giving rise to his springboard claim. It appears in the context of his considering the first springboard event in 2020 when reinsurers were approached by Marshmallow. He says:

There was a significant debate as to whether the postcode file just referred to was actually included in the DM-2.6 file. In the end I do not consider that matters much because I have concluded below that a further Mulsanne postcode file was introduced in 2019 anyway, and that later file was in place by the time of any springboard argument to which DM-2.6 is relevant, but in the light of the debate, and in case it matters, I will express my conclusions on the point.

The postcode file is an element in the engine used to produce ratings (multipliers) applicable to each postcode, because risk has been seen to be relevant to the policyholder’s postcode. There are approximately 1.6m residential postcodes in this country. The technique used to allocate the relevant rating to the relevant postcodes in this case involved allocating each postcode to a group, and then allocating a rating to that group. That was the technique used by Marshmallow when it produced the Hiscox model. It divided the 1.6m postcodes into 60 groups, and allocated a multiplier to each group. There were therefore 60 ratings (multipliers).

In 2018 Mulsanne had produced its own postcode file, which relied significantly on its own underwriting and claims experience. This had 99 postcode groupings to which 99 ratings were applied. There is a dispute as to whether it was this file that was contained in the overall ratings engine DM-2.6 or the Marshmallow one.

It is a little surprising that there is no direct evidence on this point. One would have thought that someone would have been able to give clear evidence on this point. However, I consider that on the probabilities the Marshmallow file was the one which was still included in this engine. In his covering email by which Mr Preedy sent the whole Rating Model he listed the items that he had changed and they were those listed under (a) to (f) above. They did not include a different postcode file. That suggests that he did not add the then existing Mulsanne file.

The spreadsheets representing Mr Preedy’s Rating Model DM-2.6 contained a table listing 60 postcode ratings against the numbers 1-60, not 99. The matching postcode grouping or allocation was not within the exhibit available to the trial. The fact that there are 60 ratings and not 99 strongly suggests that the relevant postcode lookup table would also generate only 60 groupings, which in turn suggests that it is Marshmallow’s overall file that was part of the engine and not Mulsanne’s. Otherwise, as Mr Hough pointed out to Mr McCauley, there would be no rating applicable to the postcodes in 39 groups and the tables would not work properly or at all. Mr McCauley suggested that the table would work with Marshmallow’s ratings and Mulsanne’s 99 groupings, because a postcode in the missing 39 groups would generate a decline. That does not seem to me to be a plausible explanation. It seems to me to be unlikely that such a hole would be allowed to exist in a carefully constructed model such as this, and it would seem to me to be likely that the number of declines thrown up would have caught someone’s attention in due course and any mis-match rectified. There is no evidence that anyone spotted such a mis-match. I have not overlooked the fact that on the first page of the exhibit, which is a sort of index with comments, there is the following comment against “Postcode category”: “Postcode category derived from the postcode database, a number from 1 to 99. Use risk postcode." This may indicate that Mr Preedy had it in mind to use the Mulsanne file, but I consider that he failed to do so.

I therefore conclude, so far as relevant, that it was Marshmallow’s postcode file (the whole postcode ratings element) that was within Item DM-2.6. It therefore cannot be relied on as Mulsanne’s confidential information in this case.

It is now necessary to consider the other elements of Item DM-2.6 that are said to be relied on (see the list above). I shall take them sequentially. Most, but not all of them, correspond to items which Mr Preedy’s email said he was sending.

Exhibit DM-2.6 contains a large spreadsheet containing some generally applicable underwriting rules under the headings “Cover”, “Drivers”, “Addresses”, “Vehicle” and “Claims/convictions”. Within those headings are rules prescribing, for example, minimum vehicle value, maximum vehicle value, minimum policyholder age, whether the vehicle is left hand drive, maximum number of prior claims and many other such factors, with the consequence if any of the rules are “breached”, which is (usually) a decline. These rules, in DM-2.6, were provided by Mulsanne in place of rules which Marshmallow had proposed to use in its Hiscox model.

It is said that some of these rules were used by Ms Coulthard when she devised underwriting rules for Marshmallow’s new business. It is also said that these rules were in the overall ratings engine when that engine was run to generate sample quotes for testing and evaluation purposes in the springboard context.

The first question is whether these rules should be treated as being confidential information of Mulsanne. Both experts agreed that the vast majority of these rules were “relatively straightforward” (Mr Rourke’s words). They could be re-created by an experienced underwriter to a similar standard. Mr Rourke also agreed that if a customer rang up and asked about one of the rules (for example, whether he/she would be excluded by age considerations) that customer would be told, so to that extent many of the individual rules would not be treated as confidential. However, he also said, and I accept, that an insurer would not disclose all the rules to a competitor because insurance companies are very sensitive to competition and the rules in bulk (the whole rule model) would not be something it would wish to disclose. To that extent I would consider the whole underwriting table or spreadsheet to have the characteristic of some confidentiality, but bearing in mind the ease with which they could be reproduced and the fact that individually they would not be kept secret I consider the level of confidentiality to be low, and the significance of any use of individual rules to be likely to be very limited in terms of damages or springboard.

The next question is whether any confidentiality in these rules was breached, which involves considering what use, if any, was made of the rules. The first alleged use was said to be in drafting underwriting guidelines.

This section of this judgment relates to part of the DM-2.6 underwriting rules, namely its exclusions - that is to say, its provisions for excluding cover. In September 2020, after Ms Coulthard had been taken on to assist in the preparation of the new business, she was asked to prepare underwriting guidelines. Those guidelines are a separate document and are normally prepared for the benefit of staff administering the policies, so that they can, for example, explain to a customer why an MTA request is refused - for example, driver too young, car too valuable, and so on. They are (at least judging by what Ms Coulthard produced) a narrative document, not part of the ratings engine. The Ombudsman might need to see the guidelines in the event of a dispute between policyholder and insurer. On this occasion the need for guidelines, at the time they were written, was so that they could be shown to the intended re-insurer Scor Re.

On 25th September 2020 Dr Holliday emailed Ms Coulthard welcoming her to Marshmallow and saying that Scor Re was keen to see the underwriting guidelines/underwriting policy. He went on:

Ms Coulthard responded to AKB:

On 30th September he sent a document under cover of an email which said:

It is the attachment to this email from which Ms Coulthard worked in producing the underwriting guidelines that she went on to produce. She did not work from anything within the model which is exhibit DM-2.6. She explained that she did not have access to the ratings engine as such because she did not have access to that part of the company’s network. The significance of the other documents that were sent was not made clear to me. The essence of the case propounded by Mr Howe and met by Mr Hough was that Ms Coulthard worked from the document she was sent, and she admitted that.

The document she was sent was not a straight extract from, or reproduction of, the underwriting rules in DM-2.6. It was a spreadsheet which set out various rules in summary form on alternate lines, with intervening lines briefly explaining the purpose of the rule. They were summaries of the sort of rules that one would expect in underwriting rules, but not formally formulated. Mr Rourke has analysed the rules as there set out and compared them with the underwriting rules in DM-2.6. There are 33 rules in the new document sent to Ms Coulthard. Mr Rourke concluded that the majority of the rules in the document had a matching rule in the underwriting rules in DM-2.6 or had a rule with an equivalent effect. Assuming his comparison is correct (and it was not challenged) his statement about the majority is true, but it is not the vast majority and the evidence needs to be considered with qualifications. His analysis is set out in table 6 of his second report. Out of the 33 rules 10 did not have an equivalent - almost one third. Against the rules said to have an equivalent he has added comments which indicate that there are not true equivalents, so the majority comes down further.

That suggests an alternative source for these rules, either in addition to or in place of what appeared in DM-2.6. There was no evidence of the making of any alterations in DM-2.6 which might have been reflected in the parts that did not match, or of who was responsible for those mismatches. On any footing someone has had some input into the table since the first rules were incorporated into DM-2.6. It was not apparent at the trial who it was. Each side might be accused on not filling in this gap with evidence, so drawing an inference from the absence of evidence would not be fair.

However, doing the best I can on the evidence I heard, it seems to me to be likely that the document had at least its roots in the original underwriting set. The descriptions of the documents in the emails, and Dr Holliday’s suggestion that they use current material, suggests that the source is a current version of underwriting rules or guidelines, which provides a link to Mulsanne material. The additional matters that do not have a corresponding entry in DM-2.6 must have come from somewhere else, and are not part of the Confidential Information which is the subject of this part of the inquiry, but the other two-thirds are likely to have come from there. True it is that most or all of them could well have been devised by a competent underwriter; but that does not mean that they had an independent source on this occasion. A finding that the source was some part of the existing model is made more likely by the fact that on the chronology, when this material was provided to Ms Coulthard, the dispute had not yet arisen and (as I find) Marshmallow were still assuming that they would be operating alongside Mulsanne. The point at which they started to purge Mulsanne material had not yet arisen. Dr Holliday and AKB would therefore be less likely to be sensitive to the potential difficulties of using existing material.

That deals with the original source of the material. I now have to deal with the use that Ms Coulthard made of it. She produced her first draft guidelines quite quickly, by 5th or 6th October. She was preparing a draft so it could be produced to the reinsurers. A finalised version would come later. In cross-examination she admitted that she would have used the exclusions from the spreadsheet and transposed them into the guidelines. The draft is a long, 30 page document with a lot more material in it than just the exclusions in the spreadsheet, and is obviously primarily the work of Ms Coulthard’s underwriting and insurance judgment and experience.

Of the 33 exclusion rules in the spreadsheet with which she had been provided, only 6 were put to her as having been transposed, derived or copied from the exclusion list in the spreadsheet. Mr Howe relied on the equivalence in the content of the rules. In final submissions Mr Howe said his case was not limited to these six instances and he relied on some sort of use of all the rules. However, he did not put any other instances, and he did not seek to demonstrate that others may have been used. His case has to be confined to those 6 instances. It is not even clear how he makes the rest of his case.

When they were put to her Ms Coulthard accepted that they derived from the list, though it appeared from re-examination that one of them was put to her on a false premise of similarity. The six items were:

Further observations need to be made in relation to those items. First, what I have just said compares Ms Coulthard’s underwriting guidelines with the spreadsheet and relates to evidence as to use. The spreadsheet is the link back to DM-2.6, so that an inference is arguable that confidential information in that exhibit was used. I have found that it is likely that to some extent the spreadsheet derives from material in DM-2.6, though the correspondence is far from total. Furthermore, in relation to three of the items on which Ms Coulthard was cross-examined the evidence of the claimant’s own expert witness, Mr Rourke, involves a finding that the spreadsheet does not exactly correspond with DM-2.6. The driver maximum age is said to be the same for new business and mid-term adjustments “but 5 years different for renewals”, and the maximum vehicle age is said to differ by 4 years as between the spreadsheet and DM-2.6. In relation to the vehicle value it is said: “The parameters for this rule are significantly different to those contained in [the relevant tab] of DM-2.6. There is therefore a serious question-mark about whether there really is a link between the draft guidelines of Ms Coulthard and DM-2.6.”

Second, Mr Rourke said that most of the content of the underwriting rules in DM-2.6 could be created using market data or could be derived by a competent motor underwriter to at least a broadly similar level, and in the context of considering similar material in another part of her witness statement Ms Coulthard comments on the similarity of motor insurers’ exclusion lists, including those she had used in previous engagements.

Taking all these points together I do not think that it has been established that Ms Coulthard made any material use of confidential information of Mulsanne when she created her draft underwriting guidelines. The apparent reproduction of 3 rules from the large set, in circumstances where those are (as I find) the sort of rules an experienced underwriter would come up with anyway is not factually significant in this context. The other two rules have not (as I have indicated) been exactly reproduced, and even if the spreadsheet and what it reflects about DM-2.6 were some sort of starting point for her then as an experienced underwriter she could have come up with the same rules anyway.

This claim is a claim that exclusions in the underwriting rules in DM-2.6 were used by Marshmallow in its 2021 models for its own business. It is said that a comparison of DM-2.6 and the April 2021 rules presents similarities which are attributable to (in effect) copying and therefore misuse. (In the interests of clarity, I should say that these are underwriting rules, for the models, not the informal guidelines referred to in the previous section.)

The final version of the rules for the 2021 models was written by Ms Coulthard at the end of 2020. During the course of this exercise she never had access to DM-2.6 (and therefore to the existing underwriting rules). There were attempts to send her material, but she could never access it at the time because (as previously stated) she did not have access to the network on which they were situated. There is therefore no case that can be made of direct copying. The misuse case has to be based on the use of the exclusion list referred to in the preceding section of this judgment. Her evidence in chief made it clear that she used this list when she created exclusions in the final version of her underwriting rules.

Mr Rourke carried out a comparison of DM-2.6 and the final form of Marshmallow Insurance’s underwriting rules. He has listed what he says are the 39 rules in the final form (dating from April 2021) and identifies cases where the rules are “similar”. Closer inspection reveals that he sometimes treats rules as similar where they are nonetheless different in content - for example, both have an exclusion for “vehicle age” and Mr Rourke considers that they are similar rules, but Mr Hillon points out that the rule for one is 4 years greater than the rule for the other. If one is analysing for (in effect) copying content then that difference is important. The claimant does not seek to make a case of misuse by copying a headline category.

Of the total of 39 exclusion rules in Marshmallow’s April 2020 model, Mr Hillon has identified that for 19 a comparable rule does not exist in DM-2.6, 10 are different in the sense that the subject matter is covered but in a different way. 6 “appear consistent”, three “appear similar” and on one he is unable to opine. I find that that analysis is fair. So that analysis does not support a suggestion of slavish adoption of large tracts of the relevant part of DM-2.6 because they were there and it seemed useful simply to port them over. It is plain that on any footing Ms Coulthard applied her judgment and added a large amount of material which differed from DM-2.6. Both experts agreed that the rules which are similar or identical could be derived from market data or by a competent underwriter.

I find that there is no evidence of misuse of confidential information in this respect. Exactly what Ms Coulthard did when she created her document is not clear, not least because she was not cross-examined on the process, and no particular misuse was put to her. All the elements from DM-2.6 which are said to re-appear are standard underwriting conditions. That does not prevent them being treated as confidential information as a body on the footing that a collection of standard rules which requires work to construct can be treated as confidential, but it does mean that when a limited number of rules are isolated and re-appear in a subsequent document, then it is possible to say that they re-appear not because they were deployed as being a useful body of existing rules to deploy, but because, having seen them, a separate underwriting judgment is applied and the same rule results. It is entirely plausible that the latter process was gone through, and if that is the case then it would not amount to a misuse of confidential information. That, in my view, is what happened. In her second witness statement, which was largely unchallenged (and not challenged at all in this respect) Ms Coulthard went through what she perceived as being the 15 items which are said to have been the subject of the misuse claims. She advanced various explanations for their inclusion as categories and/or the choice of content of each category. The categories and/or their contents were either normal in the market or the parameters were what Marshmallow wanted in its risk profile. There is no suggestion that she simply adopted wholesale. If what she did was to identify relevant factors, decided to adopt them (using her judgment) and decided on the parameters within the narrow expectations of the market, then that does not in my view amount to misuse of confidential information (even assuming, for these purposes, that some or all of those parameters came to her via the list of exclusions which in turn got them from DM-2.6). It amounts to the proper independent use of industry knowledge and experience.

I therefore reject this claim for misuse of confidential information.

It will be useful to draw those strands of DM-2.6 together and express my overall conclusions here before going forward. I have found:

The Mulsanne postcode file was not included in this file as at October 2018.

A confidential Mulsanne claims and convictions file was included in this file as at October 2018. It is available for springboard arguments later on.

A confidential Mulsanne vehicle segmentation file was included in this file as at October 2018. It is available for springboard arguments later on.

The occupation file with small Mulsanne contributions was included in this file but was not confidential information for these purposes. I would add that its part in any springboard claim would be highly insignificant anyway.

The underwriting rules in this file attract a low level of confidentiality. The rules were not misused by Ms Coulthard in her underwriting guidelines in November 2020 and there is no ongoing use into the Marshmallow 2021 engine.

This document was an amended part of the vehicle segmentation engine which was already in use. There was little debate about this at the trial. The Defence (at paragraph 22) admits its incorporation into the engine (by way of a general admission of use of the Annex 1 material), which, together with its substitutionary nature, probably explains why no-one really gave any evidence about it. A case of direct reliance in the new Marshmallow 2021 business was dropped (the experts found no relevant similarities) so the only part this document plays in the trial is as part of the springboard case based on the use of the overall engine into which it was incorporated. I have already found that the vehicle segmentation section of which it forms part is capable of constituting confidential information so this item goes into that finding, and it falls to be considered as part of DM-2.6 in due course.

This, as its name indicates, was a substituted postcode file generated by Mulsanne containing postcode tables, groupings and ratings allocations, intended to be substituted for the then existing file in the ratings engine. In his email of 25^th April Mr Preedy said:

There is, yet again, a surprising and unnecessary air of mystery as to whether the file was ever incorporated in the system. As I have already indicated, the Defence seems to accept that all the items in the Annex were used for the purposes of the Marshmallow/Mulsanne business, which would seem to involve an implicit acceptance that DM-2.10 was incorporated. The uncertainty comes from the fact that there is no positive evidence of incorporation, and Dr Holliday raised a question-mark over its incorporation in his cross-examination. Then Mr Hemming referred to his being sent a postcode file in 2020 by Anton Leacock as being the postcode file then in use, and that turned out to be the first Marshmallow file, not a Mulsanne file.

The defendants’ final submissions do not seem to take a positive position on this. They do not analyse the evidence. On the basis of the material I have I find that this file was incorporated into the overall engine so that it was there when the engine was used for what are said to be springboard events. I suppose it can be said that in the light of the pleading it seemed that no point would arise, so neither party addressed it in the evidence. Mr Howe’s case originally focused (understandably) on whether there was evidence of use of that file in the Marshmallow 2021 business files (there was none, and the direct use claim was dropped), and the significance of the presence of DM-2.10 in the engine from April 2019 did not really become apparent until Mr Howe put questions relating to his springboard case to witnesses in cross-examination. However, the point has been raised and I consider there is enough material for me to make a finding that it was incorporated, which I do for the following brief reasons.

So far as the pleading is concerned, it seems to assume incorporation. That is an important start. The email exchange set out above clearly connotes an acceptance by all sides that it should be incorporated. No reason was advanced by anyone as to why it should not be incorporated, and it therefore more likely that it was. Dr Holliday’s evidence was that he expected that it would be. Significantly, in procedural rather than evidential terms, on Day 3, when Mr Hough was cross-examining Mr McCauley about events in 2019 and referred to DM-2.10, Mr Hough said to me (in answer to some questions that I had):

I am not sure that he is right about the evidence at that point, but in any event that seems to reflect the position taken by the defendants, and when that is taken with the other material to which I have just referred it would make it wrong to allow the defendants to make some sort of alternative case (which to be fair to them they do not do in positive terms).

I therefore conclude that DM-2.10 was inserted into the overall ratings engine and was used thereafter. The significance of this is that it was part of the engine when it was used in what are said to be springboard exercises.

I need to make some findings about confidentiality here. Like some of the rest of the material contributed by Mulsanne it attracts the characteristic of confidentiality in that it is the product of an assembly of ideas. It is not (as I find) something which Mulsanne would be prepared to pass to a competitor, or to anyone else other than Marshmallow, because it might give that person a competitive advantage or put part of Mulsanne’s business know-how in the hands of an outsider. Mr Rourke categorised this file as amber, that is to say it could not have been created by a third party solely by market data. It requires more.

Whether it was misused or not is something to which I will turn in the springboard section of this judgment.

The rules referred to here relate to the situation where an applicant misrepresents the number of years of no claims discount (NCD) that he/she has accumulated. It is possible to check the claimed NCD against a database held by LexisNexis so that the claim can be verified. If the check reveals that the customer has overstated the amount of NCD years then some insurers will simply decline at that point and on that basis. Others will simply increase the premium to reflect the “correct” number of NCD years. Mulsanne’s table sent in April 2020 reflects what Mr McCauley considered to be a different approach. Mulsanne realised, from its own data, that those who misrepresent their NCD by one year have not only made a misrepresentation; that driver presents a much riskier prospect as well. Accordingly Mulsanne produced this table which provides for a loading of the premium to reflect the risk. There is a graduated loading depending on the number of years declared and the discrepancy between the truth and the represented (though actually there are only two values which are ever applied - a slighter loading and a heavy loading, with the latter applied uniformly across a wide range of discrepancies; I will not set out the actual figures here for reasons of confidentiality.)

This table was, as I understand it, incorporated into the ratings model used for the Marshmallow/Mulsanne business. What is alleged is a form of misuse by copying, though not of the whole table. When Marshmallow Insurance started its business its own ratings engine contained nothing like this table. Instead there was a blanket decline rule in January 2021. However in the April 2021 incarnation of its processes a table was introduced which is similar in concept, in that it provides for a graduated response to the amount of over-declaration of NCD by a series of graduated loadings of the premium, but the amounts and gradations are (as the experts agreed) very different from those in the Mulsanne model. There was no evidence as to who devised this Marshmallow table.

I am prepared to find that the Mulsanne table was vested with confidentiality. The contrary was not really argued, and Marshmallow claimed confidentiality in its own table. Any allegations of direct misuse by copying the table itself were not pursued. The experts both said that the actual tables were different in content and presented no evidence from which one might even begin to infer misuse by copying. Mr Hillon pointed to what he called a difference in “philosophy” in that the Marshmallow table was less punitive and less likely to throw up a loading (and therefore a price) which would price Marshmallow out of the market. In his final submissions Mr Howe invited me to find that the Mulsanne matrix was the starting point for the Marshmallow matrix. I suppose what he meant by that was that Marshmallow copied the idea, or some abstract essence of the Mulsanne matrix, but that was not made clear. He would first have to establish that the idea of a matrix, or the idea of a matrix which applied a graduated response, was confidential information which Marshmallow was not entitled to adopt. This fails on the evidence of his own expert. In his second report Mr Rourke said that the concept of a misrepresentation matrix in pricing is common and well-understood in the industry. The format of the Mulsanne matrix was a logical way to present this sort of consideration. The concept of considering misrepresentation is well understood and acted on in the market, and insurers deal with misrepresentations by pricing changes, underwriting changes and/or “operating changes”. (2nd report para 5.109). None of this supports the view that the idea of a table with loadings is unknown (and therefore capable of being confidential). I find that there is no confidentiality in the ideas behind this table, and no evidence that the idea was copied from Mulsanne. There is therefore no valid claim of direct misuse by Marshmallow in relation to this item.

Mr Howe’s Springboard Note indicated that a springboard claim was made in respect of this document. I suppose that that is based on the use of the whole ratings model late in 2020, which is considered below.

These are said to be subject of an ongoing use claim and a springboard claim. The ongoing use is in relation to parts of these rules only.

The background to this appears above, but a little more detail has to be supplied in this context. In December 2019 the Kent-Braham brothers conceived the idea of implementing open banking data to provide a broader range of data that could help better understand customers and their underlying risk. The scheme was viewed (at least by Mulsanne) as being one targeted at younger drivers. In that way they hoped to increase the profitability of the Marshmallow insurance scheme and to extend its market. The idea of using this sort of data was not entirely new to the insurance industry, but no-one had put it into operation by this time. This was not part of the initiative to provide for migrants; it was a separate initiative. The Kent-Brahams spoke to various API providers (providers who can interface with bank software in order to extract relevant data) and decided that the enterprise was probably feasible. They then decided on Plaid as being the best provider because it was an aggregator of bank connections with links to all banks in the UK. Marshmallow entered in the contract with Plaid which governed these arrangements. It was not in evidence before me.

The essence of the idea was that, with a policyholder’s consent, Plaid would extract data from the policyholder’s bank data, review it for relevant material and pass the result of the review back to Marshmallow. Exactly how this worked was not made clear at the trial, but that seems to be the gist of it. This data would feed into the policy application and be matched against certain criteria so as to produce a refusal of cover in certain cases. AKB understood that one could look at an individual’s cashflows and get an idea of their spending habits and creditworthiness. For example, it might reveal the amount of loans and the source of some income. Other examples of the useful information that might be provided and how it could be used will appear from the allegations of misuse below.

On 12th May 2020 AKB wrote to Mr McCauley saying he had put together a first draft of underwriting rules for open banking which he was keen to get his thoughts on.

A few days later there was a “brainstorming” session involving Dr Holliday, AKB, Mr McCauley and a Mr Dragne of Mulsanne. Rules emerged from the session which eventually found their way into the formalised rules in DM-2.12. The rules were actually written by or within Mulsanne. There are 12 of them. Precisely how and by whom the final rules were formulated, or how they emerged, is not in evidence. It would seem that a lot of the rules emerged from the brainstorming session. Certainly some ideas were contributed by Mulsanne, but the thrust of the evidence was that the parameters of rules, and some rules themselves, emerged from the discussion (and thereafter, probably).

The direct ongoing misuse of what is said to be confidential information arises from just two aspects of the rules in DM-2.12. It is not clear whether complaint is made about a third. Mr Rourke compared those rules with rules used by Marshmallow in 2021 and has noticed what he says is a relevant similarity in two of the rules. The first is one which seeks to identify the recipient of a payday loan or a large loan, so that they can be excluded, and another seeks to identify an individual who is employed as a delivery driver (in particular a food delivery driver, by reference to well-known companies) and exclude him/her too. He also references large loans - that is the third aspect about which the complaint, if any, is not clear.

So far as the delivery driver rule is concerned, the one line rule in DM-2.12 declines if more than 50% of income comes from one of two well-known delivery companies - the rule is as simple as that. (It was explained that being able to apply the rule would require the identification of one of those companies as payer of a credit in the banking data of the would-be policyholder.) This is said to have been misused in a rule dealing with the same subject matter in Marshmallow’s 2021 engine. In that engine the rule is as follows:

Obviously it cannot be said that the DM-2.12 rule was used by direct copying of the rule. The rule in the Marshmallow 2021 engine covers the same topic, but is very different in formulation. Mulsanne relies on the fact that Marshmallow is using a rule “based on or derived from the Open Banking Rules”, and the fact that the same concept underpins the two rules. It seems to be said that that is a misuse of confidential information.

For that claim to work Mr Howe first needs to identify the confidential information. His starting point would be that the rule itself amounts to confidential information. I do not consider that, in the circumstances of this case, it qualifies as such, or at least it was not a confidentiality that can be asserted against Marshmallow. I accept that a body of rules such as the open banking exclusions in DM-2.12 would be capable of being confidential information, the confidentiality in which could vest jointly in both Mulsanne and Marshmallow. The rules arose out of a genuine joint effort to arrive at a set of rules, with Marshmallow coming up with the overall idea and some areas to be covered and Mulsanne then contributing to their development. However, I doubt that that would have the consequence that no individual rule could be used in another venture by either of the parties. When the parties come up with their joint idea it effectively becomes part of their underwriting stock in trade. Once one has the general idea the details of the implementation become a matter of underwriting judgement and little more (though, as Mr McCauley pointed out certainly a little more) than common sense. I do not consider that the full panoply of confidentiality descends upon each individual rule as between Mulsanne and Marshmallow.

However, even if that is wrong, and the rule itself amounts to an item of confidential information which neither party can exploit outside the joint endeavour or without the other’s consent, that is not what has happened here. What has happened here (assuming for the moment that Marshmallow simply took the concept of the rule from DM-2.12 and deployed it without more) is that the idea has been used but not the substance of the rule. If Mr Howe is to succeed in his argument he must establish that the idea of declining cover to someone who has a significant part of his/her income from a delivery company, is itself vested with confidentiality. I do not consider that it is. Even assuming that the idea was contributed by Mulsanne (and there is no direct evidence of that, though it is accepted that it was not one of the original ideas in AKB’s email) Marshmallow had started the ball rolling and contributed some ideas of circumstances in which the banking data might reveal the unreliability of someone. The ideas, as indicators of unreliability, are not particularly special. Once an idea was proposed between them as being relevant to risk in an open banking data context, I consider that it became the sort of thing that an underwriter would thereafter himself or herself just know, in the same way that he/she would know that it is a bad idea to insure a 17 year old to drive a Ferrari.

The ongoing use claim based on this part of item DM-2.12 therefore fails.

The next item relied on is a payday loan exclusion. The rule in DM-2.12 is:

The allegedly infringing rule in Marshmallow’s 2021 engines is to the effect of the following (taking Mr Hillon’s translation from the java script):

A large number of “merchants” are then identified, presumably all those known to Marshmallow as providing payday loans.

Obviously the Marshmallow rule is much more sophisticated and is not a copy of the basic rule in DM-2.12. The same points arise in relation to this claim as arise in relation to the delivery driver rule, to which the same reasoning is applied. In order to succeed Mulsanne must establish that the idea of a payday exclusion is confidential. It seems to me that, given the idea of looking at a debt position was something which started with Marshmallow in AKB’s email of 12th May, the idea of looking at other badges of payment difficulty would be natural, and would be natural to any underwriter. It would not be sufficiently out of the way to amount to confidential information when it arises, and that is particularly the case when it probably arose out of a to and fro discussion in the brainstorming session or afterwards. The idea simply does not have the quality of confidentiality which is required for the claim to succeed. That assumes, in Mulsanne’s favour, that it came from Mulsanne in the first place, as to which I make no finding.

The claim for direct (ongoing) use by Marshmallow in 2021 therefore fails. It is also said in the revised Springboard Note that there was springboard use in the preparation of the January and April 2021 Marshmallow models “by referencing to/crosschecking”. That is not expressly dealt with in the section of Mr Howe’s final submissions devoted to these rules, but I assume that it is a reference to the use of overall engine at the end of 2020, to which I shall come. If that is what is intended it needs to be established that these rules became part of the engine used by Marshmallow for Mulsanne business and that they were in place by the time of the springboard events relied on in the autumn of 2020. The picture here is confused. AKB’s unchallenged evidence is that by September 2020 8 out of 10 of the open banking rules used by Marshmallow were not in the rules contained in Exhibit DM-2.12. He gives one example. I am prepared to find that AKB’s evidence is correct, though the comparison between the rules (which would require an understanding of the coding appearing in the evidence of the later rules, which is beyond my capabilities) was not carried out at the trial. In those circumstances it is not clear how much, if any, of the rules fall to be treated as something in which Mulsanne had a confidentiality interest when the springboard events arose. Mulsanne is not entitled to the benefit of the doubt in this respect and I find that it has not demonstrated a body of rules in which it had such an interest by the time the springboard events arose in the last quarter or so of 2020. Putting it another way, Mulsanne has not proved that if it had a confidentiality interest in the rules particularised by reference to DM-2.12, then that was a body of rules which was present at the time of the springboard events.

This is the full pricing specification (basically the model) for the open banking project referred to earlier in this judgment. It was based on the original Marshmallow 2017 model (which, with modifications, had been in operation throughout) but with particular modifications and insertions to cater for the open banking aspects. While a number of aspects of this model were originally said to be confidential information which was wrongfully misused by Marshmallow, the direct or ongoing misuse was eventually confined to two aspects - tables concerned with fraud scores and credit scores.

The table with a fraud score is a very short table. It contains three lines and two numeric columns. The first column sets out a fraud score with three numbers - 0, 447 and 551 - and the second column applies a relativity (multiplier) to each of those numbers. I will not set out the relativities because they are said to be confidential, though frankly I do not see how that information, even if disclosed, would be of any use to any outsider without understanding the detailed rationale of the underlying calculations which produce the score in the first place. The allegation which I have to consider at this point is an allegation of direct use in relation to this particular table. That is, strictly speaking, within the pleaded case, because the pleading is of confidentiality and misuse in relation to the whole of the detailed specification of which this table is a small part. Therefore the pleading, strictly speaking, includes this table as a part of the whole, though the table is not singled out in the pleading. I shall, however, have to consider the pleading-related points to which this gives rise. As will appear, the actual table itself is not relied on.

The purpose of this table is to produce a loading for premiums based on a computerised assessment of the likelihood of a policy being cancelled for fraud. Its job is not to predict fraud, or even to flag up applications which need to be considered for rescinding for fraud; it just produces a loading. It arose in the following manner.

Marshmallow prides itself on its fraud detection abilities. So does Mulsanne, and Mr McCauley claims (and no doubt has) an experienced underwriting nose for assessing applications for policies which are likely to give rise to fraud considerations, either at the applications stage or at the claims stage. He says there are various badges which are detectable at the application stage - for example, applications made in the small hours, or applications made quickly after the quote, or acceptances of quotes after a number of different entries into the online forms apparently designed to see how the best quote can be obtained. There are others.

From the outset of their venture both Marshmallow and Mr McCauley assessed applications for badges of fraud and cancelled policies where appropriate. Mr McCauley’s evidence, which I accept, was that, at least initially he would spend 1 to 1.5 hours a day reviewing new policies (details of which were sent through on a daily basis as they were incepted) and he would raise with Marshmallow those which he considered it right to check further, or even on occasions perhaps even reject, for possible fraud. This review process continued throughout the relationship, though after a time Mr McCauley gave way to Mulsanne’s Mr Hardip Sagoo. It resulted in sometimes large numbers of cancellations for fraud.

At some point it was decided that a computer engine should be built to seek to identify applications which demonstrated a risk of fraud and to load premium prices according to the level of risk perceived by the programme. It is not stated clearly in the evidence who came up with the idea, but looking at the evidence as a whole it seems to me to be likely to be Marshmallow. It is indisputably the case that Marshmallow built the engine and the other IT processes involved. What Marshmallow did was to train an artificial intelligence engine with the experience of claims which had been identified as having badges of fraud by feeding it with those cases where that had been done and (as I understand it) providing at least some of the parameters which pointed that way. The provision of parameters aspect was not wholly clear, because no-one gave detailed evidence as to the development of the engine, but that is what seems to me to have happened. The AI engine then “learnt” from what it was given, and was able to give a likelihood of fraud score as some sort of numeric figure. The computer then applied that figure to see which of the bands within the disputed fraud score table applied, and applied one of the three relativities accordingly. Thus was the quoted premium increased (or in one case slightly discounted).

At the trial Dr Holliday explained that the fraud engine was run a second time in order to throw up a score which was a trigger for a human being to assess the policy for the likelihood of fraud.

The choice of the numbers in the table does not seem to have been done by the computer. That seems to have been done by a human being, who must have been in Marshmallow. Thus the actual figures in the table are not said to be confidential information provided by Mulsanne. They are actually meaningless, and in fact ineffective, without the underlying computer engine (Marshmallow’s) which produces the fraud score to which the figures in the table are applied.

Mulsanne’s claim to have supplied confidential information is directed elsewhere. What was alleged at the trial was that Mulsanne’s (principally but not exclusively Mr McCauley’s) contributions in identifying potential fraud cases, and his indication of the sort of factors relied on, were confidential information which found their way into the fraud engine which was used to generate the fraud scores within the computer programme which were then applied to the fraud score table with the 3 lines and 2 columns.

Mr Hough’s first attack on this claim is that it is not pleaded. I agree. What is pleaded is confidentiality in the fraud score table in the open banking specifications. What is relied on is information provided to Marshmallow which (it is said) was fed into the AI exercise, which was used to generate a computer model, which was used to generate computer scores, which were deployed in the table. That is a very different claim. It is no answer for Mr Howe to say, as he did, that there was input at the early stages of the exercise, and that counts. It does not count. The pleading is of one thing being Mulsanne’s confidential information; the case now made is for another thing, related but somewhat removed.

That point would often matter, but it would matter less or not at all if the case was able to be advanced and met satisfactorily (even in the absence of a correct pleading) and if Marshmallow had had the chance to meet it properly; and if there had been a fair trial of it, bearing in mind the latitude I would extend to a case prepared as speedily as this one was. However, on this occasion I am not satisfied that that is the case. The case was originally advanced (on the pleadings) on the footing that the whole open banking specification was said to be confidential. Even if it would have been understood (and I am sure it was) that that whole case would not be advanced in court (because Marshmallow would have been aware that large parts of the specification still originated in the Hiscox model, with parts substituted or bolted on), the pleading gave no idea of a focus on this particular point. The first suggestion of the current point’s being run came in Mr McCauley’s third witness statement (the first one submitted for the purposes of the trial) where Mr McCauley picked up a previous reference by AKB to Marshmallow’s advanced fraud detection processes and advanced his own contributions to those processes. He described his own efforts in identifying fraud cases in the initial and later phases. Paragraph 7.21 then says that the results of the Claimant’s fraud checks were used to build “the First Defendant’s fraud models” and describes in general terms how the issues picked up in the assessments should be incorporated into fraud models “and the associated rating (in the form of the jointly developed ‘fraud score’ rating tables contained in [Item 2.13].” He referred to meetings between Mulsanne’s head of fraud (Mr David Taylor) and Marshmallow at which “general learnings and specific cases” were discussed. Mr McCauley’s confidential annex then listed 6 particular factors which were particularly relevant to identifying potentially fraudulent applications.

This evidence was met by a witness statement of AKB which challenged what he saw as an attempt to establish that Marshmallow’s counter-fraud skills were somehow attributable to Mr McCauley or the Mulsanne team, but if the remainder of the paragraph is a challenge then I do know what it is because it is redacted for reasons I do not know. Dr Holliday’s 5th witness statement also deals with fraud-related matters but is not really focused on what is now said to be the confidential information that was imparted.

I am not satisfied that there has been a fair trial of this point notwithstanding the absence of a pleading. The confidential information turns out not to be the table in the specification but some prior information which was fed into an artificial intelligence engine which produced some scores which led a human being to fix the score levels and ratings in the table. That chain shows the distance between the information and the pleading, which clearly falls outside the principles in Shenzhen. If the claim had been pleaded so as to reflect what Mr McCauley’s evidence came to be there may well have been a fuller investigation from the defendants’ witnesses into how Mr McCauley’s alleged input actually affected the input to the machine learning exercise. That would be relevant to an assessment of how much in reality of Mulsanne’s information was fed in and what impact it had. It might well have thrown a clearer light on whether it fell to be treated as confidential information, or confidential information of Mulsanne, in the fuller circumstances which would have become apparent. The actual machine learning process and how it worked was not wholly clear on the evidence I heard, with the experts having to surmise as to what it actually did. That is hardly satisfactory. I rule that the absence of a proper pleading means that this point cannot be taken.

Having said that, I consider that the claim would be likely to fail on the evidence that I heard anyway. It turns on whether the information as to cancellations and the factors involved amount to confidential information, which in turn is very much affected by the circumstances in which the information (if that is what it is) was provided. What is apparent from the evidence is that when Mr McCauley (and then Mr Sagoo) provided input on cancellations it was not done in circumstances which attracted confidentiality. It was not done so that the material could be put into a fancy computer programme. It was done so that an assessment could be made as to whether to maintain a policy or not. It was not (on the evidence) intended to generate a decision as to whether a policy should be cancelled. It was in effect a recommendation, or perhaps even an instruction, to look at the facts and carry out further investigations, sometimes with a clear steer as to what the result was expected to be. The three instances of communications passing between the parties on this topic reflect that. They do not reflect cancellation decisions being taken by Mulsanne, and then communicated for purposes which attracted confidentiality in the communication. So far as those communications also reflect Mulsanne’s views as to relevant factors, they were being communicated for the same reasons, and in any event I do not think that Mulsanne has made a case for saying that those sort of communications, for that reason, were particularly within Mulsanne’s expertise, as opposed to being within the expertise of both parties, which some of them are likely to have been. It is true that Mr McCauley gives evidence of particular meetings at which it is said that discussions about fraud took place with a view to assisting in the creation of fraud models within the overall model, but these are not particularised, let alone all linked to the particular table under consideration, and without clearer evidence I do not consider that it has been established that those matters were of a particularly confidential nature per se, or communicated for purposes which should be treated as confidential. I think that no more has been established than that there were discussions for furthering a particular aspect of their joint enterprise, and particularly for the protection of Mulsanne as the insurer, which is not enough. Again, I consider that some of the matters under discussion will have been matters which were common knowledge.

For those reasons, therefore, I consider that the case actually advanced under this head is not pleaded and cannot formally be advanced, and in any event has not been established on the evidence. It therefore fails.

The Marshmallow 2021 models (both January and April) contain tables reflecting the credit scores of individuals. The tables are lists of credit scores in one column, starting at zero and then jumping to 535 and continuing to 655 in increments of five in one column, and in another column, against each of the values for credit scores, there is a relativity to be applied to the applicant according to the credit score. The credit score is a figure reflecting apparent creditworthiness, supplied by a third party. Thus the table enables Marshmallow to load or discount from the base premium by the multiplier depending on the credit score of the applicant.

Apart from the last 4 rows, where the relativities differ, these tables are the same as those being used in the Marshmallow/Mulsanne enterprise at the end of 2020. Mulsanne claims that the figures in the table are its confidential information, or rather, according to final submissions, joint confidential information, which Marshmallow is not allowed to use without Mulsanne’s consent. It is said that Marshmallow and Marshmallow Insurance’s use in the new enterprise was a wrongful act which amounts to direct (mis)use of confidential information.

The first question is whether these figures form part of Mulsanne’s confidential information, or confidential information to which they have a joint entitlement. I consider that they do not.

Marshmallow did not produce a witness to show where these figures came from, but the clear weight of evidence before me shows that they came from Marshmallow and that any contributions from Mulsanne were not such as to give them any confidential entitlement to them. In a table prepared by Mr Gray as part of his analysis of which parts of the engines needed to be purged of Mulsanne-derived material, he records that this table contained figures proposed by Marshmallow and agreed by Mulsanne. Other evidence supports the view that they came from Marshmallow. When this table was in the course of preparation Marshmallow prepared a presentation for Mulsanne in August 2019 about the credit score rating. It extols the virtues of their creating “our own bespoke score using the underlying credit concepts”. It goes on to carry out various analyses which demonstrated that there was undercharging of lower credit scores and overcharging of the higher ones (the words obviously intended to refer to the higher ones have been accidentally omitted from the presentation document but the meaning is clear enough). The presentation proposes a credit score table which grouped scores into bands of 5 (as ultimately happened) and loading from 1.19 (in the accompanying table) at the highest to 0.8 (a discount) at the lowest. Again, this is basically what happened.

Mr Sealy (under the name Anton Leacock) of Marshmallow, who prepared this presentation and the figures, invited Dr Holliday and Mr McCauley to a meeting to discuss these proposals. The meeting took place and they were agreed as reflected in a Slack message from Mr Sealy to AKB of the same date. The figures were, it seems, put into operation, as Mr McCauley confirmed.

In January 2020 Mr Sealy proposed some modifications to the table so as to improve the position for the higher credit score owners. He prepared a presentation dealing with that, and with other proposed tweaks to the scheme. That was discussed with Mr McCauley who sent an email saying he broadly agreed with the proposals, but in cross-examination he expressed some doubt as to whether these particular proposals were implemented - he was not sure. There was no clear evidence one way or another. There are suggestions that later tweaks to this table might have been carried out, but if they were there is no evidence to suggest anything other than that Marshmallow carried out the amendments.

In his evidence Mr McCauley agreed that Marshmallow provided all the figures, but he also relied on the fact that they were discussed and agreed with Mulsanne and that the idea of a credit score table had been the subject of discussion before the first of these presentations; and indeed that Mulsanne had proposed a credit score table in the first place. Mr Howe relied on those matters plus Mulsanne’s participation in discussions about the actual figures as giving Mulsanne at least a joint entitlement to what he said was confidential information.

I am prepared to accept that the table is capable, bearing in mind its general nature and content, of attracting confidentiality, but I do not consider that even if that is right, Mulsanne has any confidentiality rights in it. The idea of a credit score table was a familiar one in the industry, as Mr McCauley accepted, so no confidentiality could be claimed in relation to that. Mr Howe needs to establish a claim to confidentiality in the actual table, and the actual figures in the table. There was no suggestion that Mr McCauley or Mulsanne contributed actual figures. The most that happened was discussion and approval. The discussion might have gone to the actual figures, but it still remains the case that the structure and the actual figures, or most of them, came from Marshmallow. It is not actually possible to identify any particular piece of material that was contributed by Mulsanne, much less that any information was imparted and received under an obligation of confidence. Nor do I consider that the process of consultation and approval that went on made the information one in respect of which there is joint confidentiality. It simply does not amount to a sufficient contribution to make the information in the table in any meaningful sense Mulsanne’s. It is accepted that elements of the engine that were Marshmallow’s creation were not confidential information to which Mulsanne could claim an entitlement, and this table falls into that category.

This part of the claim therefore fails.

So far as this price specification contained duplicate tables which I have held to be available for springboard use from the general specification, then those tables are similarly available for springboard use at the same time (in autumn 2020) insofar as those springboard events involved using the whole engine; but it is hard to see how this can add anything material to the springboard events in question.

These claims are designated “springboard” claims by Mr Howe (whether or not that is quite the right use of the word in this context) to distinguish them from “ongoing” or “direct” claims in which it is said that particular parts of the engine or rules were directly used (essentially copied) in the new Marshmallow Insurance business in 2021. The springboard claims are claims that Mulsanne’s confidential information was used for the purposes of preparing for and advancing the Marshmallow Insurance enterprise albeit that it was not then thereafter directly used in that enterprise. Thus, for example, it is said that the ratings engine, or parts of it, were produced to reinsurers during the attempts to put reinsurance in place; and, by way of another example, that the engine (containing Mulsanne’s contributions in which confidentiality is claimed) was run to generate sample quotes in order to carry out certain checks of and comparisons with Marshmallow Insurance’s own models, thereby misusing Mulsanne’s contributions. In that way Marshmallow Insurance was enabled at least to speed up the process of putting in place important elements that had to be in place before the new business could be started. That is the essence of these claims.

Mr Howe’s final submissions focused on three particular instances in which there was said to be springboard use in a relatively limited fashion compared with his original case. His original case asserted that there was a more basic use in that it was asserted that the engine, including parts said to contain Mulsanne’s confidential information, was submitted to potential reinsurers (and the ultimate quota reinsurer) and to regulators so that those bodies could be satisfied about how the business would operate. This particular assertion did not materialise in Mr Howe’s final submissions, though those submissions did not expressly abandon the case just described. For the sake of completeness I record here that Marshmallow denied the sort of use just described and I accept that evidence. Marshmallow said that it was not required to, and did not, submit the engine in the manner just described, and I find that it did not.

I shall therefore confine myself to the allegations which survived into the written final submissions.

This point arises out of the answer to a question posed in a questionnaire from the then proposed quota reinsurers (Scor Re), who in fact became the actual quota reinsurers to Marshmallow Insurance. It is said that in answering that question Marshmallow ran the whole of the engine then used in the Marshmallow/Mulsanne business and thereby misused such elements of it as were confidential.

The facts are said to be (and, save where the contrary appears, I find them to be) as follows. In anticipation of their new venture, at the beginning of 2020 the participants in Marshmallow sought to set up quota reinsurance. A quota reinsurer takes on a portion (quota) of the overall risk and receives that portion of the premium, and (at least in this case) pays a fee back to the first insurer. In the present case the plan was that Scor would take on 85% of the risk for 85% of the premium (and paying a fee back). Negotiations were conducted with Scor between January and April 2020. A number of matters were discussed at a high level, but Scor did not ask for the ratings engine. Reinsurance was finally arranged with Scor in December 2020.

In the course of this exercise, and probably in February 2020, Scor submitted a questionnaire with a number of questions, including a Question 1(a), described in its index as:

What then follows is a completed spreadsheet or spreadsheets, with figures completed by someone on behalf of Marshmallow (probably Mr Sealy), and some explanations in text boxes below. The first table shows figures for, inter alia, net premium income for the years 2018 and 2019 (including an original estimate for 2019 which was superseded by the actual figures which also appear) , average premium and loss ratio. There are also estimated figures for 2020, on which this particular part of the claim turns. In a further table there are figures for net changes in average premium for those years, and then a “rate change” column for the two years 2018 (which will have been a part year because the business started half way through that year) and 2019.

In one of the text boxes Scor asked this question:

The case of Mr Howe is that that exercise was achieved, at least in relation to the 2019 Rate Change, by taking the policies written in 2019 and running them through the claims engine as it stood in 2020 (excluding some policies as Mr Sealy explained). (It was not apparent how a figure could have been obtained for 2018.) In running that engine there was a misuse of Mulsanne’s confidential information insofar as the engine at that time contained such information. That sounded in damages, which would have to be assessed at the separate inquiry as to damages for which directions had been provided.

Mr Hough’s first response was that that case was not pleaded and it ought to have been. It was a significant matter and had the matter been pleaded evidence could properly have been directed to precisely how it was that the figures were arrived at. As the matter developed at trial, the only evidence given was by Dr Holliday who could only surmise as to how the figures were reached. In addition, Marshmallow could have sought evidence from Scor to the effect that the effect of this figure was immaterial so any breach of confidence was similarly immaterial and not appropriately taken forward to any next stage in this action. He made some further general observations on the flood of evidence and/or disclosure that would have been appropriate and at least considered, if not obtained, had the matter been pleaded.

So far as the merits were concerned, on the assumption that the point could be run, he submitted that there was a mystery as to how the figures were in fact arrived at (the mystery as to the 2018 figure infecting the 2019 figure). Furthermore, he disputed that running the engine (if that is what happened) amounted to a misuse of confidential information because the output, namely a set of deemed premiums, could not conceivably be confidential because a premium quotation is presented to anyone who inputs responses into the relevant website.

I find that the pleading point fails. Technically the complaint is really one about absence of particularisation rather than a complete absence as to pleading. The last sentence of paragraph 40 of the Re-amended Particulars of Claim reads:

There never was a detailed pleading of this particular point, but there was at least a strong pointer in Mr Howe’s Springboard Note (see above) which he provided at the beginning of Day 2 of the trial. In paragraph 3 of that document Mr Howe referred to “preparatory steps relying on Mulsanne’s Confidential Information to set up its competing business.” It gives reliance on the Scor document as an example :

That is not quite the allegation that came to be put, but it is reasonably close. When this document was produced to the court at the beginning of Day 2, Mr Hough himself had only just seen it. He said he would consider it but was not asking me to do anything about it at that point. I myself indicated that if it was necessary for Mr Hough to produce additional evidence in chief then I would be sympathetic to that, and Mr Hough seemed to accept that as a possible expedient. He also accepted my preliminary view that it would not apparently lead to a need for any further disclosure. Mr Hough did not thereafter expressly raise any clear pleading point about this document until he took the point in his written final submissions. In his opening (before the Springboard Note had been produced) Mr Hough dealt with the point very shortly saying:

It would therefore seem from that that Mr Hough was prepared, at that point, to deal with the point. The matter was put to Dr Holliday and AKB, and Mr Hough re-examined AKB on it briefly.

There is one further point which needs to be taken in consideration in determining whether the failure to identify the point more clearly in a pleading should lead to its not being admissible. It is a point about which the claimant cannot have known until shortly before the trial. The overall document of which Question 1(a) was part had been disclosed by Marshmallow as part of the disclosure in the case, but the parts relevant to the present point had been redacted in such a way that the point now taken would not have been apparent to Mulsanne. I ordered the removal of those redactions (and others) at the pre-trial review on 29th October 2021, with the unredacted document to be provided by 2nd November. That was less than 14 days before the projected start of the trial, and even fewer days before the case had to be prepared to skeleton argument stage, so the claimant did not have a lot of time to consider the document and identify the point. Strictly speaking it ought to have been pleaded out, but those circumstances make it more understandable that it was not.

In all those circumstances I do not consider that the absence of a pleading should debar Mr Howe from taking this point. The lateness of the taking of the point was caused by a redaction which obscured the point from view until close to the trial. It was opened clearly enough for Mr Hough and his clients to understand the point at least at that stage, and he had the opportunity to meet it with additional evidence if chief if he wished to (but he did not seek to adduce any). If he had raised his final submissions on additional disclosure earlier then that point could have been dealt with during the trial, and he seemed to disclaim any need for disclosure anyway. All in all I think that the point can be dealt with fairly to both sides, and I shall do so.

I find that it is probable that Mr Sealy did indeed use the engine in the manner described by Mr Howe, to produce a lot of quotations which could be compared with the premiums charged in the earlier year so as to produce the rate change figure. Dr Holliday did not himself know how the figure was produced, and said he would probably have done it differently (by just adjusting for differences which had arisen in the intervening period rather than running whole quotes) but he was minded to accept that Mr Sealy ran the whole quotation exercise. I agree. The description of the exercise in the text box, quoted above, seems to suggest just that.

The engine at that stage contained some elements (very few) which can be categorised as being Mulsanne’s confidential information, namely DM-2.4 (claims and conviction table), DM-2.5 (vehicle segmentation file), the voluntary excess ratings table, and DM-2.10 (2019 postcode file), albeit that I have found the confidentiality level of some of those items to be low. The next question that arises is whether that information was misused when it was used as part of the source of the engine whose handle was turned (metaphorically speaking) to produce the range of alternative quotations which generated the single rate change figure for 2019. Mr Hough’s answer to this question was No, because the end result of the exercise is sample quotations which are not confidential because anyone can produce identical quotes by going on to the aggregator’s website and producing the same quotes, and it is the quotations that are used, not the underlying confidential information.

I disagree with Mr Hough’s reasoning and conclusion. He looks at the wrong part of the process. The act of producing the quotations, which are the desired end result of the exercise, is what is said to be the use of the information, and I agree that it is. Although the information is not actually disclosed to anyone, it is used as part of the engine whose handle is turned for a purpose for which it was not disclosed. It was used for the purpose of Marshmallow seeking reinsurance for a separate venture. Although no-one produced authority to me on the point, I consider that analysis to be obvious. If the actual engine (created by software engineers from the information in the tables contributed by Mulsanne and Marshmallow) had not existed at the time of the dealings with Scor Re, and Marshmallow had first created that engine from the tables contributed by both parties, and then turned the handle to produce the figures which produced the rate change figure, then the act of using the confidential information to create the engine for that purpose would plainly be a misuse. On the facts, Marshmallow did not have to go through the step of creating the actual software engine at the time because it already existed for the purposes of the Marshmallow/Mulsanne business, but that seems to me to make no difference. It merely means the use is more indirect, but it is nonetheless use because the engine and its contents are being used for a purpose which is not Mulsanne’s and to which it cannot be assumed that Mulsanne would have consented or had already given implied consent - there was no real evidence of an industry practice which would have justified the activity despite the short attempts of Dr Holliday to suggest otherwise. This conclusion stands even though the bulk of the material for the software engine was not Mulsanne’s but was derived from Marshmallow’s Hiscox model.

I therefore consider that this springboard claim is made out. Marshmallow made indirect use of Mulsanne’s confidential information, albeit indirectly, for a purpose outside the scope of the purposes for which Mulsanne provided it, as an act preparatory to Marshmallow’s own enterprise. Whether that use is of any real significance in the case is a different matter.

Mr Hough submitted that Marshmallow had not made out a case of materiality, so this part of the claim should fail and should not go forward for a further inquiry as to damages (or other remedies). As part of his submissions he suggested that it can be seen at this stage that the figure cannot have made any real difference to Scor’s decision to provide re-insurance, or to have hastened the end result in any material way.

This submission causes me serious concern. There seems to me at present to be much to be said for this point. The figure is only a very small part of the information provided to Scor, as far as I can see, though there was no analysis of the rest of the information provided (and indeed the totality of the information was not put before me). It is not at all easy to see how this single figure, provided as part of a mass of other information (the nature of which was outlined in general terms in Dr Holliday’s 4th witness statement) can have played a material part in Scor’s deliberations, much less have hastened the reinsurance arrangements bearing in mind the time lapse between February 2020 and the reinsurance being in place at the end of the year.

Notwithstanding that, I do not think it right, at this particular stage, to say that there is a sufficient absence of a case on causation or damage that the point should not be available to Mr Howe at the next stage of this action. It might well be said that Mr Howe has not established “causation”, but I consider that, for these purposes, any question of causation should, by and large, go forward to the “quantum” phase of this action. This is a point I develop below in a separate section when indicating the way forward towards any “quantum” hearing. It is very much to be hoped that the claimant will take a realistic view of the strength of this particular finding of misuse and consider whether it really is something worth taking forward. I have not made up my mind on the point, but I think I can fairly say that as a separate act of misuse it is not readily apparent what the pathway would be towards any worthwhile remedy in respect of it.

This springboard claim arises out of the creation of a new postcode file created essentially by Mr Hemming on the instructions of Marshmallow in mid-2020, and in particular the activity of what was said to be comparing it and checking its viability and sense against a different Mulsanne-derived postcode file. It is the comparison exercise which is said to give rise to a springboard claim. That this was the point was something that emerged only by the end of the trial. It is not pleaded as such, and not clearly foreshadowed in Mulsanne’s opening skeleton argument, though Mr Rourke’s second report seems to advert to the point.

The central facts are as follows. In about April 2020 Dr Holliday decided that the postcode file then in use, in its then form, required re-doing. It had acquired a number of overlays and it appeared to Dr Holliday that it was no longer the file that should be used. He instructed Mr Hemming to work on a new postcode file and over the next month or so Mr Hemming did so, with some input from Dr Holliday. This file was not created purely for use in the proposed new Marshmallow business. It was produced for use in the Marshmallow/Mulsanne business, even if there was an eye to using it in the new business as well. I do not consider a final decision had been taken about that.

This new file was not produced using any Mulsanne file as a starting point. Mr Hemming produced it using market data and his own expertise, machine learning and other openly available data. I think at one stage there was a suggestion that Mulsanne’s data was used for this purpose, but by the end of the trial it does not seem that that allegation was pursued any longer, and I find as a fact that such data was not used. The overall file was Mr Hemming’s own creation using his own application called MappaR (since re-badged Lucidum). All this was done, as Mr Hemming correctly understood, so that the file could be presented to Mulsanne with a view to its being adopted in that business. The activity of the creation of this file in no way used Mulsanne confidential information.

Once the file was created some comparison exercises were carried out to see how the resulting relativities compared with other indicators. One of the things with which a comparison was drawn was a Mulsanne postcode file. As seems to be typical of postcode files in this case, that file was again surrounded with a bit of (again probably unnecessary) mystery. Mulsanne believed that that comparator file was Mulsanne’s file then in use in the business. Mr Hemming denied that, but his evidence was nonetheless that the file had been provided by Mr Dragne of Mulsanne. That could have led to a pleading-type point (because Marshmallow had conducted its defence of this aspect in part on the footing that Mr Hemming had not used the then in-use postcode file in the comparison exercise) which Mr Howe sought to cover off by a late application to adduce further evidence after both sides’ evidential cases had been closed. That, and some further argument on the point, was avoided (after significant debate in final submissions) by Mr Hough’s acceptance for the purposes of the argument, that Mr Hemming’s comparator was a Mulsanne-originated file. The consequence of that is that that file (whatever it was) falls to be treated as confidential information of Mulsanne; Mr Hough did not seem to dispute that consequence.

In a spreadsheet of information presented to Dr Holliday, Mr Hemming presented various comparisons involving the relativities generated by his new postcode file. In three columns of the spreadsheet he compared what he understood to be the relativities in the currently used postcode file , his new relativities, the Mulsanne file relativities and the old Hiscox historic relativities. Then in further columns he compared each of his MappaR relativities and the Mulsanne file relativities with the currently used relativities. The details of those comparisons do not matter. What matters for these purposes is that the Mulsanne file (whatever it was) featured in the comparison exercise.

It does not appear that Mr Hemming was instructed to carry out that comparison exercise. His evidence strongly suggests that he did the comparison exercises because he understood the material would be presented to Mr McCauley and that the latter would be interested in such things. He believed that because at the beginning of the Marshmallow/Mulsanne business he had been interested in a comparison of postcode tables which was demonstrated by heat maps. He thought it would be helpful to both Dr Holliday and Mr McCauley to present this information.

On 11th June 2020 Dr Holliday emailed Mr McCauley about three significant changes that he had been looking at so as to give rise to a “rating refresh”. His email is a long one which sets out and seeks to justify that he was proposing to refresh the occupations tables, the postcode tables and a change to the base rate. So far as the postcode tables were concerned, he attached what he described as a “summary spreadsheet”, promising to send the full proposed file under separate cover to avoid the risk of bounce-back being caused by its size.

On 16th June Mr McCauley responded:

On 17th June Dr Holliday replied by sending two spreadsheets – one for occupation groups and one for postcode areas. It seems that the latter would have been a version of Mr Hemming’s spreadsheet, with some modifications to the figures which Dr Holliday wanted (“rebalancing”), but it is also apparent that the columns containing the comparisons, and the material for them, have been omitted. Dr Holliday admitted that he did not supply those comparisons. His explanation was that he was embarking on a negotiation with Mr McCauley about the three changes that he proposed, and wished to hold something back in the negotiation in case he needed it. In the event he did not need it. There was a debate about a dozen postcodes, and although Dr Holliday did not agree with Mr McCauley it was agreed that they would take the higher of the two relativities in those postcodes. In due course, as I understand it, that postcode file was put into operation in the Marshmallow/Mulsanne operation. When the Marshmallow operation started up the same file was used but with Mulsanne’s amendments removed.

It is not suggested that use of that postcode file by Marshmallow was of itself a misuse of Mulsanne’s confidential information. Mulsanne does not claim that file as its confidential information. What it complains of is the use of the actual Mulsanne file (supplied to Mr Hemming) in the comparison exercises that were carried out. In the Springboard Note (when adjusted to substitute a reference to the Mulsanne file for a reference to DM-2.10) it is said that:

I find that this claim fails for the following reasons. No complaint is made about the process of creation of the new postcode file. It was plainly intended by Marshmallow that the file should be proposed to Mulsanne for the purposes of their venture. Dr Holliday actually wanted that to happen. I find that that was the real motivation behind the decision to revisit the postcode files. Dr Holliday had his own firm views as to the need for a new file because he was unhappy about aspects of the current file, which had acquired too many additions for his liking. I find that the comparisons were done for the purposes of their being shown to Mr McCauley, even if at the time it originated as Mr Hemming’s purpose only. That became Dr Holliday’s purpose, even if he resolved he would only show the comparisons if asked. The comparison was not done to get confirmation or comfort that the new MappaR relativities were in line with other ideas for postcodes. I do not think that Mr Hemming would require such comfort - he was satisfied with his analytical work which was designed to produce something better than pre-existing efforts. It may be that the intended deployment of the comparison would give Mr McCauley some comfort but I do not consider it likely that Dr Holliday or Mr Hemming considered they themselves needed such comfort. Accordingly, the comparison was nothing to do with the intended new Marshmallow business. It was for the purposes of the existing Marshmallow/Mulsanne business.

Mr Howe sought to support his case as to their being an illegitimate purpose by relying on the removal of the comparison columns as being somehow sinister in this respect. I do not consider that that analysis works. It makes little sense. Dr Holliday wanted to get this postcode file adopted into the Marshmallow/Mulsanne venture. If he thought that the columns would be useful to achieve that he would have to be prepared to advance them if necessary. There was no benefit to his purpose in withholding them in all circumstances, and even if he had in mind the future use of this file in his intended new venture there would still be no benefit to that venture in having had the comparisons but still withholding them from Mulsanne. If the comparison was helpful then it would be helpful to the new postcode file whoever was thinking about deploying it and, to repeat myself, he wanted the Marshmallow/Mulsanne venture to adopt the file. For these purposes there was no sinister import in his withholding the columns. I accept Dr Holliday’s explanation that he regarded himself as in a negotiation (his interest as a broker who took commission was not the same as the insurer who took on the risk), even if I do not quite understand this particular negotiating tactic.

I therefore do not accept that, as a matter of fact, the comparisons were carried out so that Marshmallow could gain some comfort, or carry out some confirmatory cross-checking, in relation to a file to be adopted in the new venture. The postcode file was to be adopted in the existing venture; the purpose of carrying out the comparison in the first place was Mr Hemming’s, who had nothing to do with any future venture when he carried out the comparison; the purpose was to satisfy Mr McCauley, and it was not to bolster Mr Hemming’s figures or prepare the way for the new Marshmallow business. Accordingly when the Mulsanne file was used as part of that exercise there was no misuse of it.

In cross-examination it was put to Dr Holliday that the motivation for the new postcode exercise was at least in part to increase premium revenue in order to make the existing business look more attractive to proposed investors in the new one. Dr Holliday denied that and said that any broker would want to increase business. The postcode and occupation files were revisited because concerns had been expressed about loss ratios and Dr Holliday himself was concerned about some of the overlays which had been imposed on the postcode file, so the existing business itself required something to be done about the postcode file. I accept that evidence.

It follows that Marshmallow has not made out its case that the comparison exercise carried out by Mr Hemming was one which amounted to or involved an improper use of Mulsanne’s confidential information. It was a proper use when the comparison was done, and the use was not the cross-checking comfort-affording purpose that Mr Howe relied on. The fact that the same postcode file with Mulsanne additions removed was then used in the new Marshmallow Insurance business does not make the comparison exercise involving Mulsanne’s file a misuse of confidential material. By then it had become immaterial, even if it was ever a material consideration (which Dr Holliday denied it was - he said it would not have made any difference to his view of the effect of the file).

In his written final submissions Mr Howe also added a reference to a “rebalancing” exercise which was carried out before the new postcode file was put to Mr McCauley. Mr Gray of Marshmallow looked at the figures which resulted from Mr Hemming’s MappaR exercise and considered whether and how they should be adjusted to iron out some differences between regions and preserve price competitiveness. The details of this do not matter. The point was discussed with Dr Holliday. Mr Howe’s submissions suggested, without development, that this too involved a misuse of Mulsanne’s confidential information in terms of its postcode file.

I do not see how this adds anything to his case. The exercise involved applying existing market experience to Mr Hemming’s figures and carrying out adjustments. That does not involve an actual comparison with a Mulsanne file, whether an existing in-use postcode file or the file provided by Mr Dragne. It was part of an exercise which was legitimate as between Marshmallow and Mulsanne to come up with an end result which was to be provided to Mr McCauley. This seems to me to add nothing to Mr Howe’s case.

The claim here is that when Ms Coulthard was preparing some of her elements of the new Marshmallow Insurance engine she received some assistance from Mr Gray of Marshmallow who ran a number of sample quotes through the then Marshmallow/Mulsanne engine in order to produce analyses which informed Ms Coulthard’s work. In doing so it is said that Mulsanne’s confidential information was used insofar as the engine at that point contained such information. These are therefore said to be further instances of the “turning the handle” type which has already been referred to. There are (now) said to be four such instances.

The first concerns the creation of a vehicle modifications table, within a new vehicle segmentation table which Dr Holliday instructed Ms Coulthard to create for the ratings engine in place of one which was, or was thought to be, Mulsanne’s. Part of that table was a vehicle modifications table - a table containing relativities to be applied to categories of vehicle modification. An email from her dated 27th November 2020 to Mr Gray indicates that she had formed her own her own views which were that there should be 5 groups, numbered 1 to 4 and then 99, in which the relativities or loading were to be no load, adjustment 1.1, 1.2, 1.3 and decline respectively.

However, she also asked Mr Gray to run an analysis to provide a “preliminary indication” of what loads or decline decisions should be made. On 23rd November 2020 he responded to her, describing the exercise which he had carried out and his results:

It is not easy to understand the nature of the analysis which Mr Gray carried out, but that particular piece of detail does not matter. What is clear enough is that he ran market available quotes through Marshmallow’s then engine in some way in order to ascertain what loading factors would keep Marshmallow Insurance’s prices competitive. He was “turning the handle” of the engine. Mulsanne’s case is that that amounted to misuse of its confidential information so far as it was incorporated in the engine at the time, and that that use gave Marshmallow an advantage in a springboard sense when it came to launch its product.

Ms Coulthard in due course prepared a table. It was not used in Marshmallow Insurance’s January 2021 model but it was used in the April 2021 model. In cross-examination she said that she discussed the figures with Mr Gray and could not remember whether she agreed with his recommendation, or whether she used her original figures. The figures used in that model for the first 4 groups were those proposed by Mr Gray in his email (that is to say the loading was 1, 1, 1.21 and 1.27 across each of Groups 1 – 4). Mr Howe submitted that Mr Grey’s analysis appears to be “causative” of the final form of Marshmallow Insurance’s table.

Mr Hough’s first point on this part of the case was that yet again the matter was not pleaded. He is (again) right about that so far as a specific pleading is concerned, though there was a reference in the Springboard Note to use of the model as at 2020 when creating the vehicle modification table. In this instance again I would not allow the pleading point to stand in the way of the claim, extending (again) the latitude that a speedy trial sometimes requires. I do not think that Mr Hough has been materially disadvantaged. In her own witness statement Ms Coulthard refers to having asked Mr Gray:

She therefore herself raised the question of Mr Gray’s analysis, and it is then clear enough from Mr Gray’s email what he did and that he used the then engine to run the quotes. Ms Coulthard’s clear understanding was that he had used the engine for that purpose. I consider that Mr Howe should be allowed to rely on this incident so far as the facts of use are concerned.

Mr Hough’s next point is the same point that he has run before – the quotes obtained are not confidential therefore there was no misuse of confidential information. The answer is the same as that which I have already given – running the engine, which contained confidential information, is capable of being, and in this case is, a misuse of that information because it is a use which is outside the scope of that for which underlying confidential information within it was provided.

Next Mr Hough says that the ratings factors produced were in line with the figures that Ms Coulthard would have produced anyway based on her own judgment. The only difference which exceeds a rounding effect is a 0.1 difference for Group 2. Bearing in mind that this is an effect for only one group of a subset of insured persons with modified vehicles, this is a minimal effect. No springboard advantage was obtained – Ms Coulthard could have arrived at the figures with her own judgement or Mr Gray could have made estimates of how to maintain the competitiveness of Marshmallow prices by other means.

There is much more force in those submissions. Mr Gray’s recommended figures were indeed very close to those of Ms Coulthard and if one imagines what would have happened if he had not done the exercise and provided those figures it is reasonably apparent that she would have come up with the same figures or something extremely close. Furthermore, there seems to have been no time advantage in the exercise because Mr Gray’s figures, derived from his exercise, were not utilised (if they were utilised) until the April 2021 model, so the process of getting Marshmallow’s Insurance business off the ground was in no way accelerated or assisted by this particular exercise. Nonetheless it is apparent that some comfort was taken from Mr Gray’s exercise. It may be slight but unless I conclude it to be trivial (which I do not) it cannot be ignored at this stage of the exercise.

Next Mr Howe relies on the same mechanism having been adopted when Ms Coulthard came to create a voluntary excess table - a table which allowed a discount depending on the amount of a voluntary excess that an individual was prepared to assume. Ms Coulthard’s witness statement says:

There is no suggestion there that Mr Gray carried out a similar exercise involving “turning the handle”. There is nothing in the disclosure on this point similar to that which seems clearly to demonstrate what Mr Gray did in relation to the vehicle modification table. Mr Howe’s case rests on a single question and answer of Ms Coulthard in cross examination:

Mr Hough’s objections to this claim are the same as those in relation to the vehicle modification table. By a very narrow margin I allow it to be run. While it is not supported by a clear email such as Mr Gray’s just referred to, the table is at least mentioned in the Springboard Note, so Marshmallow could see it coming. While the matter was not investigated beyond the single question and answer just set out, that is evidence which ties in with what Ms Coulthard said in her witness statement even if it is not wholly foreshadowed by it, so she had, to that extent, already embarked on the point (albeit not in face of a pleading).

However, Mr Hough makes a more compelling point when he says there is no serious case which can be made for there having been any springboard effect from the exercise in this instance. Ms Coulthard’s unchallenged evidence was that she set the rates in this table (which was a single table within a very complex overall table within a complex ratings engine) according to her own experience. It’s very hard indeed to see how Mr Gray’s exercise, even if carried out, can have had any material effect on the light of that. It seems to me that any misuse was temporary, not deliberate and likely to be insignificant in the overall context.

Next Mr Howe advanced a similar case in relation to two other tables - NCD (no claims discount) and NCD protected tables. Ms Coulthard said in her second witness statement:

Again in cross-examination she confirmed her belief that Mr Gray conducted the same exercise of running quotes through the engine.

This time these two tables are not even mentioned in the Springboard Note. On this occasion I would accede to Mr Hough’s pleading point. One would normally expect some particulars to be given of a misuse which is going to be alleged, and despite the latitude to which I have extended to a claimant in a speedy trial case, and particularly in a misuse of confidential information case where details of the misuse may well not emerge until disclosure, there are limits to that latitude and in this instance they would be exceeded if I allowed the point to be taken. This point is not foreshadowed in any document and emerged only as a result of a short question in cross-examination. Ms Coulthard did not conduct the exercise herself, and there is no disclosure equivalent of the email which we have in relation to the modifications table which makes it clear what has happened. I would not allow it to be advanced. If these points had been added to the last two in some sort of pre-trial indication then it may be that the defendant would have wished to adduce some evidence in relation to it, or given some more disclosure. I would not allow the point to be run.

For the sake of completeness I record that the Springboard Note suggested that the “Marshmallow Vehicle Segmentation File (vehicle groupings table)” fell into the same category of claim as the above, but no case was advanced in relation to this in final submissions. In any event Ms Coulthard’s clear evidence was that she prepared the vehicle segmentation file without reference to the existing engine (as she was told to do, in order to avoid challenge by Mulsanne). I accept that evidence.

It is necessary take this point at this stage because of a residual dispute as to what parts of Mulsanne’s confidential information were in the engine as at November 2020 so as to be capable of being misused when the handle was turned. I have dealt with some items above in dealing with the list of items in paragraph 536 of Mr Howe’s skeleton argument. In his skeleton argument section on the springboard claim Mr Howe draws attention to four other parts of the engine which he says were added as part of the open banking specification. He submits that they should all be considered as confidential information which was misused when the handle was turned in the springboard incidents to which I have referred.

None of them qualify as confidential information which was misused for these purposes. They are as follows:

Under this heading Mr Howe raised various incidents in which Mr Gray sought to make the engine tables available to Ms Coulthard, or on which it seemed he was prepared to use the engine to check pricing (other than the incidents referred to above). It would seem that these activities were not offered with the culpable intent of deliberately using confidential information. It would seem to be more likely that it was not necessarily appreciated that use of the tables might (but would not necessarily) be inappropriate in Ms Coulthard’s work (depending on what she would have done with the information). When links to the tables were sent to Ms Coulthard she could not access them because she did not have the relevant access permissions, and I accept her evidence that she did not see or use them.

Mr Howe accepted that there was no evidence that any of these incidents led to actual misuse on the evidence, but somehow still sought to say (in his written submissions) that “these were acts of disclosure or at least attempted disclosure for purposes outside the TOBA”. I find that they were patently not acts of disclosure, and even if they were attempted acts of disclosure they did not succeed and therefore do not constitute a cause of action within these proceedings.

So that the terms are gathered together in one place for ease of reference, and so as not to encumber this part of the judgment with that sort of detail, I have set out the relevant terms of the TOBA in an Annex to this judgment. I will in due course set out those clauses out of which issues directly arise when I come to consider those issues. Others are there to provide context for questions that arise in this case.

The expressed effective date of the TOBA is (as on the front sheet) 11th May 2018. It is designed to create and govern an agency agreement under which Marshmallow was to market and sell Mulsanne’s policies on Mulsanne’s behalf, and while it creates an agency in relation to the receipt of premiums (clause 10), it does not expressly confer authority to enter into policies and issue documentation. However, it is common ground that Marshmallow had that power and authority and, to that extent also, was an agent of Mulsanne.

The overall issues on the TOBA claim are twofold. First, when was it terminated (it is common ground that on any footing it has been terminated now). Second, whether the first defendant has been guilty of post-termination breaches of TOBA and/or general agency duties. The date of termination goes to the question of whether, at any given point in time, Marshmallow was guilty of post-termination breaches. The choice is between January or February 2021 (Mulsanne’s case), or August 2021 (Marshmallow’s case). If the true date is the latter then the scope of the alleged post-termination breaches is very much more limited in terms of time, and potentially in some aspects of liability.

Mulsanne’s first termination case is the easiest to deal with. On 20th January 2021 NRF, solicitors to Marshmallow, wrote to HL, solicitors for Mulsanne, with whom they had been in correspondence for some months, claiming that Mulsanne were in material breach of the TOBA by virtue of data protection failings and claiming to terminate under clause 19.1.5 - right to terminate immediately if there is a material breach of the agreement. In a letter of 25th January HL denied the breaches and said that NRF’s own letter was itself a repudiatory breach and reserving its client’s rights. It pointed out that if Marshmallow wished to terminate it could give 180 days notice under the varied clause 19.1.2 and went on:

On 28th January NRF responded:

Notwithstanding that, on 2nd February 2021 HL wrote purporting to accept what it described an act of repudiation by Marshmallow, saying:

Mr Christie argued that Mulsanne was entitled to disregard the withdrawal because the withdrawal was inadequate and/or ineffective because it was caveated as being “for the present time” only and because the letter maintained the position that Marshmallow had the right to terminate for material breach.

I do not accept that argument. I shall assume for these purposes, without deciding it, that Marshmallow’s attempt to terminate was indeed capable of being a repudiation. Nonetheless, Mulsanne plainly invited Marshmallow to withdraw its notice, and Marshmallow did so. In saying that the withdrawal was “for the present time” it was merely indicating that it might seek to do it again in the future. That does not qualify the withdrawal of the then notice. Its express reservation of the right to revive the point in the future is not a qualification of the presentness of the withdrawal either. There is no inconsistency in withdrawing the intended effect of a document for the present whilst indicating that it might be done again in the future. If Mulsanne is correct in saying that the termination was capable of being a repudiation, then if it happened again in the future there would be another repudiation; but for the present there is a withdrawal of the notice as invited. Essentially Mulsanne was inviting Marshmallow to withdraw its notice and Marshmallow accepted that invitation without any relevant qualification of the acceptance.

Another way of looking at the matter is that, with knowledge of an active repudiation, Mulsanne nonetheless affirmed the contract. It is a plain affirmation of a contract when one party invites the other to withdraw a notice purporting to terminate it, and the reference to the ability to terminate on 180 days notice is also an acknowledgment or affirmation that the contract is to be treated as subsisting. Accordingly, on those normal contractual principles of affirmation after a repudiatory breach, the contract survives.

Accordingly, Mulsanne has not terminated the contract on this basis.

If Mulsanne does not succeed on that basis it advances various other grounds based on breaches of other provisions of the TOBA, and further express grounds under the TOBA itself (with a large amount of cross-fertilisation between those two heads). The further grounds were set out in HL’s letter of 2nd February in the following terms (in which MICL is a reference to Mulsanne):

Some of those grounds were repeated in a further letter of 1st March 2021 in which HL relied on material subsequent to the letter of 2nd February as being the basis on which it could, so far as the TOBA had not already been terminated, terminate on the basis of Clause 19.2.7 and on the audit basis of a continuing failure to allow an audit.

I shall take those claims in a different order from that appearing above and from that which Mulsanne used in its final submissions, starting with that set out in sub-paragraph (f) - intention to transfer business.

At the outset of the relationship Mulsanne was the sole capacity provider for Marshmallow’s broking business. It was Marshmallow who approached Mulsanne to provide “capacity”; the approach was not the other way round. While nothing in the TOBA prevented Marshmallow in contractual terms from taking on more business, whether motor business or not, as a matter of fact it did not do so at or shortly after the inception of that relationship, or indeed until Marshmallow started its operation. It would not have been a breach of contract to do so - there was no sole agency relationship. However, the TOBA did contain a clause operating in the event of a competing motor insurance capacity provider being taken on, namely clause 19.2.7. That gives Mulsanne a right of immediate termination if:

Mulsanne claims that as Marshmallow worked up its plan to introduce Marshmallow Insurance into its business it became its intention (if it was not always its intention) to put a major part of the motor insurance, which would otherwise have gone to Mulsanne, into Marshmallow Insurance and thus the right to terminate under that clause is triggered and has been exercised.

Marshmallow’s riposte is to submit that the clause does not have anything to do with what happened when new contracts of insurance were entered into or existing contracts were renewed. It was intended to operate in circumstances in which Marshmallow terminated Mulsanne policies (in sufficient numbers) and arranged a substitute policy for the insured in its place, or in which it procured insureds to terminate contracts part way through their period (which the Mulsanne contracts permitted) with a view to covering the risk in new contracts with a new insurer. Marshmallow submitted that it did not cover a situation in which expired contracts (in sufficient numbers) were renewed into a different insurer, or new customers were entered into a Marshmallow policy rather than a Mulsanne policy.

This is a pure question of construction. There is no material dispute of fact as to what Marshmallow intended to do, or whether that was in relation to a “major” part of the business placed with Mulsanne. On Day 2 Mr Hough accepted that prior to the notice of termination Marshmallow intended to get what amounted to a major proportion of the Mulsanne policies into Marshmallow Insurance by way of renewal, and that was confirmed in correspondence (the word “substantial” rather than “major” was used in correspondence but no point was taken on that in final submissions). The effect of that is that if Mulsanne is correct on its point of construction then it will succeed on this ground of termination. I would also be prepared to find that at least from January 2021 if not before Marshmallow intended to place a lot of new customers into Marshmallow Insurance as well - probably a “major” part.

I was referred to the familiar cases on the construction of contracts - Wood v Capita Insurance Services Ltd[2017] UKSC 24 and Arnold v Britton[2015] UKSC 36. I do not need to set out the oft-repeated citations from those cases because there was no significant dispute in this case as to what their effect is. My task is to ascertain the objective meaning of the language used by the parties, giving appropriate weight to business common sense in the business context in which the agreement was reached. Mr Hough pointed up the role of the contra proferentem rule in resolving ambiguity, while acknowledging that it was not to be used to create ambiguity where none otherwise existed (see Financial Conduct Authority v Arch Insurance (UK) Ltd [2020] EWHC 2448 at para 761 per Flaux LJ).

Mr Hough sought to argue that the “business placed” with Mulsanne and referred to in that clause is only the set of actual contracts which are entered into (“placed”) through Marshmallow, so that clause 19.3.1 must be referring to a form of transfer of those contracts and nothing else. That is said to be the natural construction, whereas Mulsanne’s construction is unnatural and does not coincide with business common sense because if Mulsanne were right then Marshmallow would be inhibited from even entering into discussions with other capacity providers for providing capacity even where those alternatives might provide better and cheaper cover than Mulsanne, and that would be a strange result bearing in mind the absence of an exclusivity clause and the fact that the TOBA makes it clear that the customer connection remains with Marshmallow (clause 15). Contractually Marshmallow was entitled to put business through another insurer, and internal documentation of Mulsanne from December 2019 showed that it acknowledged that that was the case. What protected Mulsanne from the sort of thing which has now happened was the 30 day termination clause in the original documentation. That could be invoked if, instead of arranging mid-term transfers, Marshmallow did what it has now in fact done, which is wait until contracts expire and invite renewals with a different underwriter. If there is ambiguity then the contract should be construed against Mulsanne as the proferens.

Mr Hough’s case is arguable, but in my view it is not correct. In the circumstances of this case it attributes too narrow a meaning to the meaning of “transfer” and “business placed with” Mulsanne. The word “transfer” cannot be taken literally vis-a-vis Mulsanne because Marshmallow’s agency did not include such a power. Nor did the insurance policies give any power for Marshmallow itself to do any switching mid-contract on behalf of the insured. A “transfer” must describe the end of a process in a broader way, looking at its effect. The “business placed with [Mulsanne]” is capable of including the opportunity to renew at the end of the contract as part of the “business”, in a context in which Marshmallow devised the product, prepared a large part of the ratings engine which was used and had expressly reserved “client ownership” to itself - see clause 15. In fact clause 15 supports a wider construction than somehow referring only to existing policies. It forbids Mulsanne from soliciting “the business away from” Marshmallow. The “business” here cannot refer to just the policies currently in force. They could not be solicited away from Marshmallow in the normal sense of the word “solicit”. What could be solicited away from them would be the opportunity to enter into new contracts, and, within that, the opportunity to offer and conclude renewals. That is a virtually inevitable construction of this clause and it demonstrates that “business” is not necessarily confined to concluded policies. True it is that clause 19.2.7 refers to “business placed with Us”, which does refer to policies actually concluded, but the wider import of “business” still applies, in my view, so that when read with “transfer” it is capable of meaning a transfer by a renewal.

This is a much more likely construction in the circumstances of this case. That this is a commercially plausible construction is borne out by what Marshmallow said to potential excess of loss insurers in about November 2020. A Powerpoint slide describes what is planned as “Transfer all current book placed with current carrier Mulsanne over to our own insurer.” That statement does not make what happened a transfer, but it does demonstrate what commercial men in the area thought they were doing, which affirms the plausibility of Mulsanne’s construction (though I would have reached the same decision without it).

It is also supported by some of the material in Julien Praet et Cie S/A v H G Poland Ltd [1960] 1 Lloyds Rep 420 (the first instance decision). This case was much relied on by Mr Christie as demonstrating that seeking a transfer mid-policy would be something that would be seriously in breach of an agent’s (broker’s) duties. I think his point was that this clause cannot have been purporting to deal solely with something which was an egregious breach of duty anyway. I am not sure I agree with this line of reasoning, but to my eyes other aspects of this case are more helpful to him. The case seems to have involved, not an attempt to terminate mid-policy of the kind posited by Mr Hough, but an attempt to bring about a termination at an available termination date with a view to offering and procuring a new policy with a new underwriter. There are various references in the report to this being a “transfer” of the policies - see how the matter was presented (using the word “transfer”) in circulars to policyholders as set out at pages 436, 437 and 438, and the use of word “transfer” to describe what was happening at pages 438 (“Some 98 per cent of the policyholders made the transfer …)” and 441 (“transfer operation”). All this demonstrates that to use the word “transfer” to describe what was going to happen to policies was one well within the bounds of what commercial men would or could understand by the word.

The 30 day termination provision does not gainsay this conclusion. It is highly unlikely that commercial men would think that the transfer mid-policy was such a serious matter (if it was of a majority of the policies) that immediate termination should be allowed, but that a planned removal such as that said to have been planned by Mulsanne should be something which can wait for 30 days. They could both be equally serious. Insofar as there are ambiguities they are not such as to make invocation of the last resort contra proferentem rule appropriate. The position is clear enough without that.

In the light of that reasoning I find that what Marshmallow intended amounted to a transfer of the majority of the business placed with Mulsanne for the purposes of clause 19.2.7, so that Mulsanne was entitled to terminate, and did terminate, the TOBA by its letter of 2nd February 2021 on that basis. Insofar as necessary, it was an effective basis of termination under a following letter of 1st March too.

As is apparent from the letter of 2nd February, a range of termination events are relied on. In this action some of them are also relied on as amounting to repudiations which Mulsanne was entitled to accept. In the light of my conclusion on the transfer of business point I do not strictly need to deal with them in the context of termination, but they, or one of them, may go to damages if a breach is established. I will therefore have to deal with them at least to the extent of considering breach.

Under heads (b) and (d) Mulsanne complained about breach of data protection provisions. In its final submissions it seems to rely on only one matter, namely an alleged lack of co-operation amounting to deliberate obstruction at the end of 2020, which its final submissions describe as “the crux” of Mulsanne’s case. I shall therefore deal only with that point (subject to one additional point I make at the end of this section).

The point arises out of an exchange of emails between Mr Jaimin Patel, Marshmallow’s head of compliance, and Ms Clare Hodson of Mulsanne. On 2nd October 2020 Mr Patel had given notice to Mulsanne of anticipated complaints to the Information Commissioner’s Office from customers who received calls from claims management companies (or purported claims management companies), having made a claim under their policies. He urged the importance of finding out where a leak of information came from. He gave Ms Hodson some details and said he would provide further information on request. On 16 October 2020 Ms Hodson wrote back explaining steps that Mulsanne had taken in order to investigate the matter. On 21st October Mr Patel wrote back expressing a little disappointment at the lack of scrutiny applied by Mulsanne and wondering whether the investigation would be reopened, and the correspondence about it ended with Mr Patel saying (on the same day) that if Mulsanne were comfortable with the investigation then as far as he was concerned the matter could be left there. He professed an intention to continue to send complaints pursuant to clause 17.2 of the TOBA.

Three weeks later Ms Hodson sent an email the response to which is the subject of the complaint under this head in this action. On 12th November 2020 she wrote to Mr Patel:

Marshmallow’s documents reveal that there had been a meeting between Mr Patel and AKB, and it would seem that for some reason they decided to delay giving the response that Mr Patel ultimately gave. AKB confirmed in evidence that he was not minded to be cooperative in this instance.

Ms Hodson’s response was to complain of non-compliance with the TOBA. She said:

There was no direct response to that email. There was further email traffic in relation to customer complaints, but no attempt to comply with the request made by Ms Hodson.

The contravened provisions of the TOBA are said to be clauses 13.8 and 13.11 (the latter appearing in the addendum). Clause 13 deals with data protection matters. Paragraphs 13.1 to 13.4 deal with restrictions on the use of personal data. Clause 13.5 deals with Mulsanne’s right to share personal data, and 13.6 imposes an obligation on Marshmallow to notify loss of data incidents. Clause 13.7 imposes an obligation to store data properly and clause 13.9 imposes an obligation to preserve the confidentiality of information. The first clause relied on by Mulsanne under this head is clause 13.8:

Clause 13.10 is of no significance to this dispute. The second clause relied on is clause 13.11 which provides:

Clause 13.12 requires the parties to work together to ensure that each can process personal data lawfully, and clause 13.13 contains definitions.

I do not consider that the refusal to supply the document sought was a breach of clause 13.8. That clause requires co-operation with the requirements of the rest of clause 13. I have described what the rest of clause 13 requires, and it is not apparent to me which of those stated contractual requirements, at least in the context of the request, need the disclosure of Marshmallow’s procedures for the purposes of co-operation. Mr Christie’s submissions did not identify how this was supposed to work. Nor was it made plain how disclosure of the procedures would assist Mulsanne to comply with its obligations under date protection law for the purposes of clause 13.11.2. Ms Hodson’s email of 2nd December does not really articulate a reason, under clause 13.8 or 13.11 (which is what is now relied on), or why the procedures should be disclosed.

I am therefore not satisfied that this particular act, which is all that survives of a wider data protection claim, is a breach of the terms alleged, much less that it was a repudiatory breach (which is Mulsanne’s other claim).

I also add that the attempt to establish that Marshmallow was in breach of its own data protection obligations, which was the nature of the rest of the complaint about data protection, seems to me to have been rightly abandoned (if abandoned it was). It was unrealistic to expect that breaches on the part of Marshmallow could be established on the basis of the evidence at this trial.

Next, Mulsanne claims to have terminated for breach of the TOBA (either under the contractual termination rights or as a result of repudiation at common law) for failing to allow an audit as is said to be required by the TOBA

The relevant provision of the TOBA upon which this claim is based is clause 16.8:

The first request for an audit came in a letter dated 11^th November 2020 from HL to solicitors then acting for Marshmallow (HFW). The letter starts by setting out various categories of confidential information which it was alleged must have been used by Marshmallow in setting up the competing business of which Mulsanne knew by then. Section 7 seeks the audit:

On 27th of November 2020 HFW responded (in relation to the audit) as follows:

At the end of November Mulsanne’s solicitors sent to Marshmallow a “High-level audit scope” prepared by Deloitte, whose content confirmed that Deloitte would be conducting an investigation of (inter alia) “the Marshmallow production and development environment on the use and application of MICL data and processes”. They required data preservation which included “Email and other communications of individuals who have discussed technical specifications of MM applications with potential new business partners”.

HFW responded on 4th December, drawing a contrast between previous audits which had concentrated on individual policies and policyholders, and the then present proposals, which were said to be “very different”.

The letter then goes on to ask precisely what Mulsanne would be searching for and requiring the identification of any confidential information which is said to have been misused.

HL responded on 13 January 2021. Paragraph 4 of its letter made it clear that the audit would focus on Marshmallow’s compliance with its obligations of confidentiality under clause 13.9 of the TOBA. It averred that Marshmallow’s confidence would be protected by the same provision and said if Marshmallow was concerned that any more specific protection was required then those measures should be devised and agreed following a scoping call which was to take place between Marshmallow and Deloitte. The letter attached various heads of alleged confidential information. They differ in some respects from those which ended up being the subject of this action, but there was some overlap.

By 20 January 2021 NRF had replaced HFW as solicitors to Marshmallow and they responded to the correspondence. So far as an audit was concerned it said:

That stance was disputed by HL on 25th January, in a letter which asserted that the audit rights extended to ensuring Marshmallow’s compliance with all and any of its contractual obligations including its obligations of confidentiality. In a further response dated 28 January 2021 NRF said that its client did not object to a clearly delineated audit that would enable Mulsanne to check Marshmallow had complied with the terms of the TOBA:

That takes the story to the letter of 2nd February from HL terminating the TOBA. However, there was limited correspondence thereafter which might be relevant to the scope and effect of any breach of the audit provision so I will deal with it here.

In a letter of 24 February 2021 NRF returned to the question of an audit. It disputed that Marshmallow had committed any breach of the audit provision in the TOBA and asserted that it had repeatedly indicated that Marshmallow was open to an audit in accordance with the TOBA, “i.e. an audit of its conduct of its duties under that agreement”. It claimed that Marshmallow had properly refused to agree an unfettered fishing expedition, and made proposals for any audit:

The letter went on to assert that if, as Mulsanne claimed, the TOBA had been terminated, then the right to audit went as well because termination extinguished that right.

Mulsanne maintains that what Marshmallow did was refuse to allow an audit which was expressly permitted by the TOBA. That is said to be material, and indeed significant, because Marshmallow was holding Mulsanne’s pen and Mulsanne was entitled to an audit which confirmed whether Marshmallow was taking adequate steps to protect Mulsanne’s and Mulsanne’s policyholders’ confidential information. The breach is said to be not only material for the purposes of clause 19.1.5 of the TOBA; it is said to be sufficiently serious to be repudiatory.

Marshmallow’s case is that it did not breach the audit provision because it never refused a proper audit. The clause should be read as meaning that Mulsanne was entitled to carry out such formal audit as was reasonably appropriate to confirm compliance with the TOBA (which means its functions as an agent) while respecting confidential information of Marshmallow. It was within its rights to establish some procedures which would secure protection of its confidential information in relation to the commencement of its new business. It committed no breach of contract in seeking to discover the scope of the audit and agree a fair and appropriate methodology. It had never refused such an audit. Mulsanne described the attempts to do that as being the raising of unjustified and illegitimate barriers.

There is no doubt that the audit intended by Mulsanne went a long way beyond checking such things as the proper accounting for money, or whether or not proper steps were taken before a policy was incepted. Mulsanne did indeed intend to investigate whether Marshmallow had misused what it described as Mulsanne’s confidential information on the formation of Marshmallow’s new business. On the correspondence, Marshmallow disputed that right. The first question that needs to be addressed is whether that is within the scope of the audit provision.

Neither side made any real submissions as to what “audit” meant in the context of this provision, presumably each impliedly relying on the proposition that they were self-evidently right. So I received little help on the point. Dictionary definitions do not help - one can find definitions which assist each side. So working it out for myself, in my view the most telling factor in this case is the context of clause 16.9 in the TOBA. Clause 16 is headed “Disclosure and administration”. The introductory paragraphs (which are not numbered) make the point that the relationship is one of principal and agent, and then impose an obligation to give proper advice, obviously in that context. There then follow seven Undertakings before one gets to the eighth, which is the audit provision. In that context the more natural meaning of the word is one which limits the function of the audit to checks that the obligations contained in clause 16, and perhaps similar obligations arising otherwise, have been complied with. The use of the word “audit” in that sort of context is familiar and understandable. The audit that is required is one of auditing the transactions under the agreement, which is what is contemplated by clause 16.7. That is consistent with the use of the word in clauses 9.5 and 10.4. What Mulsanne’s audit would require is not a process of checking or verifying so much as conducting a roving inspection to see if there has been a breach of the agreement in relation to what is said to be Mulsanne’s confidential information. I do not consider that that is naturally referred to as an audit. This approach the matter is in line with the main, non-figurative, definition in the Oxford English Dictionary, which is:

Mulsanne’s process does not fit within that. I prefer that construction to the wider one which would be required for Mulsanne to succeed.

The consequence of that is that Mulsanne was asking too much. They were demanding access for a process to which they were not entitled. Marshmallow was entitled to refuse that demand. As an alternative Mr Christie seemed to complain that Marshmallow refused any audit. If that was his submission it fails. The correspondence above demonstrates Mulsanne’s stance clearly enough, and especially NRF’s letter of 20th January. Despite the excessive demands of Mulsanne, Marshmallow seems to have been willing to allow an audit, and perhaps one going beyond that which it had to concede, if appropriate terms could be agreed to protect its interests. That is not a refusal of any audit. It is an attempt to agree an audit which protects Marshmallow’s own legitimate commercial interests and is in compliance with what I have held the obligation to be.

The claim that Marshmallow was in breach of the agreement for refusing an audit therefore fails, as does reliance on such a breach as a reason for determining the TOBA. It follows that Mulsanne was not entitled to terminate the TOBA on this basis (if that matters).

Mulsanne claims to be entitled to terminate for misuse of confidential information both under the contract and by way of accepting a repudiatory breach. Its claim encompasses the specifically pleaded material (the “Confidential Information”, the term of art used in the pleadings to describe the parts of the ratings engine which I have considered above), and also confidential information (with no capital letters) more generally. An issue arises as to whether it is entitled to rely on this latter form of confidential information (which I will call “the other confidential information” in this part of this judgment). I shall consider that point first.

Mulsanne’s Particulars of Claim refer briefly to this other kind of confidential information in just three relevant paragraphs, without in any way seeking to define or particularise this other confidential information. In paragraph 28.9 there is a pleaded “duty to refrain from disclosing confidential information and/or documents, including the Claimant’s Confidential Information …”; in paragraph 45.3 there is a pleading that there was a repudiatory breach by “sharing and/or using the Claimants’ confidential information including but not limited to the Confidential Information”; and in paragraph 55.5 there is a complaint of a failure “to avoid and/or refrain from disclosing confidential information and/or documents, including the Claimant’s Confidential Information … and instead … disclosing it to the Second Defendant”. In paragraph 20 the contractual duty to protect confidential information is pleaded in relation to both types. Otherwise the claim looks as though it is all about the Confidential Information (as defined).

Mulsanne’s case now is that Marshmallow shared its confidential information (presumably including both types) with “a whole battery of [unauthorised] third parties”, including Artex (the company assisting with regulatory matters in Gibraltar), the Gibraltar regulator, reinsurers and potential investors. The sort of disclosure that is now relied on is of such matters as loss ratios and other measures of the Mulsanne business which Marshmallow conducted. That was never pleaded in terms of those to whom disclosure was made, and the information was never particularised. Marshmallow’s first complaint is that the information was not particularised, was therefore not before the court, and was not properly addressed in disclosure or evidence. There is much to be said for this point, even allowing the sort of latitude to Mulsanne to which I have already alluded more than once in this judgment. Without a proper pleading Marshmallow just does know what case it has to meet, and the importance of proper pleading appears from the authorities referred to above.

The second basis of Marshmallow’s opposition lies in the history of this matter. On 6th September 2021 Marcus Smith J considered an application to amend the Particulars of Claim by adding new categories of information to the Confidential Information that was already expressly pleaded. The proposed new material was some of the metrics of the Marshmallow/Mulsanne business - loss ratios and claims information.

Marcus Smith J refused the application. He acknowledged that this was a new claim which was extensive in its scope (rejecting the submission that it was modest) and would require “revisiting of all aspects of the trial process ranging from pleadings to disclosure to witness statements” (para 17); he acknowledged that the existence of confidentiality in this area was “nuanced” (para 28) and that a lot more work would have to be fitted into the timetable if the new claim were to be ready for trial. Then he considered the lateness of the application, and came to the conclusion that it could have been made much earlier than it was. He pointed out (para 39) that in a letter as long ago as 22nd October 2020 HL expressed the view that it was highly likely Marshmallow had shared “historic loss ratios, forecast loss ratios and other critical data with third parties such as those identified in this section - reinsurers, regulators and so on.” No step was taken to include those matters in the claim when issued and when Marshmallow’s then solicitors said they assumed the claim had been dropped there was no contradiction by Mulsanne (judgment para 40). Marcus Smith J rejected a submission that an earlier pleading was vulnerable to being struck out. There was no satisfactory reason why the claim had not been made before. He therefore disallowed the amendment.

As Mr Hough pointed out, the attempt to bring the contractual claim of infringement of confidentiality into the contractual claim is an attempt to run an unpleaded case where an attempt to plead it has already been refused on the basis that it was too late to plead it. He said that was impermissible, and I agree with him. Furthermore, after Marcus Smith J’s decision Mulsanne took out of its witness statements a significant number of paragraphs which related to that point, so when those statements were confirmed in the witness box there was no evidence on the point from the claimant. That makes it even more inappropriate to allow Mr Christie to rely on this other confidential information. I also remind myself that I relied on the other confidential information not being in issue in this case when I refused to order some additional disclosure.

For all those reasons it is not open to Mulsanne to run a case that other confidential information was actually confidential or that, if it was, it was misused. In those circumstances it cannot be the basis of a repudiation claim to the extent that it might otherwise have been a breach of contract. Nor can it be the subject of a claim to terminate for breach under the express terms of the contract.

So far as there has been found to be misuse of the Confidential Information, it was not sufficiently serious to amount to a repudiatory breach. Such breaches as I have found are more in the nature of incidental, and almost accidental, breaches, some of them in the context of a situation in which Marshmallow was conscientiously trying to avoid the use of Mulsanne material.

So far as concerns the claim to have terminated under the contractual terms for misuse of the Confidential Information, had it been necessary to decide it I would have decided that the case failed because this was not a breach which was specified in the termination letter of 2nd February.

Having found that the TOBA was terminated by the letter of 2nd February, it becomes necessary to consider claims of post-termination breaches by Marshmallow. The claim here is that Marshmallow was in breach of various obligations in failing to provide certain information which Mulsanne claims would have enabled it to take over the administration of its policies effectively, and in particular to be in a proper position to offer renewals to policyholders when their policies expired. It is said that such breaches caused damage, again principally by virtue of lost renewal opportunities. This is a serious issue in this case, because the thrust of Mulsanne’s case is that it was unable to have an opportunity to renew any of its Marshmallow policies, and since all policies will have had a renewal date in the year following the inception of the new Marshmallow business, and since policyholders wishing to renew will have renewed into Marshmallow policies by now, that opportunity is said to have been permanently lost to Mulsanne. If I find breaches any damages will, so far as appropriate, be assessed on a separate inquiry.

The alleged breaches (already identified in outline above) can be summarised again as follows (more detail follows in the sections devoted to each breach). They were failures to supply details of:

Most of the material now sought was set out in an appendix to the Letter of Claim dated 11th February 2021 from HL to NRF which demanded it in paragraph 14(c), although the shape of the demands shifted in relation to some of that material in the course of the debate in correspondence and at this trial. For the sake of completeness I note that item 1 (“Update to records already provided by [Marshmallow] - For all policies, the current complete in force policy details per customer”) was not pursued as a continuing demand. It is now common ground that this material has always been available, because at the end of every day Marshmallow reported to Mulsanne what policies had been put in place, with details of the policy inception date; the car (group, whether left-hand or right-hand drive and other material); no claims discount; excess; the policyholder age, address and telephone number; the claims and convictions history; and the premium.

In addition the complaint about the audit technically falls in part under this head of post-termination breaches, but I have already held this not to have been a breach, so I do not need to deal with it further under this head.

Mulsanne relies on both the terms of the TOBA and general principles of agency law as entitling it to the information sought. The relevant provisions of the TOBA are the following.

Clause 19.3 provides for what happens on a termination:

Clause 20 refers to “Records” and provides for the maintenance of records for 6 years after the expiry of the risk, and goes on:

Clause 15 preserves client or customer ownership in Marshmallow. That seems to me to have a potential relevance to some of the questions that arise in relation to the post-termination claims, though Mr Hough did not seem to rely on it in this context.

I should also draw attention to clauses 6.2 and 8.1. The latter, but not the former, was pleaded. These require (respectively) compliance with FCA rules and principles, and the adoption of Good Industry Practice, as defined in clause 1.1. They have a significance because it is part of Mulsanne’s case that a duty to act fairly vis-à-vis customers is said to qualify and expand what was required of Marshmallow. That duty is said to arise from regulatory standards. Principle 6 of the FCA’s Principles for Businesses says:

Mr Christie relied on an amplification of this in an FCA publication entitled Fair Treatment of Customers from March 2015 which requires regulated entities to achieve 6 customer outcomes, which include the following:

In addition to the contractual duties referred to above, in this context Mulsanne also relied on common law and equitable principles applicable to agents. Mr Christie drew attention to the following:

Reliance was also placed by Mr Christie on the position of a sub-agent. He submitted, and I accept, that unless Mulsanne consented to or authorised it, Marshmallow was not entitled to delegate its agency obligations so as to give rise to a direct relationship between the sub-agent and Mulsanne. Marshmallow remained liable if the “sub-agent”, appointed without consent, failed to perform. I have little difficulty in accepting those propositions, but more difficulty in seeing how they apply in the present case. Marshmallow does not rely on any sub-agency (in that sense) in this case, and neither does Mulsanne.

Last, it is necessary to deal with clause 19.3.1, because on analysis it figures largely in Mulsanne’s case as to what Marshmallow has to do following a termination. As will become apparent a large part of Mulsanne’s case depends on the obligation to deliver up records under 19.3.2. That obligation is readily understood, but there is a dispute as to whether Marshmallow should have gone further and procured documents not under its control or “facilitated” Mulsanne’s (or a nominated intermediary’s) being able to do so; or to take positive steps to put Mulsanne (or a nominated intermediary) in a position in which it could administer the policies itself. This further obligation would have to arise out of clause 19.3.1 itself, or an implied term in the contract. I prefer to deal with this possibility in a separate section, and to do so after considering the other alleged breaches, which will make more sense of the debate about this clause and implication.

Having set out the legal parameters of the obligations, I now turn to the particular breaches which are relied on.

In the letter of claim Mulsanne asked for:

In a letter dated 24th February NRF agreed that Marshmallow would provide these documents. The offer to supply the details was repeated in a letter of 28th April, and they were provided on 17th May.

The wrongful failure to provide these details earlier was clearly set out in Mulsanne’s written submissions. Marshmallow did not deal with this point at all in its final submissions. There is no dispute (and can be no dispute) that documents falling under this category fall for disclosure under the obligations referred to above, so the only question is whether the delay in supplying them amounted to a breach. No evidence was given to explain or justify the length of time it took. Doing the best I can, I find that there was a breach because it is hard to see that it can really have taken over three months to provide this material. There would be a lot of policies to consider, but even so three months seems to me to be an excessive time. I therefore find a breach. I can make no finding as to when they should have been provided because, as I have just said, I received no evidence on the point, but whenever that point of time is it is short of three months from the demand. If it matters that point can be investigated on an inquiry as to damages, although at the moment it is not easy to see how significant damage can have been caused by this breach.

What was sought here in the letter of claim was:

This information has never been provided. Mulsanne says it was required so that Mulsanne could effectively service the policies when it took back administration of them; Marshmallow says (in essence) that this information is Marshmallow’s, relating to Marshmallow’s business in which Mulsanne has no interest, and Mulsanne is not entitled to it.

As already explained, where a customer required it Marshmallow would sell other policies or services, breakdown cover and/or legal expenses cover. In doing so Marshmallow acted as agent for the service providers. If commission or fees were payable then they went to Marshmallow. It has not been suggested that Marshmallow was in breach of any duty owed to Mulsanne (fiduciary or otherwise) in doing so. Nor can it be suggested that it was part of the agency contract with Mulsanne. This was Marshmallow’s own business which it operated alongside the business which it offered to customers when the arrangements started. It was doubtless viewed as part of the “customer ownership” or “client ownership” under clause 15. True it is that the sale of the motor insurance contract was the context in which the add-on policies were sold, and that absent that contract they would not have been sold, but where the principal has consented to this activity as being Marshmallow’s own activity (which consent or permission is actually acknowledged in a letter from HL dated 15th October 2021) then Marshmallow is entitled to treat it as its own separate activity in relation to which (at least on the facts of this case) there is no obligation to account, or provide details to its principal on the motor insurance.

There is therefore no obligation on Marshmallow to account for any financial benefits that might accrue to it from these activities and no obligation to provide details of a business in which Mulsanne, as principal under the motor insurance arrangements, had no interest (subject to a different argument about servicing, which I deal with below). The details sought are not records of “Ours” within clause 19.3.2 of TOBA, and the add-ons are not “transactions being carried out in the course of this agreement at the date of termination” (clause 19.3.2). Not being part of the Mulsanne agency arrangements, they are not documents within the fiduciary obligation to provide documents relating to the agency.

Nonetheless Mulsanne submitted that the add-on cover was treated as part of the policy if one looks at the documents. Thus a sample policy schedule produced at the trial says:

On another sample schedule the sort of details one would expect from a motor policy are set out and at the bottom of the first page there is a “Price Summary”, in which the annual premium and Marshmallow’s fee are set out, and then “Add ons” with a total.

Mr Christie also relied on the Policy Wording which referred to the need to read the policy, the policy schedule and the certificate together. “The policy schedule tells you which sections of the policy apply to you…” I do not see how this assists him to improve his point because the wording of the policy document that he puts forward refers only to the motor insurance and not to the terms of any add-ons.

He also relied on “renewal” documents from April 2020 which he said demonstrated that the breakdown cover add-on was treated as part of the policy which would be renewed. The third page of that document set out the following information:

Then in the right-hand column there are two boxes. The first shows fees, excesses and cancellation fees. Then below that there is a third box with the contents:

He submitted that that demonstrated that the add-on was part of the policy which would be renewed. AKB confirmed in his evidence that the add-ons were regarded as an integral part of the policy, which depended on the existence of the motor insurance element. This is a central part of Mr Christie’s case in relation to his client’s entitlement to add-on information.

I do not consider that this argument is correct. It is true that as far as the customer is concerned the two different types of policy were lumped together in the documentation, but that does not affect the legal analysis of whether the activity was one which gave rise to fiduciary obligations to provide information, or a contractual liability to disclose documents.

Mulsanne had what turns out, on analysis, to be a further line of argument with a different foundation. It was submitted that the evidence demonstrated that if the information was not provided then Mulsanne would not be able to take over the administration of the policies concerned (or indeed any policies, because without Marshmallow’s information it would not be known whether any given policyholder had the benefit of the add-ons) because customer confusion and unfairness would be caused if one and the same company could not deal with both add-ons and with the motor policy, whether on a renewal or an MTA. He relied on cross-examination of AKB as demonstrating the likelihood of confusion.

In order to make this point Mr Christie has to rely not on the normal contractual and fiduciary duties of an agent, because they do not get him where he wants to be, but on regulatory obligations. He puts forward the regulatory duty to act fairly which I have set out above. I do not consider that, without more, he can rely on that material. That is a regulatory obligation which is not necessarily enforced as a matter of contract. Its enforcement is a matter for the regulator, not for a contractual counterparty.

I have considered whether Mr Christie could rely on clause 6.2 in order to import the duty in these circumstances. It provides:

It was not pleaded in the Particulars of Claim, but he invoked it in his final submissions, so I will consider whether it can be made to work for him in terms of incorporation of the duty.

I do not consider that it can. The obligation arises “when acting on Our behalf under this agreement”. When arranging the add-ons Marshmallow was not acting on Mulsanne’s behalf, and was not acting under the agreement either. It was acting collaterally to the agreement. That position is not changed at the end of the TOBA. While it might be possible to extend the meaning of the words “when acting on Our behalf under this agreement” to hand-over activities in relation to activities where Marshmallow was acting on behalf of Mulsanne, I do not consider that they extend to material which did not arise when Marshmallow was acting for Mulsanne.

The same applies to clause 8.1:

Accordingly, I do not consider that Mr Christie can get in the duty to act fairly or to where he wants to be via these routes.

In case I am wrong about that then I will express a view as to whether Marshmallow would be in breach of a duty to act fairly if it were incorporated into the contract and in the event of a failure to provide details of add-ons. The factual basis on which the case is put is customer confusion. It is apparently said both that Mulsanne would not be able to act fairly without the information, and that Marshmallow would not be acting fairly in withholding it because of that confusion.

Even if there would be some customer confusion if he/she had to deal with Mulsanne over motor policies and (if necessary at all) with Marshmallow over the add-ons, I do not consider that that confusion means that Marshmallow would not be acting fairly within the duty in withholding the information. Although the point was not fully developed in the evidence, I consider it likely that any confusion would be removed or mitigated by sensible dealings with the customers. Thus if Mulsanne were offering a renewal it could make inquires of the customer about add-ons and offer its own. If a customer insisted on using Marshmallow’s add-on insurers (which is unlikely) then the customer could be referred to Marshmallow. If a question arose on an MTA then any confusion would be resolved by Mulsanne telling the customer that inquiries about add-ons should be referred to Marshmallow. Any “confusion” could be swiftly and easily resolved. I do not consider that any confusion would amount to culpable unfairness within the regulatory regime, and I certainly do not consider that Mulsanne could not take on the administration of the policies without information about the add-ons because it could not act fairly vis-à-vis the customer in those circumstances. A customer might well be irritated, but I do not consider that that would be a breach of a duty to act fairly.

Accordingly I find that Marshmallow was not in breach of the post-termination provisions of the TOBA in respect of the add-ons.

This is item 6 in the schedule to the letter of claim of 11th February. What was sought is:

The complaint goes first to credit card details, and it has changed into a complaint with two elements. First, the complaint is that Marshmallow did not provide the information; and second, that it did not “facilitate” access to the person who did hold it (as it transpired, companies known as Stripe and Checkout).

The facts in relation to the data are simple. As set out above, separate companies (first Stripe, then Checkout) took credit card numbers and moneys from customers and passed the moneys to Marshmallow. Those companies, and not Marshmallow, then held on to the data (the numbers) under appropriate and very strict security. Marshmallow does not have that data (other than the last 4 digits for transactions since mid-2020), and cannot simply call for that data because it does not have the necessary security arrangements in place to hold it. It is not known what the contracts say about this - the precise arrangements between Marshmallow on the one hand and Stripe/Checkout on the other were not available at the trial. Marshmallow does not seem to have volunteered them as part of its disclosure, and I assume that Mulsanne did not ask for them as part of the request-based disclosure put in place by Marcus Smith J. The result of that is that part of the debate on this point took place in the twilight if not in the dark as to precisely what Marshmallow can do under the Stripe/Checkout contracts in relation to this data.

There seems to be a dispute about what Marshmallow offered, and when. That makes it necessary to consider the correspondence passing between the parties on the point.

NRF’s response to the first request for the data in the letter of 11th February was to say (in a letter of 24th February):

HL did not pursue that point in its next letter of 1st March, but NRF returned to the point in its letter of 28 April 2021 about a “transfer plan”. It said:

The letter proposed a transfer plan in relation to the book of policies, and as part of that plan one of the “Policyholder Records to be Provided” was the last four digits of payment cards used by customers when making annual payments.

HL responded on 5th May. It said that further explanation was required of this matter and pointed out that in renewal notices there were statements to the effect that payment details would be used on renewals and amounts should be collected “by charging the credit or debit card that we have on file for you”. NRF were invited to explain how that could happen if Marshmallow did not hold payment details.

NRF responded on 17th May 2021:

That explanation was not satisfactory to HL who accused NRF of being “deliberately evasive” in a letter of 18th May. HL challenged NRF as to how the representations could be made in the renewal documents if Marshmallow did not keep payment details. The letter went on:

That last accusation seems to me to be inaccurate. Marshmallow had not refused to facilitate access to the third party provider (yet); it had not been asked to do so.

NRF responded on 24th of May 2021. In relation to this matter it said:

This letter does contain an apparent refusal to “facilitate” access to Stripe, whatever that may mean. This is considered further below.

HL responded on 26th of May by saying:

NRF responded on 28 May:

HL continue the debate in its letter of 30th July:

Paragraph 6 of the letter returns to the theme of HL’s being unable to understand how it is that it can be said that Marshmallow did not hold payment information when various customer-facing documents refer to the fact that moneys would be collected from a credit or debit card on file. Paragraph 7 continues:

NRF responded at length on 29 September 2021. Most of section 2 of that letter is devoted to a reiterated explanation of the position. So far as relevant it reads:

HL’s response on this point points out the agency status of Marshmallow and complains that Marshmallow has “consistently refused to recognise that withholding the payment information which it uses to process payments as agent of Mulsanne amounts to a failure to discharge its contractual obligations under clause 19.3.” It goes on:

In the last letter in the chain NRF repeats that it was not obliged to pass on credit card details which Marshmallow could not obtain from its own payment services providers. Paragraph 3.3 asserts that the TOBA does not oblige Marshmallow to take any active step to introduce Mulsanne or an alternative intermediary to the payment services providers. It goes on:

In what follows I put on one side for the moment the question of what, if any, breaches there were, or might have been, in relation to delivery up of the last 4 digits of the credit card number. That is not what this point is really all about, though I return to it when I have considered the main point.

The first task that arises in relation to the correspondence is to determine whether or not Marshmallow was in breach of its obligations to deliver up records. It is plain enough that payment details of the kind sought would be part of the agency documentation and information that Marshmallow would be obliged to deliver up (both in contract and under its normal fiduciary obligations) if and insofar as it held them. That much is at least implicitly if not expressly conceded by Marshmallow in its agreement to deliver up the last four numbers on the card. If information is held by a third party over whom Marshmallow has sufficient control to order its delivery up then by the same token Marshmallow would be under an obligation to procure the delivery up. One does not need the arguments about sub-agency advanced by Mr Christie to determine that.

However, I find it to be the case that Marshmallow neither holds the information itself (that no longer seems to be challenged) nor can it compel Stripe/Checkout to deliver it up to Marshmallow. The determination of the latter of those two propositions has been dogged by the fact that no-one seems to have asked Stripe/Checkout what the position is, and the agreement(s) with Marshmallow have not been produced. However, doing the best I can I find that it is unlikely that its production can be compelled. AKB gave evidence about the security requirements surrounding such information, and that was not challenged. Nor was it challenged that Marshmallow does not have those arrangements in place itself. I accept his evidence. The sensitivity of that information is obvious. In those circumstances it is unlikely that proper data protection principles would allow Marshmallow to be able simply to direct the handover of the information to a new insurer or its intermediary. What Marshmallow had was an arrangement under which it can call for Stripe/Checkout to take money from the card and pass it on (if the customer consented). Quite what would be permitted in terms of handover of information was not investigated by either side, but it has not been established that Marshmallow can demand the transfer of the information as such. Mulsanne bears the burden of proof on this matter, and although the contract is in the hands of Marshmallow, Mulsanne has not called for it. It could have made its own inquiries as to the position, but if it did so the position has not been revealed. I do not consider that this is a case in which the principles in Armory v Delamirie (1722) 1 Strange 505 operate in favour of Mulsanne (as Mr Christie said they did). It would not be right on the evidence to draw the inference that would have to be drawn in Mulsanne’s favour about control of the data which would allow Mulsanne to succeed on the question of control. Accordingly Mulsanne has not demonstrated a breach of any obligations in Marshmallow’s failure to deliver up all the credit card information or procure its delivery up.

Mr Christie advanced an argument to the effect that if it is the case that Marshmallow cannot compel the production of the information to Mulsanne as its principal then that lack of access is a breach of the law of sub-agency, or perhaps the provisions of the agency itself.

This line of argument was not particularly developed (particularly by Mr Hough) in submissions before me, but it is not clear to me how the principles of sub-agency assist Mr Christie. There was no sub-agency in the sense of a permitted delegation with the consent of Mulsanne so as to give rise to privity between principal and sub -agent (see Wright J in Calico Printers Association Ltd v Barclays Bank Ltd (1930) 38 Ll L Rep 105, cited in Equitas Ltd v Sande Investments Ltd[2021] EWHC 631). No-one contended that there was such consent.

Nor is it even clear to me that there was a meaningful delegation of any of Marshmallow’s agency duties to Stripe/Checkout. It was under a duty to collect in premiums itself, and it did not delegate that duty. What it did was to set up a collection mechanism involving Stripe/Checkout as collectors of credit card moneys, but that did not delegate the collection to Stripe/Checkout any more than a collection directly by its bank would have been a delegation. This was in the nature of an incidental act in the performance of the agency, like (for example) the renting of office space from which to conduct the activities, or the acquiring of internet services to allow a website to be run. If it matters, this sort of thing was anticipated in the trade - see HL’s letter of 30th July 2021. And again, if it matters, and contrary to some of the suggestions made by Mr Christie that Mulsanne knew nothing about arrangements with Stripe, it would seem that Mr McCauley was aware of the involvement of Stripe, and the fact that it held credit card data, from a relatively early stage. In an email of his to Mr Berg of Marshmallow of 16th May 2019, in which he was raising questions with Marshmallow as to certain policies which had recently been incepted, he said (and I set this out with the response from Marshmallow which followed):

Mr McCauley was not cross-examined on this exchange, and there was no reference to this particular exchange at the trial (though other parts of the email chain did figure in the cross-examination of Dr Holliday), but it is clear enough that he knew Stripe, and therefore a third party credit card payment collector, was involved, and that information about the card would have to be obtained from Stripe. He apparently did not consider it likely that Marshmallow held that information itself. As an experienced insurer, he would be likely to know about the standard arrangements that are made with third party processors, which his own solicitors’ conduct referred to. The engagement of Stripe/Checkout was an unsurprising incidental act which does not qualify as a form of sub-agency, in my view.

However, that does not deal with the whole of Mr Christie’s point because he submitted that if Marshmallow put in place an arrangement which did not allow it to extract all the credit card details for onward transmission to the principal, then Marshmallow was in breach of its own agency obligations.

This requires a determination of whether Marshmallow’s agency obligations required it to have arrangements in place which provided for such access where the credit card information was stored for potential future use. I do not consider it was in breach of duty in not collecting and storing the information itself. It was not obliged to adopt any particular mechanism for gathering in premiums. Was it under an obligation to make sure its services provider stored numbers for later use? Again, I consider not. That was a business judgment it was entitled to make.

The key question now is whether, if Marshmallow did make arrangements for the storage of numbers for future use in future transactions, it was under an obligation to make sure they were available to Mulsanne or an intermediary nominated by Mulsanne in the event of a termination of the agency. I consider that if it were demonstrated that such an arrangement was possible then Marshmallow was under an obligation to achieve that result. It was an agent, and the contract contemplated its termination and the introduction of a substitute intermediary - see clause 19.3.1 - and that substitute (or Mulsanne) ought to be able to have access to data acquired for the purposes of the agency if it were capable of being made available. I consider it would be a breach of the agent’s obligations were it to enter into arrangements for the storage of numbers for future use by another intermediary (or Mulsanne) if it could be arranged that the principal should have similar access but those arrangements were not made.

However, the difficulty for Mulsanne is that it has not been demonstrated that such arrangements could have been made. It is far from obvious that they could. One would expect the Stripe/Checkout contract to be between Stripe/Checkout and Marshmallow - indeed, I am sure it will have been. It is not at all obvious that Stripe/Checkout would have even been permitted (under data protection rules) to make the data somehow available to a new intermediary, let alone that it could have been obliged to put such arrangements in place. Mr Christie’s case would have required an obligation to put terms in place which would have required Stripe/Checkout to enter into a replacement contract with the new intermediary, and it is far from plain, or even probable, that such an obligation would be undertaken by Stripe/Checkout. One can think of all sorts of reasons why Stripe/Checkout would not be prepared to bind itself in advance, and it may even be that data protection principles (a minefield not investigated at the trial) would prevent their doing so.

I therefore consider that it has not been established that there was a breach of obligation on the part of Marshmallow in failing to make sure that the new intermediary (or Mulsanne) could exploit the stored data.

That leaves one last point to consider. Mulsanne complains that Marshmallow failed to “facilitate” access.

It is not apparent what the source of such an obligation would be. It cannot be within any of the express obligations to give up data actually held. I have determined below that no implied term has been established to the effect of requiring positive steps to make sure that the intermediary or Mulsanne can take up administration of the policies. In those circumstances the source of the obligation to “facilitate” (which is an uncertain term anyway) is not apparent.

Even if there were one it is not apparent that Marshmallow was in breach of any such term under this head. It was never called on to take any particular steps to assist in transferring the benefit of the held data. The most that was put forward was a general demand to “facilitate” Mulsanne’s access to Stripe (and Checkout) - see paragraph 18 of HL’s letter of 18th May 2021. What seems to have been meant by that is that Marshmallow should actively require Stripe/Checkout to give access to the data to Mulsanne. That is the context of the implicit demand - paragraph 15 required that access be given by Marshmallow positively arranging for Mulsanne to access the data in the hands of the service provider. When NRF responded in its letter of 24th May (para 2.3) it was dealing with an apparent demand in those terms - see the terms of that paragraph. True it is that the letter does contain an apparent refusal to “facilitate” access, but it must be read in the context of what it was being required to do. HL was demanding that Marshmallow give details that it simply did not possess (apparently refusing to accept that Marshmallow did not have the data or access to it as such), and demanding that Marshmallow should bring about some sort of direct relationship between Mulsanne and Stripe/Checkout. As I have pointed out, it is not apparent that Marshmallow could have done that. It is in that sense that the refusal to facilitate should be understood. It should not be understood as a general refusal to do anything at all. Marshmallow had not been asked the right question. HL seemed rather obsessed by the idea that Marshmallow had or controlled the data.

In NRF’s letter of 29th September Marshmallow makes its position clearer. It suggested Marshmallow would consider steps to “facilitate” contacts or dealings between Mulsanne and Stripe/Checkout provided they were identified. I consider that this was a proper stance. If contrary to my finding, there had been a prior breach then Marshmallow would not have been in breach after this letter anyway. It is not apparent on the evidence what more they could usefully have done, and they were certainly not asked to do anything appropriate or achievable before this time, and they did not refuse to do any such thing. In evidence AKB said that if approached for some relevant consent to Mulsanne’s dealing with Stripe/Checkout it would have consented. I accept that evidence. I think it likely that if HL had asked the right question earlier it would have got that answer.

Mr Christie’s case under this head, as under others, involves an averment that Mulsanne could not properly (or fairly) administer the policies without these credit card details. If it matters, I would reject that case on the facts. It might be the case that it would have been more seamless if they had had them. Increased payments could have been taken more easily and without requiring the policyholder to give the card details again if Mulsanne had made its own arrangements to that end; but the latter act is hardly difficult. Repayment of premiums, whether on an MTA or renewal, would have been more tiresome because of the money-laundering requirement to re-credit the card that the money came from. But it is not obvious why Mulsanne, could not have repaid moneys to the particular card with the benefit of similar arrangements to those which Marshmallow had reached with Stripe/Checkout with the benefit of the last four digits (which were proffered) to identify the card.

It follows, therefore, that I do not consider that Marshmallow was in breach of any obligation in respect of credit card information in relation to a failure to supply the full credit card details. That leaves the question of whether the initial failure to supply the 4 digits that Marshmallow did hold was a breach. As appears above NRF originally said that their clients held no data. It corrected that on 28^th April and the data was provided on 17^th May. Mulsanne maintains that the initial failure to supply this information was a breach of obligation to deliver up an agent’s information held for its principal. The significance of this point can be judged from the fact that it was not developed in Mr Howe’s final submissions, and only crept in by virtue of a reference to the dates in one paragraph of the relevant section of his written final submissions.

Nonetheless Mr Howe invites a finding. I find that there was a breach of duty in failing to hand over the data (which Mulsanne accepted should be handed over) within a reasonable time of the demand. There was no evidence or debate as to what a reasonable time was, and doing the best I can to fix that time I note that it took just under 3 weeks to supply the information after NRF said its client had it. I therefore find it ought to have provided it within 21 days after the initial request, which means it was in breach from 24^th February 2021. This was an accidental breach arising from a mistake made in good faith.

A significant proportion of Marshmallow’s customers (60% according to one piece of evidence) wished to pay their premiums by instalments. Where that was the case Marshmallow arranged a loan for them for the whole of the premium. That loan was from Premium Finance (Premfina), as set out above in the section describing add-ons. In making these arrangements Marshmallow acted as the agent of the customer (according to AKB’s second witness statement) and fees were payable to Marshmallow by customers in respect of this activity. Marshmallow was also the loan administrator and it emerged in oral evidence that Marshmallow guaranteed repayment of the loans. It seems that Marshmallow collected and held direct debit details in respect of repayment of the loans - I infer from that that it received the loan repayments and passed on appropriate amounts to Premfina. It is those direct debt details which are (at least partly) the subject of this claim.

The records sought under this head were within the same demand as the demand for the credit card information:

As already pointed out, on 24th February 2021 NRF said its clients did not store any payment details. However, on 1st March HL raised the question of monthly premium collections and said:

On 28th April NRF corrected the position about payment information, as it had done in relation to credit card details:

HL responded on 5th of May:

On 17th May NRF reiterated that Marshmallow did not collect direct debit details as agent of Mulsanne but as part of its distinct role in arranging such loans for customers. HL responded with an apparently wider demand on 18th May:

Thus the debate seems to be moving from a demand for payment details to a demand for some sort of “access” to Premfina. NRF responded on 24th of May (in terms already seen in relation to the credit card information, but I set them out here again for convenience):

HL continued to engage on the need for some sort of unspecified arrangement about Premfina. In its letter of 26th May it said:

NRF sought to rise to this challenge in a letter of 28th of May:

On 30th July HL responded reiterating its demand for “access” to Premfina:

The letter went on to justify why it was said that access to details about premium finance were necessary to the administration of the policy by Mulsanne or CCG. The mechanisms involved are discussed below when I consider the evidence.

That letter was not responded to by NRF until 29 September 2021. It responded:

Later paragraphs of the letter deal with the mechanics of how additional premiums or refunds might be achieved by a new intermediary and paragraph 3.6 points out that:

HL responded on 15th October challenging what was said about the mechanics of additional premiums and refunds and complaining that what was offered fell short of the discharge of Marshmallow’s obligation to facilitate a transfer of servicing and enable direct dealing:

The documents/information never were delivered up, but it appears by then that Mulsanne did not really want them in the absence of compliance with the other demands it was making.

From this narrative it appears that Mulsanne demanded two things over the course of the correspondence. First, it demanded just records. Those were refused by Marshmallow until 29th September, so if it was obliged to provide them then it was in breach until then. Second, it subsequently demanded access to Premfina, and that was refused until the offer of facilitation of contacts on the same date. If anything more than that was required then Marshmallow remains in breach of the obligation.

The first battleground is the records and whether Marshmallow is obliged to deliver up the records of direct debit payments (which was what was originally asked for, though more documents were ultimately offered) pursuant to its contractual and general obligations as an agent. I do not consider that it is. There were two different transactions. The policyholder was not paying the premium in instalments (though he/she may have thought of it in that way). First there was a loan under which instalments were paid; and then there was an insurance policy under which the complete premium was paid in one tranche. Mulsanne was not paid in instalments; Premfina was, in respect of its loan. This dichotomy is actually anticipated by the TOBA – see clause 11.10. The records which Mulsanne seeks were generated solely in the former (loan) transaction.

In each of the two transactions Marshmallow acted in different capacities. In the loan transaction it acted as agent for the customer, and then for Premfina as loan administrator. Mulsanne and its insurance business had nothing to do with that at that stage. Then in the insurance transaction Marshmallow acted as agent of Mulsanne and took in the whole of the amount of the premium in that capacity and then accounted to Mulsanne. The documents which Mulsanne seeks came into existence in the first agency, not the second.

Not being documents generated in or for the purposes of the Mulsanne agency, it follows that they are not documents or records for which Mulsanne is entitled to call under the contract or the normal agency principles. Marshmallow is not in breach of the delivery up obligations accordingly.

That deals with the normal contractual claims and common law/equity based claims. On the facts of this case Mulsanne asserted that the records had to be delivered up so that it or its new intermediary could function effectively in the administration of the policies. Its case was that if a new intermediary tried to deal with an MTA which required an adjustment of the premium (either an additional premium or a refund) it could not effectively do so, in the light of customer expectations, without the payment records, because a customer would expect any additional premium to have to be added to the loan (so that it would be paid by instalments). Mr McCauley said that in his experience it would invariably be the case that additional premiums would be added to the loan, unless one were right at the end of the policy or the amount was very small. Furthermore, any refund ought to go back to Premfina and could not be satisfactorily paid unless it was. Unless the payment data were available this could not happen, and as a result Mulsanne (or its intermediary) could not take over the administration because they could not fulfil customer expectations and comply with the duty to act fairly. Accordingly there was a breach of clause 13.9.1 (or at least I think that is the case - that was not said expressly in written final submissions).

Mulsanne seems to advance the same case for the wider claim to have access to or an introduction to Premfina. It is said to need that so that it can administer the policies for an MTA.

As will appear from a later section of this judgment, I do not consider that clause 19.3.1 imports any obligation to do any of these things. Furthermore, insofar as the claim is based on what was necessary to achieve the practicalities of administration, the provision of banking details alone would not solve the problem. All Mulsanne would have would be the customer’s banking details. That would not assist in adding a sum to the principal of the loan, and would not assist in getting money back to Premfina. Additional arrangements would have to be made, and Mulsanne’s activities are not advanced by the direct debit banking details. What Mulsanne would really need in order to overcome these problems is a relationship with Premfina and some sort of agency arrangement which would enable them to add to or repay the customer’s loan. That means it really has to make something of the claim to have some sort of access to Premfina.

At this point the claim fails again because the obligation that is asserted is too uncertain (among other reasons). It is not clear what was meant by the demands, or that anything Mulsanne might have wanted was achievable by Marshmallow anyway. If there was to be some sort of new arrangement with Premfina then Mulsanne would have to negotiate and enter into that arrangement itself. It is not apparent what it was that would have worked and would have been achievable; or that Marshmallow’s pro-active involvement would have been required anyway.

In addition, I consider that Mulsanne has made too much of the difficulties and problems involved. Even if there had been the difficulties which have been described, I do not consider that in practical terms they were all that difficult, or that it would have meant that Mulsanne could not have acted fairly vis-a-vis the policyholders. It would have been more inconvenient if extra premiums could not have been added to the original loans, but a policyholder was not entitled to expect that anyway. Refunds could have been paid direct to the customer (if necessary by asking him/her what his bank details were), and the customer could, if he/she wanted, have repaid Premfina or left the full loan running. The operation would not have been seamless, but it would not have been all that difficult and it would not have been unfair.

For all these reasons, therefore, I consider that there was no breach of any obligation in the non-provision of payment details relating to Premfina loans, and no breach in Marshmallow’s not taking some (unspecified) steps to achieve some sort of novation of the Premfina arrangements with Mulsanne.

The letter of claim of 11th February requires:

The response of NRF to this in its letter of 24 February 2021 was:

NRF changed tack in its letter of 28th April and said:

In further correspondence which follows (which it is unnecessary to set out in detail) HL continued to assert its client’s right to the information which was acquired in the course of the agency and NRF stated that its client was working on a way to provide the data in a “compliant” manner (without ever specifying the precise difficulties), and giving projected delivery dates which were never complied with. NRF also asserted that the data was not needed for the time being in order to service policies.

I should record that in Mr Hough’s skeleton argument, repeated in his written final submissions (see paragraph 511), there was a suggestion that the open banking information was held by Plaid and there was none held by Marshmallow. However, such a posture was not adopted in the correspondence, and it did not form the basis of Mr Hough’s submissions to me. Since Marshmallow has not shrunk from deploying a “we have not got it so we cannot give it” defence in relation to other items, and does not explicitly and clearly run it here (or in the rejoinder which refers to a willingness to give the data if certain issues can be resolved) I proceed on the basis that Marshmallow has open banking data. That is what one would expect anyway looking at the limited evidence about the point given to me.

Having come to that conclusion it is plain enough to me that it should have been disclosed on request. It was information acquired by the agent for the purposes of the agency, because it was deemed to be significant to the assumption of risk. It has the same quality as obviously important information such as the type of car or the age of the driver. It falls squarely within the common law or equitable duties to disclose, and squarely within clause 19.3.2.

The main reason for non-disclosure is said to be a data protection one. This concern was only ever expressed in general terms, and in the absence of some elaboration I do not see how it can provide a defence. The data is obviously sensitive, but one would have thought that if it was acquired by an agent for the purposes of the business of the principal then disclosure to the principal would be fairly obviously justified. Neither I nor Mulsanne were ever told why it was not. Nor is the consent of Plaid necessarily an obstacle. The mechanism via which Plaid might be able to impose a need for consent was not made apparent, and the Plaid contract was not disclosed by Marshmallow. If Marshmallow wished to rely on such matters the burden was on Marshmallow to justify them by the production of that and any other relevant documents (and to make a case under data protection law). Mulsanne is entitled to the information, and to make the matter clear, it should be full disclosure, and not subject to anonymisation as Marshmallow once suggested. The usefulness of this sort of data, without knowing to which policy it relates, would be seriously reduced if that link were not apparent to the person who assumed the risk.

The other point made by Marshmallow is that this information is now of no real use to Mulsanne because, as historic data which applied at one point of time, it would be of no use in an MTA or renewal. Mr McCauley agreed that it probably would not be of use on an MTA, but he did make the point that it was reasonable that he, as underwriter, should actually get the data he requested as part of the scheme, and I agree with him. It matters not to what purpose the principal wishes to put the data. As principal he/she is entitled to it and should have it. In any event, one can well imagine circumstances in which an insurer would wish to examine the data in order to assess the effectiveness of the portfolio and the open banking product.

This data has never been provided. I find that there was and continues to be a breach by Marshmallow in this respect.

I should make it clear that I am dealing only with data in the hands of Marshmallow. I am not dealing with data in the hands of Plaid. That was not the subject of the claim or the debate.

This is a point which arises in relation to several heads of the post-termination claim. Where Mulsanne seeks documents or materials held by third parties and which Marshmallow cannot itself provide Mulsanne seeks to expand its claim by reference to clause 19.3.1 as giving rise to an obligation (in substance) to put Mulsanne in a position in which it can administer the policies in the same way as Marshmallow did. Thus, for example, it is said that Mulsanne (or its new intermediary) should have had what it needed to make sure it could deal properly with add-ons where a customer required an MTA, and what it needed was full details of those products.

That line of argument makes it necessary to consider whether clause 19.3.1 imports any such obligation.

I consider that clause 19.3.1 cannot be construed or applied in this way. Clause 19.3 deals with what happens “in the event of a termination”. A number of sub-clauses then impose clear and direct obligations on Marshmallow when that happens, and they are clear enough and apparently a complete code (and note that the TOBA contains a complete agreement clause in clause 3). Clause 19.3.1 does not in its terms impose an obligation. It contains a reservation of right and expresses the consequences of that right being exercised. Those consequences (in that clause) do not include an obligation on Marshmallow to do anything. There are no express words in that sub-clause which contain an obligation, so any obligation would have to arise as a matter of implication in the contract.

If an implication is relied on it was not articulated in the Particulars of Claim or elsewhere. That poses difficulties in establishing an implied term because the implication is not clear, which will usually be fatal. The first step in alleging an implied term is to set out what it is. Without that it is not possible to apply any of the standard steps. This is not a mere pleading point. It lies at the heart of any debate about contractual implication. Since a term has not been set out Mulsanne cannot be allowed to rely on one.

However, I would add the following in relation to any attempt to imply a term which might assist Mulsanne. In Marks & Spencer plc v BNP Paribas Securities Services[2016] AC 742 Lord Neuberger considered a number of authorities. His starting point was to cite with approval the tests in various established authorities, and he obviously approved those formulations, while adding his own elaborations. He set them out in paragraph 16 of his judgment:

Without a clear formulation to work on it is not easy to apply those dicta to the case which Mulsanne apparently seeks to make out, but at the heart of all those formulations of a test as to the implication of terms are the notions of necessity (in a business efficacy sense), obviousness and the idea that both parties would have accepted the implication if it had been proposed to them. I have difficulty in identifying any term which would assist Mulsanne in this area which would fall within those parameters. A general obligation to do what was necessary to put Mulsanne in a position of being able to deal directly with the providers of third-party services would be far from obvious, because it cannot be supposed that Marshmallow would commit itself to something so vague and potentially dangerous. A series of specific obligations geared to the sort of services that were rendered would be equally vague and non-obvious. Furthermore, post-termination arrangements can be made to work without what Mulsanne says is missing; not as Mulsanne would really want them to work and not as conveniently as Mulsanne would wish, but they can be made to work. There would be some more administration where dealings in relation to Premfina loans, add-ons or credit cards, were concerned, but in my view Mulsanne would manage. The policyholders would know if add-ons or loans were involved and would be able to tell Mulsanne. The necessity test cannot be fulfilled, and it cannot be said that Marshmallow would obviously have agreed to specific or general implications which might assist Mulsanne at the outset. Mulsanne has express and implied terms which provide for it to have information, and with that information it could contact policyholders to invite them to renew or to apply to Mulsanne for MTAs, or for other purposes. It would have all the details it needed about its own policies. One can well imagine a serious argument from Marshmallow if it had been asked to commit, in advance, to a term requiring positive assistance or facilitation if it was apparent that that would require the delivery up of its own information, or a debate as to how it could “facilitate” arrangements with its own third party service providers. And as a final point, it should be noted that clause 11.10 anticipates lending by a third party but the agreement does not then go on to say anything about that other than what is in that clause.

I therefore do not consider that there is any implication arising out of this contract which would assist Mulsanne in the case it seeks to make.

This trial concerns “liability” with “quantum” to follow as appropriate (to use the wording in Marcus Smith J’s order). Mr Howe and Mr Christie urged on me that any issues as to causation should be confined to the quantum part of the exercise, and that I should confine myself to liability only. Liability means liability only. Causation questions should be treated as part of the inquiry as to damages (quantum) - what damage has been cause by the wrongs that have been committed. That is where the inquiry naturally belongs.

Mr Hough submits that I should have regard to causation and not order an inquiry, despite my findings of liability, for at least two reasons. First, the common law/equity claims require some damage as part of the ingredients of the cause of action, and none has been established at even a prima facie level - see Allied Maples Group v Simmons & Simmons[1995] 1 WLR 1602. Second, the court should not order an inquiry if it is not satisfied that Mulsanne has suffered loss which will realistically give rise to an award of damages - see Reed Executive plc v Reed Business Information Ltd[2004] RPC 40, and Beautimatic International Ltd v Mitchell International Pharmaceuticals Ltd[2000] FSR 267. Both of those considerations applied here. So far as the non-contractual claims are concerned, damages was an essential ingredient of liability, and Mulsanne had not established even a prima face case of loss. As to the second, the evidence showed (Mr Hough said) that Mulsanne did not want to take over the administration of the policies and was putting up false obstacles in the form of unreasonable demands (in the solicitors’ correspondence) in order to disguise that and in order to try to get damages instead of having to do the administrative work.

In my view Mr Hough has not succeeded in his objectives, or at least not yet. I consider that on a fair reading of the order of Marcus Smith J, and in particular in its context of an order for a speedy trial, and whatever the position might be in other cases as a result of Allied Maple, in this case the findings should be confined to technical liability matters without considering causation and prima facie damages material. A proper consideration of those additional matters would be likely to have required additional evidence and additional disclosure, and that evidence and disclosure was not available other than some limited evidence of Mr McCauley in chief and then some cross-examination of him. I do not believe that Marcus Smith J will have had such matters in mind when he set this case on its course towards an expedited trial.

Second, Mr Hough could not achieve his objective of no inquiry without dealing with the contractual claims anyway because technically they do not require a finding of damage as part of a finding of liability. He seeks to avoid an inquiry as to damages on the contractual post-termination breaches by saying that causation of loss, in the primary sense of a lost opportunity to get money from renewals by Mulsanne, can plainly be seen to be a lost cause because on the evidence he has established Mulsanne’s disinclination to take over the policies. While he may well have material which would enable him to advance that case, I do not think it right to make that finding on the evidence that I heard. Although Mr McCauley gave some evidence in cross-examination about the point, and some limited evidence in a witness statement which responded to Marshmallow’s point which AKB had made in a witness statement, I do not think that Mulsanne has had a fair opportunity to develop its case on the point. That may well be attributable to the fact that it was not taken in the Re-Amended Defence, which (in this respect) contained merely a non-admission as to damages. A positive case such as Mr Hough’s really ought to have been pleaded. I am not saying he is not entitled to run it; I am merely saying that I do not consider that it was properly developed at the trial.

Furthermore, Mr Hough’s case on this point does not take into account what remains of the pre-termination misuses of confidential information, so far as they have survived.

I shall therefore not determine at this stage that there will be no further inquiry as a result of this judgment. However, what I have decided does require that further consideration may have to be given to the point before a full inquiry is embarked on. My decisions on the extent of misuse of confidential information mean that, on that part of the claim, Mulsanne has not got much of its original claim left, and the extent of any financial remedy may be limited (of course, I do not decide the point). At last one of the successful springboard claims appears at the moment to be very insignificant in the context of this action. Some of those remarks also apply to the post-termination breaches. I have decided in Mulsanne’s favour on only two of those points, and whether they give rise to substantial damages is a question that will have to be addressed. Mulsanne will obviously want to reflect on these points, which I stress I have not yet decided.

The correct way of dealing with this point is for it to be properly addressed in the light of my actual findings at the hearing dealing with consequential matters. The parties can make submissions as to the scope, and indeed usefulness, of any inquiry bearing in mind that any inquiry would have to be confined to the limited infringements that I have found. I shall therefore not consider causation matters further at this stage, and will certainly not decide that there should be no inquiry.

This claim arises out of the circumstances in which Marshmallow started to offer its own policies on the occasion of the expiry of a Mulsanne policy and in place of that policy. It is said by Mulsanne that Marshmallow was guilty of passing off by substitution. The essential allegation is that when Marshmallow sent out renewal notices it was not made sufficiently clear that the proffered new policy was with a different underwriter, and indeed the renewal notices in some respects (it is said) represented that the insurer would be the same.

Mulsanne only sells through intermediaries. It has no directly customer-facing business in the conventional sense. Even when there are claims the customer goes to claims administrators known as Key Claims Ltd (a company in Mulsanne’s group), not to Mulsanne. When selling through some intermediaries Mulsanne’s own branding would appear fairly prominently in, for example, policy documents, and examples of that appear in the papers, but that, according to Mr McCauley, was unusual. It would seem that the more usual state of affairs was that the identification of Mulsanne as the insurer was more subdued and the usual state of affairs was that the polices would bear the branding of the intermediary or distribution channel.

That was the position with Marshmallow. The branding was very much the Marshmallow brand. Marshmallow’s name is the prominent name in all the documentation, though it is still the case that Mulsanne was identified as the insurer, albeit in small print, where appropriate or necessary. Thus the Marshmallow policy document has “Welcome to Marshmallow” in large print on the front page, and the first defendant’s name appears in a footer on every one of the 31 pages. Mulsanne was identified in the normal print of the document as the underwriter on page 4, and its corporate details are provided there, and the policy document bears a facsimile signature of someone who signs for Mulsanne; but this is in a section headed prominently “About your Marshmallow Insurance policy”. The only other reference to Mulsanne is in the section dealing with data use, which says that “Your insurer Mulsanne Insurance Company Ltd” would use data in accordance with its data privacy policy; but if the policyholder wanted any information about that he/she was to write to Marshmallow, not Mulsanne. Mr McCauley accepted that there was nothing unusual in this. It was usual for policies to bear the branding of the intermediary, though the insurer’s name would (usually) appear on the certificate and schedule.

That was the case with Marshmallow policies. Marshmallow’s name appeared prominently on quotations and Mulsanne was not mentioned in them, though the online quotations contained a link to “policy wording”, which I assume would take a customer to the document referred to in the previous paragraph of this judgment, which did refer to Mulsanne. Marshmallow’s name appeared prominently in large letters at the top of the insurance certificate. Towards the bottom of the page, in the normal font size (though in block capitals) Mulsanne was identified as the authorised insurer, and there is a statement that the policy was “arranged by” Marshmallow, who represented Mulsanne. Likewise, the Motor Insurance Schedule, which was to be read in conjunction with the certificate and policy documents, had Marshmallow’s name prominently at the top, and Marshmallow’s name (in tiny print) appeared in footers as being the correspondence address. Mulsanne was identified as the insurer in equally tiny print in a footer on page 1. Mr McCauley was prepared to accept, without looking at it, that the main branding of Marshmallow’s website was Marshmallow’s. AKB’s evidence was that the website of the aggregator Comparethemarket (where Marshmallow’s product appeared) did not even mention Mulsanne.

It is therefore apparent that the main branding of the Marshmallow product was Marshmallow, and that Mulsanne was mentioned only in the small print on odd occasions when it had to be.

That is important background to what happened from January 2021 when Marshmallow started selling its own policies. Mulsanne’s case rests on what was sent out by way of renewal notices. The notices sent out from January to March contained the following wording:

From about March 2021 a different version of renewal notice was sent out. These notices, so far as is material, read as follows with a prominent “marshmallow” at the head of the first page:

The starting point is the classic statement of Lord Oliver in Reckitt & Colman Products Ltd v Borden Inc (No 3)[1990] 1WLR 4981 at 499:

The first area of dispute in this case is whether Mulsanne had any relevant goodwill at all for the purposes of this claim. Goodwill was defined by Lord MacNaghten in IRC v Muller & Co’s Margarine Ltd [1901] 217 in the following terms (at pp223-4):

Goodwill does not necessarily have to be associated with a name - a business can have goodwill for these purposes even if customers do not have the actual name of the business in mind - Birmingham Vinegar Brewery Co Ltd v Powell[1897] AC 710; Wm Edge & Sons Ltd v Wm Nicholls & Sons Ltd[1911] AC 693. What is sufficient is that the customer is accustomed to receiving the particular goods or services in question.

I do not need to develop, in this section, the elements of misrepresentation and damage.

There is a dispute as to whether and to what extent Mulsanne had any relevant goodwill in this case. Mr Howe claimed his client had general goodwill in the market generally, and pointed to its large number of vehicle-years sold (over 300,000), its website presence and the number of claims handled. It is not clear that Marshmallow disputed the presence of goodwill in the UK market generally, and I am prepared to find that it had that general goodwill.

Marshmallow’s dispute was different. It claimed that Mulsanne had no relevant goodwill in relation to Marshmallow customers. Mr Moss’s case is that the Marshmallow insurance was all about Marshmallow, and nothing really to do with Mulsanne, as far as customers were concerned. He pointed to the prominence of Mulsanne’s name in some documents, and the low prominence given to it in other documents (as set out above). Although he did not put it in this way, I think the essence of his case was that the centrality of the Marshmallow branding forced out any customer association with Mulsanne. His submission was that customers were not really interested in who the underwriter was. They were interested in the fact that it would be Marshmallow. This submission was said to be supported by a comparison of the level of reviews of Marshmallow and Mulsanne on the Trustpilot review site. The reviews are said to show that as at 30th April 2021 Marshmallow (the first defendant) had higher levels of customer satisfaction from a higher level or reviews (4/6 out of 5, from 3,642 reviews) compared with a rating of 2.1 out of 5 from just 10 reviews for Mulsanne. In the 6 months that followed Marshmallow had garnered almost 4,000 more views (82.5% excellent) while Mulsanne’s reviews had gone up by 6. That was said to support the fact that in the context of Marshmallow policies Mulsanne had no material goodwill. It was, said Mr Moss, worthy of note that Mulsanne was seeking to rebrand Mulsanne’s policies under a new name - Abacai. That is said to show that customers give little or no weight to the name Mulsanne in the market place.

Mr Moss submitted that the case of Medgen v Passion for Life[2001] FSR 30 provided a powerful analogy with the present case, while acknowledging that all cases in this area ultimately depend on their particular facts. That was a case in which it was held that a distributor of goods, not the manufacturer, owned the goodwill in the product, and Mr Hough pointed to various findings of fact which showed how it was that the distributor was responsible for all the marketing, the packaging contained no reference to the manufacturer, the advertisements were effected by the distributor which “for the most part” carried the distributor’s name, and wholesalers and retailers would look to the distributor if there was a defect or problem. That was, he said, very like the present case.

Care must be taken in vesting too much significance in that authority. First, as was acknowledged, these cases depend on their facts, and all cases are different. Second, in the present case there is a potentially key difference in that Mulsanne’s name is mentioned in the documents - see above. Nor is it true that those documents are documents which only come into existence after the contract was concluded. As I have pointed out, and contrary to the submission of Mr Moss, there is some evidence that documents bearing the name of Mulsanne as underwriter were available to potential customers through the link in the quotation. Furthermore, references to Mulsanne in the post-contract documents are references in what become pre-contract documents where renewals are concerned. Accordingly, a lack of customer concern with who the underwriter was cannot be inferred from necessary ignorance at that stage. To that extent, at least, the present case is materially different from Medgen.

If it matters, I do not think it would be safe to infer that no customers would ever be concerned as to who the underwriter of Marshmallow’s policies was. Some may well have been interested enough to note the underwriters, even if after the policy documents were issued, and perhaps take comfort from that. On the other hand, I accept that the predominant connection will have been with Marshmallow. This means that if there was goodwill it was limited. However, I do not think it necessary to dwell on the point and consider whether that level of appreciation by customers gave rise to goodwill because even if it did and Mulsanne had some goodwill in relation to Marshmallow policies, Mulsanne’s case clearly fails at the misrepresentation stage.

I will deal first with the January-March renewal documents.

If that is to have occurred in this case then it must be by virtue of the renewal notices, the relevant terms of which have been identified above. It was submitted that they were confusing.

Mr Howe’s first point is that the transaction is described as a renewal, from which a customer would have assumed that his original insurance with the original underwriter was being continued into another year. That was wrong, because the underwriter was different. This effect on customers would be compounded by the fact that the policy number was the same. Thus there was a representation that the renewed policy was with the same underwriter as before, that is to say Mulsanne and there was passing off by substitution.

Next, Mr Howe relied on the statement that the policy would be underwritten, sold and managed by “the same company group”. To a customer who does not know that Mulsanne is not in the same group as Marshmallow this will indicate that the activities will be done by “the same group as before”. That, again, is said to indicate inaccurately that the same underwriter, Mulsanne, was involved.

Third, Mr Howe submitted that the statement about not being able to drive other people’s cars under third party cover gave the impression that the policy was being underwritten by the same entity but with a specific change of terms.

In my view these arguments fail. While it is obviously legitimate to look at individual elements to see what they say, the document as a whole also has to be looked at, and it is important not to ignore important elements, which Mr Howe’s submission does.

The most important element is the clear indication of “important changes”. Under this section only one “important change” is mentioned, namely the change of underwriter to Marshmallow Insurance. A reason is given - to deliver a better customer experience. Those customers who had noticed that Mulsanne was previously the underwriter cannot have been misled into thinking the underwriter was the same in the face of this statement unless there was something else in the document which indicated that it was in fact the same. There is nothing in the document which does that.

The word “renewing” might, by itself, be taken to connote a renewal with the same underwriter, but it is also capable of referring to a policy which is being replaced with one placed by the same intermediary. Chapter 6 of the FCA’s Insurance: Conduct of Business rules deals with Product Information, and Section 5 deals with what has to be disclosed on “Renewals”. A “renewal” is defined in paragraph 6.5 1(2) as:

That shows it is within industry understanding to use the term in the way that Marshmallow uses it. That sense of the word is particularly apt to the sort of situation in which the intermediary is promoting the scheme in the manner in which it was done in the present case. Mr Moss was able to produce a similar renewal notice from another broker (Devitt Insurance) for a broker branded policy which referred to a renewal into a different underwriter. Accordingly, the use of the word “renewing” is not inapt, it does not necessarily connote a further continuation of the Mulsanne underwriting and it in no way detracts from the clear wording about the change of underwriter.

Next Mr Howe relies on the “same company group” point. This point goes nowhere. The natural use of the wording is to describe the three activities as taking place in the same group, which describes accurately what was to happen. It had not happened under the Mulsanne regime. For someone who never noticed the references to Mulsanne in the previous documentation it will not be taken to be “the same as before” in a way which refers to Mulsanne because that person will not have Mulsanne in mind. For someone who did notice Mulsanne, the statement of the change of underwriter will have a greater significance, and that person will not be led into thinking that Mulsanne was still doing the underwriting.

The maintenance of the policy number does not assist Mr Howe. It is incapable of overriding the clear statement of the change of underwriter. What the notice means is that despite the change of underwriter, there will be no change of policy number.

The reference to not being able to drive third party cars indicates a change in cover in the new policy. When the document is read as a whole it does not indicate that the underwriter remains the same but the scope of cover is changing.

For all those reasons Mr Howe’s case that there was a misrepresentation in the January-March renewal notice fails.

I record that while both parties produced the results of institutional surveys and reports on what people did on renewals by way of reading and checking, I did not find that evidence particularly helpful.

The case for a misrepresentation in the April renewal notice is even poorer. This document expressly discloses that a new insurer is on the scene, and the old one (“an external third party”) is not. There is absolutely no case for saying that this contained any form of representation that the old underwriter’s product was continuing, and Mr Howe’s written final submissions did not seem to attempt to make one.

For those reasons the passing off claim fails and falls to be dismissed.

My conclusions on the diverse claims in this action which survived as far as final submissions are therefore as follows:

Items DM-2.3A and DM-2.3B – these items lack confidentiality and the claim fails.

Item DM-2.4 – confidentiality has been established and this item is available for springboard events involving the engine in which it was incorporated at the time.

Item DM-2.5 – confidentiality has been established and this item is available for springboard events involving the engine in which it was incorporated at the time.

Item DM-2.6 – this engine contained confidential material at the time it was used for springboard use, namely the claims and conviction file and vehicle segmentation file. There is no claim for misuse of the underwriting rules, whether an ongoing use claim or a springboard claim.

Item DM-2.9 – confidentiality has been established as part of item DM-2.5, as part of that item it is available for springboard events involving the engine in which it was incorporated at the time.

Item DM-2.10 – confidentiality and incorporation has been established and this item is available for a springboard event involving the engine in which it was incorporated at the time.

Item DM-2.12 – this claim fails.

Item DM-2.13 – Open banking price specification – two specific ongoing use claims fail. Insofar as this specification was used on a springboard event, and contained matters which I have held to be available for a springboard event, then those tables are similarly available.

There was misuse of confidential information in February 2020 when the ratings engine was used to generate a pattern of quotes for the purposes of information provided to a potential re-insurer.

There was no misuse of a Mulsanne postcode file in or about June 2020 when Mr Hemming conducted a postcode file comparison exercise.

There was a misuse of confidential information when analyses were done to support the activities of Ms Coulthard in or about November 2020 in preparing a vehicle modifications table and a voluntary excess table. The misuse was of such parts of the ratings engine as contained confidential information as found in this judgment, though the end effect or benefit may well be insignificant.

The TOBA was terminated on 2^nd February 2021 by Mulsanne’s election to do so under clause 19.2.7.

Marshmallow was not in breach of the TOBA by dint of its refusing an audit.

Marshmallow was in breach of the TOBA for failing to deliver up credit checks, ID checks and the results of Lexis Nexis NCD checks timeously.

Marshmallow was in breach of the TOBA for failure to deliver up open banking data held by it.

Marshmallow was technically in breach of the TOBA in failing to provide the last 4 credit card digits, but not otherwise in breach of any obligation in respect of credit card details.

Marshmallow was not in breach of the other provisions of the TOBA as alleged.

The passing off claim fails.

The next steps in this action will be determined when the parties have had a chance to reflect on this judgment.