Class actions continue to grow throughout Europe, and the United Kingdom remains the most developed market for them[i]. In this series we look at class actions jurisdiction by jurisdiction, not from the perspective of the law, but from that of the data underneath it: what has to be found, preserved, reviewed and produced to establish liability, and what has to be verified, matched and paid out once liability is established or a settlement is agreed.

Our earlier article, Data in class actions: know your case, understand your class, described a class action as a data problem in two phases: liability, where a party must know and produce its own record, and distribution, where claims must be verified, valued and paid at scale[ii]. This series applies that lens country by country, starting with the UK, where the second phase is proving harder than many expected.

The UK is a useful place to start for two reasons. First, disclosure is the default rather than the exception. Under Practice Direction 57AD, a party must disclose the adverse documents it knows of, including those that damage its own case, whereas most continental systems usually order the production of specific documents only, even after recent EU reforms. The liability phase therefore bites harder in the UK than almost anywhere else in Europe. Second, the UK is the European jurisdiction where opt-out distribution has been tested furthest in practice: a settlement has been distributed, damages have been awarded after trial, and the results are now visible. What happens in the UK today is likely to shape what claimants, defendants and courts expect elsewhere in Europe.

Key Takeaways

  • The mechanism sets the data challenge. Opt-in proceedings start with an intake problem. In opt-out proceedings the question is whether the class can be matched to identifiable people at all.
  • Disclosure is the default. Unlike in most of continental Europe, parties in the UK must disclose documents that hurt their own case.
  • Human in the loop. The CAT accepts AI-assisted review but expects human checks and a protocol that can be shown.
  • Distribution remains a major unresolved challenge. In the first CAT opt-out case to reach distribution, class members claimed around £216,500 of up to £25 million. The route to class members should be planned at certification, not after judgment.

A £25 million settlement, £216,500 claimed

The first opt-out case before the Competition Appeal Tribunal (CAT) to reach the distribution stage was Gutmann v Stagecoach South Western Trains, concerning boundary fares[iii]. The settlement made up to £25 million available to the class. Class members claimed around £216,500[iv]. Of the unclaimed balance, which could otherwise have gone towards the class representative’s costs, £3.8 million was paid to the Access to Justice Foundation[v]. Claimants had to show they had bought a particular type of ticket, and many who started the claims process dropped out when asked for their bank details[vi].

For the class representative the outcome was a win, but for most class members it made no difference. Nobody held a list of who had bought the relevant tickets, few class members had kept proof of a journey made years earlier, and every extra step in the claims process cost participants. A legal outcome only becomes compensation through a workable process of identification, verification and payment. Data underpins that process, but its design decides whether people complete it.

The larger cases now follow the same path. In Merricks v Mastercard, the CAT approved a £200 million settlement, of which at least £100 million is set aside for a potential class of some 44 million people who shopped in the UK between 1992 and 2008[vii]. Once registration opens, class members will have six months to register following a national awareness campaign, and the fund will be divided equally among them[viii]. At the time of writing, registration had not yet opened. Payments, estimated at roughly £45 to £70 each, will follow the close of the registration window, and whether that estimate holds depends on how many people come forward[ix].

In Kent v Apple, the first opt-out case to result in a damages award after trial, the CAT found in October 2025 that around 36 million iPhone and iPad users had been overcharged and awarded approximately £1.5 billion[x],[xi],[xii]. Its order of July 2026 leaves distribution for later: the class representative must propose next steps within two months of the end of any appeals[xiii]. How to reach 36 million people is still an open question. Unlike in Merricks, however, the defendant holds account-level purchase records for the class. That does not mean everyone can be found and paid, but it is a very different starting point.

In many opt-out cases, the class exists only in data until someone has to be paid.

The UK regimes, seen through data

The UK has four routes to collective redress. The legal detail is well covered elsewhere. For data purposes, the first distinction is whether class members opt in or are included automatically, because that sets the initial data challenge.

MechanismForum and scopeOpt-in or opt-outWhat it means for data
Group Litigation Order (CPR 19, Section III)[xiv]High Court; any cause of actionOpt-inClaimants are known from the start. The challenge is intake: identity checks, a group register, duplicates and UK GDPR compliance for claimant data.
Representative action (CPR 19.8)High Court; any cause of actionOpt-out, but the ‘same interest’ test is strictRarely viable for damages after Lloyd v Google and Prismall v Google[xv][xvi]. If loss can only be shown person by person, the class cannot be framed this way.
Collective proceedings (Competition Act 1998, s 47B)[xvii]CAT; competition claims onlyOpt-out for UK-domiciled class members; opt-in for othersThe class is defined and sized from data before certification. Economists need sales, price and margin data[xviii], and the Tribunal looks at how compensation will reach the class.
Group proceedings (Scotland, 2018 Act)[xix]Court of SessionOpt-in (the Act allows opt-out, but no court rules for it exist yet)[xx]Similar to a Group Litigation Order: an intake problem. Recovery of documents is narrower and must be ordered by the court[xxi].

In opt-in proceedings the data work starts immediately: collecting identity and eligibility information from thousands of claimants, removing duplicates across claimant firms and claims-gathering platforms, and handling it all under UK GDPR[xxii]. In opt-out proceedings the class is defined at certification from sales volumes, prices and user numbers, and the decisive question is a different one: can the class definition be matched to identifiable people using records that actually exist? Where the defendant holds account-level data, as in Kent, much of the class may be identifiable. Where the available records cannot reliably link purchases to individuals, as in Merricks, class members largely have to come forward themselves. Most cases sit somewhere in between, and where a class sits on that spectrum shapes everything that follows, from disclosure to distribution.

Liability: knowing and producing your record

For defendants: the first weeks matter

The duty to preserve potentially relevant information arises as soon as litigation is contemplated, and UK class actions tend to reach far back in time. Merricks covers 16 years and Kent nine. For the defendant, that means legacy systems, archived mailboxes, decommissioned enterprise systems and billing platforms, and former employees whose accounts may already have gone. In our experience, the first practical question is rarely ‘what does the data show?’ but ‘does the data from 2012 still exist, and can anyone still open it?’

In the Business and Property Courts, disclosure follows Practice Direction 57AD, which is issue-based and requires the parties to agree how technology will be used before the first case management conference[xxiii]. The CAT applies its own rules and tends to order disclosure in stages, in proportion to what is needed[xxiv]. In collective proceedings, much of the early disclosure is not email at all. It is structured data for the economists: transaction records, prices, commissions and margins, often over a decade or more. That data needs a data dictionary, a documented extraction method and results that can be reproduced, because the experts on the other side will test every assumption.

Where documents do need reviewing, the courts have been receptive to technology for some time. Technology-assisted review was approved in Pyrrho in 2016[xxv]. In Gormsen v Meta, the CAT held in December 2025 that AI-intensive review tools are not generally inappropriate, provided stringent human checks are in place, and that using AI to supplement human review cannot be considered experimental[xxvi]. It also placed responsibility for monitoring the tools on a designated solicitor and made clear that the review protocol may have to be disclosed if the CAT needs it[xxvii]. For anyone running a review, that means keeping a record of how the tools were tested and checked from the start, because the CAT may ask for it.

For claimants: data to get through certification

At certification, the CAT asks whether the class representative’s expert has a credible methodology and whether the data to apply it is likely to be available[xxviii]. Claimant teams therefore need an early data map: what is public (published commission rates, market reports, regulatory decisions), what sits with the defendant and must be requested, and what can only come from class members themselves.

The defendant’s experts will look closely at the data the claimant’s expert relied on, not only at the methodology. Claimant teams should keep a clear record of every dataset an expert receives: its source, its version and what was done to it. If the other side can show that a dataset was changed along the way without anyone recording it, the expert’s conclusions become much easier to attack.

Distribution: from a class on paper to people who get paid

The CAT has made clear that it looks at whether compensation can realistically reach class members, at certification and when approving a settlement[xxix]. In Gutmann, it went further and indicated that class representatives should provide evidence of likely take-up from the outset. That turns distribution from an afterthought into a data question that must be answered early. Four parts of it stand out.

  • Finding class members. In an opt-out case, there is often no list. Notice depends on media campaigns unless someone holds contact details, and using a defendant’s customer data to contact class members raises its own data protection questions[xxx].
  • Verifying claims. Proof requirements can deter fraudulent claims, but they can also deter honest ones, so the trade-off between verification and participation must be weighed deliberately. Gutmann shows how quickly take-up falls when people are asked for documents they no longer have.
  • Paying out. Bank transfers, vouchers and account credits each need different data. Merricks will offer bank transfer or gift vouchers; a defendant with live customer accounts could in principle credit them directly.
  • Dealing with what is left. Unclaimed funds can go to charity, revert to the defendant or go elsewhere, depending on the order or settlement[xxxi]. Low take-up is not only a fairness problem; it shapes how the CAT views the whole case.

How much data distribution needs also depends on how compensation is calculated. In UK collective proceedings, the CAT may award damages without assessing what each class member individually lost[xxxii]. That aggregate approach is what allows a scheme such as Merricks to pay every eligible person the same amount, so that distribution only has to establish eligibility. Where compensation follows each person’s own transactions or loss, the scheme needs records detailed enough to value every claim individually, and those records rarely sit with claimants. Account-level data, as in Kent, is a strong starting point, but it does not by itself mean that every eligible person can be reached or paid.

Even with good records, getting from a purchase record to a payment takes a fair amount of work. Historical transactions must be matched to an eligible person, including people with several accounts, shared family accounts or changed details. Usable contact and payment information must be separated from records that are years out of date. The eligibility and compensation rules must be applied consistently, and uncertain matches routed to human review rather than paid or rejected automatically. Each step must be documented, and each has data protection implications. If this work is done carefully, a large part of the class can be paid directly. If it is rushed, eligible people may still miss out, even when the defendant’s records would have allowed them to be found.

This is the same point our first article made from the defendant’s side. The records a defendant masters to defend liability are the reference data that later make verification and payment possible. For defendants, that is also a source of influence over how a settlement is designed. For claimants, it is a reason to put distribution data on the agenda well before judgment.

Two trends and what they mean for data

Funding uncertainty puts data earlier in the process. Since PACCAR (2023), funders have become more selective[xxxiii]. Legislation to reverse it was announced in December 2025 but left out of the King’s Speech in May 2026[xxxiv],[xxxv]. An early, data-based assessment of class size, likely damages and the availability of evidence is increasingly part of the case for funding, and gives defendants an early view of their exposure.

Cross-border claims bring cross-border data. Municipio de Mariana v BHP, in which the High Court held BHP liable in November 2025 for a dam collapse in Brazil, combines opt-in intake from hundreds of thousands of claimants with evidence in other languages and under other data protection regimes, such as Brazil’s LGPD[xxxvi].

Practical tips for UK class actions

For defendants

  1. Map the claim period to your systems. UK claims often cover ten years or more. Identify legacy and decommissioned systems, and who can still access them, before the first case management conference and before a Disclosure Review Document has to be agreed.
  2. Assume adverse documents will be disclosed. Under PD 57AD, known adverse documents must be disclosed whichever disclosure model applies. Know what they are before the other side does.
  3. Suspend deletion and record that you did. Include chat, mobile devices and the accounts of departing employees, and keep a record of when and how holds were applied.
  4. Treat structured data as evidence. Prepare a data dictionary, a documented and repeatable extraction method, and version control for every dataset produced to the other side.
  5. Write the review protocol as if the Tribunal will read it. After Gormsen v Meta, assume it might.

For claimants and class representatives

  1. Build the data map before certification. The CAT expects a credible methodology and evidence that the data to apply it exists. Separate what is public, what the defendant holds and what only class members can provide, and plan disclosure requests accordingly.
  2. Keep a chain of custody for expert data. Record every dataset given to the experts, its source, version and any transformation.
  3. Design opt-in intake properly from the start. Identity checks, de-duplication and UK GDPR compliance are much harder to add once thousands of claimants have signed up.

For both sides: distribution

  1. Plan distribution at certification, not after judgment. After Gutmann, expect the CAT to ask for evidence of likely take-up from the outset, and establish early whose data could reach class members directly.
  2. Keep the claims process light. Each additional proof step can reduce take-up. Apply baseline checks to every claim and reserve enhanced scrutiny for anomalies.
  3. Measure and report. Track take-up, rejection rates and fraud indicators throughout, so you can show how the process is performing if the Tribunal or the class asks.

Conclusion

The UK shows both phases of a class action in practice. In the liability phase, the CAT has accepted AI-assisted review and much of the evidence is structured data, but both come with an expectation of validation, reproducibility and transparency. In the distribution phase, the UK is still searching for answers, and the first results show that winning a case and getting compensation to people are two different achievements. As the first European jurisdiction to test opt-out distribution at scale, the answers it finds are likely to set expectations elsewhere.

For both claimants and defendants, the lesson is the same: the records used to fight a case are often the records needed to deliver compensation, and decisions about them made in the first weeks determine what is possible at the end.

About the Authors

Mathieu van Ravenstein
Partner at FORCYD
Mathieu is co-founder and owner at Forcyd. He has over 25 years of experience in legal and policy advice and eDiscovery. He has worked with global corporations, law firms and international institutions advising on many high-profile and complex investigations and disputes as well as on litigation in areas such as corporate fraud, money laundering, competition law and international trade.

vanravenstein@forcyd.com
+31(0)6 458 727 46
Julian Geraerds
Cyber & eDiscovery Expert at FORCYD
Julian is an expert in legal technology and digital investigations. He is experienced in complex, large-scale matters relating to litigation, regulatory compliance, internal investigations and arbitration. He works with large corporations and leading law firms, focusing on high-stakes issues such as competition, anti-corruption and bribery, and data breaches.

geraerds@forcyd.com
+31 (0)6 398 111 71
Sofia Piazza López
Cyber & eDiscovery Analyst at FORCYD
Sofia specialises in the intersection of data, technology, and legal proceedings. She supports data-intensive litigation, investigations, and regulatory matters. Her interests include class actions and collective redress, regulatory compliance, and cybercrime with a particular focus on how emerging risks and evolving regulatory frameworks influence modern dispute resolution and investigative practices.  

piazzalopez@forcyd.com
+31 (0)6 413 578 08

The authors would like to thank Liwia Jarczewska, who contributed to this piece.


[i] CMS, European Class Action Report 2025, https://cms.law/en/gbr/publication/cms-european-class-action-report-2025

[ii] FORCYD, Data in class actions: know your case, understand your class (July 2026), https://forcyd.com/data-in-class-actions-know-your-case-understand-your-class/

[iii] CAT, Gutmann v First MTR South Western Trains, Judgment (SSWT Collective Settlement), 10 May 2024, https://www.catribunal.org.uk/sites/cat/files/2024-05/13047719%20Justin%20Gutmann%20v%20First%20MTR%20South%20Western%20Trains%20Limited%20and%20Another%20-%20Judgment%20%28SSWT%20Collective%20Settlement%29%20%2010%20May%202024_0.pdf

[iv] Legal Futures, CAT approves £3.8m payment to charity from unclaimed damages (September 2025), https://www.legalfutures.co.uk/latest-news/cat-approves-3-8m-payment-to-charity-from-unclaimed-damages

[v] CAT, Gutmann v First MTR South Western Trains, Judgment (Stakeholder Entitlement) [2025] CAT 72, 7 November 2025, https://www.catribunal.org.uk/sites/cat/files/2025-11/13047719%20Justin%20Gutmann%20v%20First%20MTR%20South%20Western%20Trains%20Limited%20and%20Another%20-%20Judgment%20%28Stakeholder%20Entitlement%29%20%207%20Nov%202025.pdf

[vi] The Law Society Gazette, Are class members out of reach?, https://www.lawgazette.co.uk/commentary-and-opinion/are-class-members-out-of-reach/5127061.article

[vii] Slaughter and May, Merricks v Mastercard: settlement approval decision, https://www.slaughterandmay.com/insights/new-insights/merricks-v-mastercard-settlement-approval-decision/

[viii] Global Legal Post, Merricks prepares to distribute £100m to consumers after settling dispute with Mastercard class action funder, https://www.globallegalpost.com/news/merricks-prepares-to-distribute-ps100m-to-consumers-after-settling-dispute-with-mastercard-class-action-funder-1889650995

[ix] New Law Journal, Merricks approaches payout day (12 August 2026), https://www.newlawjournal.co.uk/content/merricks-approaches-payout-day

[x] CAT, case page Dr Rachael Kent v Apple Inc and Apple Distribution International Ltd, https://www.catribunal.org.uk/cases/14037721-dr-rachael-kent

[xi] LexisNexis, Competition Appeal Tribunal makes first collective proceedings damages award (Kent v Apple), https://www.lexisnexis.com/en-gb/legal/news/competition-appeal-tribunal-makes-first-collective-proceedings-damages-award-kent-v-apple

[xii] Macfarlanes, Competition Appeal Tribunal rejects claim against BT in first collective action to reach full trial, https://www.macfarlanes.com/insights/102jrtl/competition-appeal-tribunal-rejects-claim-against-bt-in-first-collective-action-t

[xiii] CAT, Kent v Apple, Order of the Tribunal, 27 July 2026, https://www.catribunal.org.uk/sites/cat/files/2026-07/14037721%20Dr.%20Rachael%20Kent%20v%20Apple%20Inc.%20and%20Apple%20Distribution%20International%20Ltd%20-%20Order%20of%20the%20Tribunal%20%2027%20Jul%202026.pdf

[xiv] Civil Procedure Rules, Part 19: Parties and Group Litigation, https://www.justice.gov.uk/courts/procedure-rules/civil/rules/part19

[xv] UK Supreme Court, Lloyd v Google LLC [2021] UKSC 50, press summary, https://supremecourt.uk/uploads/uksc_2019_0213_press_summary_67375c4f68.pdf

[xvi] Prismall v Google UK Ltd [2024] EWCA Civ 1516, https://caselaw.nationalarchives.gov.uk/ewca/civ/2024/1516

[xvii] Competition Act 1998, section 47B, https://www.legislation.gov.uk/ukpga/1998/41/section/47B

[xviii] Clifford Chance, The blueprint (certification of collective proceedings, October 2023), https://cliffordchance.com/content/dam/cliffordchance/briefings/2023/10/the-blueprint.pdf

[xix] Civil Litigation (Expenses and Group Proceedings) (Scotland) Act 2018, section 20, https://www.legislation.gov.uk/asp/2018/10/section/20

[xx] Law Society of Scotland Journal, Why Scotland is weighing bold expansion of group proceedings five years on (February 2026), https://www.lawscot.org.uk/members/journal-hub/articles/why-scotland-is-weighing-bold-expansion-of-group-proceedings-five-years-on

[xxi] Shepherd and Wedderburn, Recovery of Evidence (Scotland) v Disclosure (England) – a tale of two countries (August 2024), https://shepwedd.com/knowledge/recovery-evidence-scotland-v-disclosure-england-tale-two-countries/

[xxii] ICO, Lawful basis for processing, https://ico.org.uk/for-organisations/guide-to-the-general-data-protection-regulation-gdpr/lawful-basis-for-processing/

[xxiii] Practice Direction 57AD, Disclosure in the Business and Property Courts, https://www.justice.gov.uk/courts/procedure-rules/civil/rules/part-57a-business-and-property-courts/practice-direction-57ad-disclosure-in-the-business-and-property-courts

[xxiv] The Competition Appeal Tribunal Rules 2015 (SI 2015/1648), https://www.legislation.gov.uk/uksi/2015/1648/contents/made

[xxv] Pyrrho Investments Ltd v MWB Property Ltd [2016] EWHC 256 (Ch), https://caselaw.nationalarchives.gov.uk/ewhc/ch/2016/256

[xxvi] CAT, Gormsen v Meta Platforms Inc [2025] CAT 85, Ruling (Disclosure), 16 December 2025, https://www.catribunal.org.uk/judgments/14337722-dr-liza-lovdahl-gormsen-v-meta-platforms-inc-and-others-ruling-disclosure-16-dec

[xxvii] Hogan Lovells Cadwalader, Generative AI in UK disclosure: the rules haven’t changed yet, but the baseline is moving, https://hlc.com/en/publications/generative-ai-in-uk-disclosure-the-rules-havent-changed-yet-but-the-baseline-is-moving

[xxviii] Gateley, There’s method(ology) in the madness: expert methodology in CAT collective actions, https://gateleyplc.com/insight/article/theres-method-ology-in-the-madness-expert-methodology-in-cat-collective-actions/

[xxix] Hogan Lovells Cadwalader, Distribution in Collective Proceedings (Legal 500), https://www.legal500.com/guides/hot-topic/distribution-in-collective-proceedings

[xxx] ICO, Legitimate interests, https://ico.org.uk/for-organisations/uk-gdpr-guidance-and-resources/lawful-basis/legitimate-interests

[xxxi] Monckton Chambers, First CAT post-distribution decision (November 2025), https://www.monckton.com/first-cat-post-distribution-decision/

[xxxii] Competition Act 1998, section 47C (damages and costs in collective proceedings), https://www.legislation.gov.uk/ukpga/1998/41/section/47C

[xxxiii] Chartered Institute of Arbitrators, PACCAR: key issues for third party litigation funding agreements in England and Wales, https://www.ciarb.org/news/paccar-key-issues-for-third-party-litigation-funding-agreements-in-england-and-wales/

[xxxiv] Dechert, UK Government to legislate on litigation funding to mitigate the effects of PACCAR (January 2026), https://www.dechert.com/knowledge/onpoint/2026/1/uk-government-to-legislate-on-litigation-funding-to-mitigate-the.html

[xxxv] Akin Gump, UK Government omits PACCAR fix from 2026 King’s Speech (May 2026), https://www.akingump.com/en/insights/alerts/uk-government-omits-paccar-fix-from-2026-kings-speech-leaving-litigation-funding-reform-uncertain

[xxxvi] Judiciary of England and Wales, Municipio de Mariana v BHP Group [2025] EWHC 3001 (TCC), judgment and summary, 14 November 2025, https://www.judiciary.uk/judgments/municipio-de-mariana-v-bhp-group-2/