The Disputed Claim: What Firms Are Asked to Evidence

The stage that resists pricing

Article 6 closed on the cascade’s most volatile stage: the disputed claim, its cost set by the evidence held. This article takes that stage apart and defines the evidence.

Ask a claims team what a disputed case costs and the answer depends on a variable fixed before the dispute. If the evidence exists, the stage is administration. Without it, the stage is reconstruction, escalation and a decision made on what remains.

Evidence is an architecture decision

The instinct when a claim is challenged is to gather evidence and for one category that works. Policy documents, configuration records and terms presented can be produced months later, because they describe the system, not the incident. What this customer saw and did in the session where the compromise happened can only be captured at the time.

That asymmetry is why the disputed claim resists pricing. The evidence adjudication weighs most heavily is the evidence that cannot be gathered retrospectively. So the cost of the dispute stage is decided at architecture time, years before any individual claim.

Evidence of the event is captured at the time, or it does not exist.

Inside the disputed claim: where the burden falls

The pathway runs faster than most complaint processes

A complaint about a fraud or scam involving a payment service runs on a shorter clock than general complaint handling. The firm must reply within 15 business days, with a final response or an explanation of the delay.¹ A final response must follow within 35 business days, a shorter window than other complaint types allow.¹

If the customer disagrees or the clock runs out, the complaint can go to the Ombudsman. The case fee there is £680 after the £2,000 annual allowance.² Sending PSPs must also complete the service’s business response form for authorised-payment scam cases, a checklist of what the firm is expected to show.¹

A second ground appears in PSR scope: the consumer standard of caution, where the firm bears the burden of proof.³ Neither the caution exception nor the excess of up to £100 can be applied where vulnerability had a material impact on the customer’s ability to protect themselves.³ The Ombudsman also considers the FCA’s vulnerable-customer guidance in such cases; vulnerability raises the firm’s evidence bar, never lowers it.¹

Outside PSR scope the contested ground changes but the evidence question does not. For members who withdraw to a nominated account, the dispute arrives as a Duty question: was harm foreseeable and what did the firm do about what it could see (see article 5)? The Ombudsman’s fair-and-reasonable review and the vulnerable-customer guidance apply either way and the same event evidence answers both.

The volumes and the odds

The Ombudsman recorded 18,900 fraud and scams complaints about current accounts in 2025/26.⁴ Across all products the overall uphold rate was 30 per cent.⁴ Current accounts were the most complained-about product between April and June 2026, with fraud and scams accounting for half.⁵

Fraud complaints have historically run above the all-product uphold average.⁶ In the first quarter of 2024/25 the service upheld 44 per cent of them, against a 37 per cent average.⁶ Cases under the former CRM Code ran at 49 per cent against 36 per cent outside it.⁶

What firms are asked to evidence

Configuration or event

The Ombudsman’s guidance on authorised-payment scams is explicit: each case turns on its own facts and requirements vary with it.¹ Its unit of adjudication is the individual incident, not the control environment. A firm can hold a complete record of its control environment and still not answer the questions the incident raises.

In an authorised-payment scam complaint, the service states it will look at “the facts and evidence from both you and your customer”.

Financial Ombudsman Service, guidance for businesses: APP fraud and other scams involving authorised payments or withdrawals

That single line carries the distinction this article rests on. Facts and evidence about the case are event evidence: what happened, to this customer, in this session. Configuration evidence describes what the system was built to do and answers a question nobody in the dispute is asking.

Five characteristics adjudication rewards

Assembled from the guidance and the response form’s expectations¹ and the reimbursement regime’s burdens,³ event evidence has five characteristics:

• Contemporaneous: captured at the time of the session, not reconstructed after the claim.
• Session-level: specific to what was displayed and done in the session in question, not what the journey was configured to show.
• Customer-specific: relevant to this customer’s circumstances, which is where vulnerable-customer considerations enter.
• Intervention-complete: recording not only that a warning or intervention occurred but how the customer responded.
• Trailed: a decision record showing what the firm knew, when and what it did, which is what the caution-exception burden requires the firm to discharge.

A generic warning log fails the second and fourth characteristics at once. It shows what was sent, not what was seen, engaged with or overridden and the case-by-case review asks about the latter.¹

The standard nobody has written

No regulator, standards body or industry code we know of has published an evidence standard for pre-transaction fraud interventions. The five characteristics above are assembled observations from adjudication guidance, the response form’s expectations and the reimbursement regime’s burdens. They are what the sources ask for, not a checklist anyone has endorsed.

That absence is a finding, not a footnote. The industry is accumulating adjudication outcomes against a standard nobody has codified. Firms diverging most sharply are, on our reading, those with the least event evidence to bring.

The variance: one regime, divergent application

Same rules, order-of-magnitude spread

Frontier Economics observed the divergence directly: most firms saw Ombudsman complaints on around 1 to 2 per cent of APP scam claims.³ Four firms in its sample ran at 4 to 13 per cent.³ On the caution exception, one firm rejected more than a quarter of claims while others rejected none.³

The regime those firms operate under is uniform; the application plainly is not. A spread that wide is hard to explain by fraud exposure alone. Our reading places it in claims and evidence practice, though the divergence itself is the evaluation’s observation, not our inference.³

The evidence audit

Five questions for your next risk meeting

The framework in article 6 prices the stages of a fraud case. These five questions test whether its most volatile stage is priced by you or for you:

1. For any given customer session, can you show what warnings were actually displayed, or only what the journey was configured to display?
2. Can you evidence whether the customer saw, dismissed or acted on an intervention and when?
3. If a claim from a customer in vulnerable circumstances were disputed, could you evidence the support offered at the time?
4. How long does assembling the evidence pack for one disputed claim take and from how many systems?
5. Which of the five characteristics could your architecture satisfy for an incident that happened yesterday?

Answers of “only configuration” to the first two are not a compliance failure; they are the state of most architectures. They are also why the disputed stage is volatile and where cascade cost meets system design.

The claims hardest to evidence and costliest to dispute concentrate in a specific customer cohort. That concentration is a risk-management fact, not a welfare framing and article 8 examines it.