Reading time: 7 minutes
Key points
- The uphold rate by product is the default headline metric in complaints MI because it is easy to calculate and looks crisp. On its own it is a thin signal.
- Uphold rates lag the underlying issue, are noisy on small samples, are sensitive to mix shifts, and conflate two very different problems — a firm getting decisions wrong, and a firm interpreting a rule differently to the ombudsman.
- A defensible starting basket of complementary measures is: rate of change versus peer median; share of cases citing vulnerability; share of cases involving professional representatives; and the redress band distribution.
- The right complaints MI is a small basket of measures read together — not a single headline number.
The thin headline
Most boards see one number in their complaints MI: the uphold rate by product. It is the default because it is easy to calculate, it feels objective, and it has been the industry-standard headline for years.
It is also a thin signal on its own. Uphold rates do not move quickly enough to drive action. They are vulnerable to volatility on small samples. They mask important variation in severity and root cause. They tell you nothing about direction — whether the firm is improving, holding steady, or drifting against peers. And they conflate two very different things: a firm getting decisions wrong, and a firm interpreting a rule differently to the ombudsman. Those have very different remediation pathways.
This piece sets out where the headline rate falls short, and four complementary metrics that risk and compliance, MI and complaints leadership teams should be reading alongside it.
Why the headline uphold rate falls short
They lag. A FOS decision is published months after the underlying customer interaction. By the time an uphold rate moves, the issue that caused it is well-established. A leading indicator this is not.
They are noisy on small samples. A product line with twelve cases per quarter can swing between a 25% and a 50% uphold rate without anything having changed in handling quality. The FOS Watch Insight Hub suppresses incomplete months and very small samples for exactly this reason. Most firm-internal MI does not, and reads the noise as signal.
They are sensitive to mix. A portfolio-level uphold rate can rise or fall because the mix of cases shifted, not because anything changed in handling. A surge in one product line — perhaps because of a market event the firm did not cause — can move the headline number without anything being newly wrong.
They carry no directional information. A 40% uphold rate by itself does not tell you whether you are getting better, getting worse, or holding steady. It does not tell you whether peers are moving the same way. Without direction and comparison, a rate is just a number.
They conflate decision-quality and judgement-divergence. A high uphold rate can mean two very different things. The firm may be reaching the wrong answer on cases that should have been resolved at first response. Or the firm may be reaching a defensible answer that the ombudsman simply interprets differently. Both matter. They call for different responses — one is a training and process problem, the other is a rule-interpretation and escalation problem.
They do not capture severity. One hundred upheld cases averaging £150 of redress and one hundred upheld cases averaging £8,000 of redress have the same uphold rate. They do not represent the same risk.
None of this is an argument against tracking uphold rates. It is an argument against reading them alone.
Four metrics that matter more
1. Rate of change versus peer median
The most important question your complaints MI can answer is not “what is our level?” but “where are we heading, and are we heading there with the market or against it?”
The metric: month-on-month or quarter-on-quarter change in your uphold rate, indexed against the median change for a defined peer group over the same period.
What it tells you that the headline rate does not: direction and divergence. A firm whose uphold rate is rising while the peer median is flat has a firm-specific problem. A firm whose uphold rate is rising in line with peers may be exposed to a market-wide pattern that warrants different action — different conversations with the FCA, different remediation framing, different board narrative.
This is also the metric closest to the FCA’s own approach to oversight. The regulator does not judge a single firm in isolation; it benchmarks against the cohort.
2. Share of cases citing vulnerability
The metric: the proportion of FOS cases involving your firm where the decision text references vulnerability — financial difficulty, bereavement, mental health, capacity, communication needs, or other Consumer Duty Outcome 4 indicators.
What it tells you that the uphold rate does not: whether vulnerable customers are disproportionately reaching the ombudsman. That is both a Consumer Duty governance signal and a leading indicator of operational gaps. Vulnerability is a multiplier on conduct risk — a firm with a rising vulnerability share is producing harder cases, not just more cases.
The FCA’s vulnerability work has sharpened in 2026, including the joint FCA / ICO statement of 27 March 2026 on firms’ approaches to vulnerability. FOS decisions are one of the few external surfaces where the firm’s actual treatment of vulnerable customers can be observed independently.
3. Share of cases involving professional representatives
The metric: the proportion of your FOS cases brought by a professional representative — a claims management company, solicitor, or other paid intermediary — rather than directly by the consumer.
What it tells you that the uphold rate does not: whether internal complaint handling is failing earlier in the journey. A high or rising PR-led share typically signals that consumers are not getting the answer or the experience they wanted at final response, and are seeking help to escalate.
Hypothesis (label it): PR-led complaints tend to involve different fact patterns than direct complaints — more documentary evidence, more confident rule-citation, and a different uphold profile. Firms that track PR share alongside the uphold rate have more granular visibility into where their final response process is breaking down. This is a working view from observed FOS patterns rather than a tested claim across the market.
4. Redress band distribution
The metric: the distribution of FOS-awarded redress against your firm by band — for example, £0–£500, £501–£2,000, £2,001–£10,000, £10,001+ — broken down by product and by quarter.
What it tells you that the uphold rate does not: severity. A firm with a heavy tail of high-value redress awards has a different risk profile to a firm with a flat distribution at the lower end. The shape of the distribution also tells you about case complexity, the seriousness of the underlying harms, and the financial exposure being created.
This is the metric that translates most cleanly into the language of risk appetite. A board can discuss “high-value tail exposure” in a way it cannot discuss “the uphold rate is 38%”.
How to use them together
The right complaints MI is not a single headline number. It is a small basket of measures, each answering a different question, read together. A defensible board pack would include all five:
| Metric | Question it answers | Cadence |
|---|---|---|
| Uphold rate by product | Level, by product | Monthly |
| Rate of change vs peer median | Direction and divergence | Monthly |
| Vulnerability share | Consumer Duty Outcome 4 governance | Monthly |
| Professional representative share | Final response health | Monthly |
| Redress band distribution | Severity and exposure | Quarterly |
The FCA’s February 2026 good practice guidance on Consumer Duty board reports favours MI that tracks outcomes, supports decision-making, and shows action taken — rather than dashboards that report numbers without narrative. A basket of complementary measures is closer to that standard than a headline rate alone.
Building these measures in practice depends on having FOS data structured consistently — tagged for product, root cause, vulnerability references, representative type, and redress band. Tools that tag decisions against a deterministic taxonomy make this easier, because the underlying numbers are reproducible and auditable. FOS Watch was built for this work and developed with three UK Tier 1 lenders and insurers.
In short
Uphold rates are not wrong. They are thin. A complaints function that reports only the headline rate is leaving most of the available signal on the table — and producing board narratives that are weaker than they need to be.
Four complementary measures — rate of change versus peer median, vulnerability share, professional representative share, and redress band distribution — turn a single number into a basket. Each is straightforward to calculate from FOS data tagged consistently. None requires new infrastructure. All sharpen the conversation.
The bar for complaints MI under the Consumer Duty has moved. The metrics on the board pack should move with it.
Sources
- Financial Conduct Authority, Consumer Duty board reports: good practice and areas for improvement, February 2026.
- Financial Conduct Authority, Consumer Duty pages (joint FCA / ICO vulnerability statement, 27 March 2026), accessed 27 May 2026.
- Financial Conduct Authority, DISP — Dispute Resolution: Complaints sourcebook, accessed 27 May 2026.
- Financial Ombudsman Service, Annual complaints data, accessed 27 May 2026.
This article is intended as analysis and commentary, not as a substitute for legal or compliance advice on specific circumstances. Atago Green is a UK-based software provider and is not a regulated firm.