Reading time: 7 minutes
Key points
- FOS decisions involving vulnerability in the context of collections, forbearance and default rose by around 70% year-on-year — from 3,276 cases in 1 April 2024 – 31 March 2025 to 5,559 cases in 1 April 2025 – 31 March 2026.
- Decisions involving vulnerability alongside bereavement or capacity references rose by around 87% over the same period — from 2,108 to 3,952.
- Yet uphold rates fell by about 16 percentage points across both cohorts — from around 43% to around 27%.
- The combined pattern — sharply higher volumes alongside materially lower uphold rates — raises three competing readings, each with different implications.
- The absolute number of upheld vulnerability cases is still rising in both cuts. The operational pinch points where they concentrate are clear, and have not changed.
The patterns
Atago Green’s FOS Watch tags every published FOS decision against a deterministic taxonomy — type of vulnerability cited, operational context (collections, bereavement, fraud, account closure and others), rules engaged, and outcome. Two structured cuts of the 2025/26 data produce complementary and striking patterns.
Cut 1: vulnerability in the operational pinch points
The first cut filters for decisions tagged as Vulnerability not recognised / supported alongside one of three operational contexts — Collections conduct, Forbearance / support, or Default / termination / repossession — with decision text containing any of the keywords vulnerable, vulnerability, vulnerable customer, collections or arrears.
[FOS Watch search definition — screenshot to be added]
The results compare two consecutive twelve-month periods.
| Period | Decisions matched | Uphold rate |
|---|---|---|
| 1 Apr 2024 – 31 Mar 2025 | 3,276 | 43.3% |
| 1 Apr 2025 – 31 Mar 2026 | 5,559 | 27.0% |
| Year-on-year change | +70% volume | −16.3 percentage points |
[FOS Watch monthly volumes chart — screenshot to be added]
Cut 2: vulnerability alongside bereavement or capacity
The second cut filters for decisions tagged as Vulnerability not recognised / supported alongside Bereavement and estate administration or Capacity / Power of Attorney / Third Party Authority, with decision text containing any of the keywords vulnerability, vulnerable, bereavement, bereaved, capacity or mental capacity.
[FOS Watch search definition — screenshot to be added]
The results, for the same two periods:
| Period | Decisions matched | Uphold rate |
|---|---|---|
| 1 Apr 2024 – 31 Mar 2025 | 2,108 | 42.4% |
| 1 Apr 2025 – 31 Mar 2026 | 3,952 | 26.5% |
| Year-on-year change | +87% volume | −15.9 percentage points |
[FOS Watch monthly volumes chart — screenshot to be added]
The two cuts rhyme. Volumes have grown sharply — by 70% and 87% respectively. Uphold rates have fallen by around 16 percentage points in both. The “Vulnerability not recognised / supported” filter sits across both cuts as the upstream tag: these are decisions where the FOS identified a vulnerability-handling failure as part of the case.
Three readings of the combined pattern
A falling uphold rate, on its own, looks like good news. Read against a sharply rising volume, the picture is more nuanced. Three competing readings deserve attention.
Reading 1 — firms are improving at the margin. Increasing volume reflects rising consumer awareness, broader claims-management activity, and a post-Consumer-Duty willingness to escalate. Falling uphold rates reflect better firm handling at final response, particularly in operational areas where firms have invested in vulnerability training and process redesign over recent years. If this reading dominates, the underlying conduct risk is reducing — even though absolute upheld-case volumes are still rising.
Reading 2 — weaker cases are reaching the FOS. Post-Duty consumer awareness, CMC activity, and the increased prominence of vulnerability in public discourse may be driving more marginal cases into escalation. Firms’ final response decisions hold up at FOS in a higher proportion of cases, but the underlying conduct picture has not materially changed. The headline numbers move because the denominator has shifted.
Reading 3 — the FOS is recalibrating. The same patterns could reflect a tightening of the FOS’s threshold for upholding cases — perhaps in anticipation of the March 2026 reforms that align the “fair and reasonable” test more directly with FCA rules. A more rule-grounded approach may translate into fewer overturns on borderline cases.
Hypothesis (label it): the combined pattern probably reflects a mixture of all three readings rather than any single one. The relative weight is impossible to determine from FOS data alone — it would need triangulation with firm-level complaint data and FOS commentary. But the practical implication is the same regardless of attribution. The absolute number of upheld vulnerability cases is rising in both cuts. A falling rate is not, on its own, evidence of improving conduct.
Where the patterns concentrate
The taxonomy filters used in both cuts point directly to the operational journeys where vulnerability cases concentrate at the FOS.
Cut 1 concentrates in financial-difficulty journeys. Collections conduct, forbearance and support, default and termination. These are the journeys where the customer is already in financial difficulty — which is itself a vulnerability under the FCA’s FG21/1 guidance on the fair treatment of vulnerable customers — and where standard process applied without adaptation is the most common upheld-case reasoning. Three recurring failure modes show up in this reasoning: standard collections process applied to customers who have disclosed vulnerability; forbearance options available on paper but not offered at the right moment; and signposting to debt charities or money guidance that becomes a tick-box rather than a paused-and-handed-off interaction.
Cut 2 concentrates around life events. Bereavement, estate administration, capacity, power of attorney, third party authority. These are cases where speed of response, sensitivity of handling, and propagation of customer information across teams matter most. They typically carry a heavier tail of redress because compounded harm is recognised — a customer faced a serious life event, the firm’s handling made it materially worse, and the redress reflects both the financial harm and the distress.
The “Vulnerability not recognised / supported” filter sits in both cuts because it is the upstream failure. It captures decisions where the firm should have identified that the customer needed a different journey, and didn’t. This pattern repeats across UK financial services — banking, insurance, building societies, and consumer credit — not in any one sector alone.
Implications for risk and compliance teams
Three implications follow for risk and compliance, MI and complaints leadership teams across UK financial services.
Read the volume and the uphold rate together. A falling uphold rate on its own can read as good news. Read against rising volume, it tells a different story. The absolute number of upheld vulnerability cases against UK financial services firms is still rising; the rate is just hiding it. A complaints MI function that reports only the rate will be reporting the comforting half of the picture.
Pressure-test the operational pinch points. Collections, forbearance, bereavement, capacity routing. The taxonomy filters used in both cuts above point directly to the journeys where vulnerability cases concentrate. Process maps, FOS reasoning, and complaints data should be reviewed together — not in isolation — for these journeys. The gaps in upheld cases are operational, not policy.
Track rate of change against peers, not just internal trends. A firm whose vulnerability-related FOS cases are rising in line with the sector pattern is exposed to a market-wide trend. A firm whose volume is rising materially faster than peers has a firm-specific issue. The FCA’s February 2026 Consumer Duty board reporting good practice guidance explicitly favours MI that supports decision-making and shows action taken — which means board-level reporting should show direction and comparison, not just volume.
The Consumer Duty Outcome 4 test on supporting vulnerable customers is no longer abstract. The data is here. The patterns repeat across firms. The question for R&C teams is whether the firm-level picture matches the sector picture — or diverges from it.
A note on the data
Both cuts above are population-level — every published FOS final decision matching the taxonomy filters and keywords for the periods stated. The keyword logic in both cuts is OR-broadened, so a decision matches if any of the listed terms appears in its text. The taxonomy filters are AND-combined with each other and with the date range.
FOS Watch tags decisions; it does not host or republish them. The full FOS public database remains the canonical source for any individual decision. Aggregations are produced by rule-based methods against the taxonomy and are reproducible — every number above can be re-derived from the same filter and keyword combination.
Sources
- Atago Green, FOS Watch — Population cuts for 1 April 2024 – 31 March 2025 and 1 April 2025 – 31 March 2026, accessed 27 May 2026. Filter and keyword combinations as set out in the article.
- Financial Conduct Authority, FG21/1: Guidance for firms on the fair treatment of vulnerable customers, February 2021.
- Financial Conduct Authority, Consumer Duty pages (joint FCA / ICO vulnerability statement, 27 March 2026), accessed 27 May 2026.
- Financial Conduct Authority, Consumer Duty board reports: good practice and areas for improvement, February 2026.
- HM Treasury, Review of the Financial Ombudsman Service — Consultation response, 16 March 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.