How these figures are produced
Every number on this site traces to a published document. This page explains exactly where each one comes from, and — as importantly — what it does not mean.
Where the decisions come from
The Financial Ombudsman Service publishes every final decision an ombudsman issues. We read the published record for the “Banking, credit and mortgages” sector and store the full text of each decision.
Which decisions are counted
The Ombudsman decides complaints across six sectors. Every headline figure on this site — uphold rates, issue rates, the scorecard, the odds tool — covers banking, credit and mortgages only.
That is deliberate. Firms sit very differently across the sectors (some large banks are over 40% non-banking by decision count) and the sectors behave differently: banking upholds at about 30%, PPI 28%, insurance 39%, investments 44%. Averaging them would produce a headline whose meaning changed with a firm’s product mix. The other sectors are kept and shown as a separate, labelled breakdown on each firm’s page.
Where the outcome comes from
The Ombudsman’s own search exposes the outcome as a filter, so a decision’s outcome is recorded from the Ombudsman’s classification, not inferred from its prose. We also parse the operative sentence independently as a cross-check; where the two disagree, the decision is flagged rather than silently resolved.
Partial upholds are the one case the Ombudsman’s binary filter cannot express, so those are identified from the decision text and counted separately.
What an uphold rate here is — and is not
It is not the same as a firm’s complaint uphold rate. Published final decisions are cases that went all the way to an ombudsman. The overwhelming majority of complaints are resolved long before that, by the firm or by an investigator. A firm’s rate here therefore describes its contested cases, not its complaints overall.
Where a firm reports figures to the FCA or the Ombudsman under its statutory returns, we show those alongside, clearly labelled, so the two measures can be compared rather than confused.
Thresholds
No uphold rate is published for a firm with fewer than 30 decisions carrying a recorded outcome, and none for an issue with fewer than 20. A percentage over a handful of cases is noise, and presenting it as a finding would be misleading.
Attribution
A decision counts toward a firm only where the text names that firm unambiguously. Where a name could refer to more than one company, the decision is left unattributed rather than guessed. Firm counts are therefore a floor, not a total — we would rather undercount than publish a wrong statement about a named company.
Quotes
Quoted wording is shown only where the exact text was verified to appear in the decision it is attributed to. Anything that fails that check is not shown.
What the provision layer is — and is not
Some decisions name the specific clause of the FCA Handbook they rely on, such as DISP 3.6.4. We record which clause each decision cites by pattern-matching the decision text we already hold. It is not taken from any FCA index, and a clause the pattern does not recognise is not recorded.
Most decisions do not cite a clause by number, and how many do depends on the rulebook. Measured on 30 September 2026: of the banking decisions citing DISP, 4,481 of 6,224 (72.0%) name a specific provision; for CONC it is 3,162 of 6,640 (47.6%); for BCOBS, 26 of 192 (13.5%); for the Consumer Duty, 147 of 1,881 (7.8%). Across the whole banking corpus only 6,907 of 190,616 decisions (3.62%) cite any provision by number. We state coverage per instrument and never as one site-wide figure, because a single figure would describe none of them.
A clause is listed only where at least 20 distinct banking decisions cite it, the same threshold used for issues.
Handbook wording is shown only where the cited reference matches a current Handbook provision exactly. The FCA publishes only current wording, while decisions cite the numbering in force when they were written, and chapters are sometimes replaced whole. For example, CONC 5 today is 78 cards, numbered CONC 5.2A.x, CONC 5.4.x and CONC 5.5A.x, while CONC 5.3.1 (647 decisions), CONC 5.2.2 (273) and CONC 5.2.1 (183) are not in it at all. Those are marked as no longer in the current Handbook and shown without text. The wording in force when the decision was written cannot be obtained, because the Handbook’s dated links redirect to the current chapter, so we do not try to reconstruct it.
A citation that names only a chapter or section, such as DISP 2.3 (128 decisions), is marked as section-level, because no single provision exists to quote.
Thirteen references clear the 20-decision threshold but have no page to appear on. Seven are provisions whose text we hold — PRIN 2.1.1 (162 decisions), SYSC 3.2.6 (123), PRIN 1.1.1 (80) and four MCOB provisions; four name only a chapter or section (MCOB 13, cited in 81, along with MCOB 11.6, MCOB 11.7 and PRIN 2.1); and two belong to a part of the mortgage rulebook since replaced. Nine concern mortgages, three the Principles and one the senior-management rules. That is a known gap: the site tracks twelve named instruments, and the Principles, the senior-management rules and the mortgage rulebook are not among them.
Handbook text is reproduced under the Open Government Licence v3.0, © FCA. The FCA’s own copy is the definitive one.
What this site is not
It reports what was decided. It does not predict what will be decided in any other case, and nothing here is legal advice. If you are complaining about a financial firm, the Financial Ombudsman Service is free to use and you do not need anyone’s help to approach it.
Corrections
If a figure here is wrong, we want to fix it. Every published decision has a page here carrying its reference and a link to the Ombudsman’s own document, so any number can be checked against the source. Decisions we could not attribute to a named firm with certainty are counted but not published, so a firm’s total is a floor.