File sampling is not outcomes monitoring — and what is
An FCA reviewer does not ask “are your files good?”. They ask: show me your methodology. Show me this quarter's outcomes. Show me the files you checked and what you found.
“We sample three a quarter and they seemed fine” is an answer to a different question — one nobody asked.
The FCA's requirements review update, held in its Consumer Duty publications library, made this explicit. What counts as monitoring is structured, repeated, evidenced review — proportionate to your size, but real.
The smallest defensible version: every file scored against a fixed rubric, results logged, gaps tracked, a one-page outcome summary for the board.
That is a process you can run in Excel. Most firms do not, because Excel makes it an afternoon every quarter. It does not have to be an afternoon.
What the FCA says about monitoring versus sampling
Under the Consumer Duty, a firm must act to deliver good outcomes for retail customers across the four outcome areas and be able to demonstrate, openly and honestly, that it is doing so. The FCA's Consumer Duty pages set out the four outcomes — products and services, price and value, consumer understanding, and consumer support — and the demonstration is what gets reviewed: not whether a compliance check happened, but whether customers are getting good outcomes, which is a monitoring question.
The requirements review update and the good and poor practice examples in the publications library draw the line between a compliance check and monitoring. A check asks whether individual files were compliant at a point in time. Monitoring asks whether outcomes are being delivered across the book over time — and the poor-practice examples are disproportionately about firms that confused the two.
What outcomes monitoring looks like at file level
The smallest defensible version of monitoring: every file scored against a fixed rubric, results logged with sign-off, gaps tracked with owners, and a one-page outcome summary for the board. Nothing in that list needs more than a spreadsheet — but every element has to be there, every quarter, in the same format.
Sampling answers “were these N files compliant at review time?”. That is a valid compliance check — firms run it to keep themselves honest between full cycles, and it can be documented and risk-based.
Monitoring answers “are customers getting good outcomes across the book over time?”. It needs a rubric, a trend, and a response — the gap list and the board page that show findings turned into change. A sample with no rubric, no trend, and no response produces none of those, whatever its size.
A worked example: two answers to the same request
Before
“We sample three a quarter and they seemed fine.” No rubric version, no scores, no rationale for the three, no trend. The reviewer cannot read results, coverage, or change from this — only that files were looked at.
After
Rubric rev 2, 12 checks. Every new file scored. 40 graded this year: 31 pass, 6 amber on cost disclosure, 3 fail on capacity-for-loss evidence — all worked and re-scored. The sampling rationale (adviser tenure and product line) was documented and chosen before the review. The trend line and the board page read as one answer.
The second answer is not more opinion; it is a different question being answered — across the book, over time, with a response attached.
Common mistakes when sampling stands in for monitoring
- Assuming a clean sample proves the book. Three passing files tell a reviewer nothing about the other files — or about whether outcomes are improving.
- Changing the rubric between rounds. A rubric that shifts every quarter makes the trend meaningless. Fix the version; score against it; change it deliberately, with a documented migration.
- Equating files reviewed with coverage of the client bank. Twelve reviews a year across a 150-client book is a check; describing it as cover of the book is the claim a reviewer will probe.
- Choosing the sample after the fact. A sample picked once the results are known is a selection. Document the rationale before the review — by adviser tenure, product line, or risk area — and keep it.
Sampling vs outcomes monitoring
| Question | Sampling | Outcomes monitoring |
|---|---|---|
| Question answered | Were these N files compliant at review time? | Are customers getting good outcomes across the book over time? |
| Evidence produced | A list of files checked, often with no scoring rationale. | A graded log against a fixed rubric, trended and board-reported. |
| What a reviewer can read from it | That a check happened — not what it found or what changed. | Results, trends, and the response to them. |
| Role in the evidence chain | A supporting, documented, risk-based check. | The core demonstration of good outcomes. |
Related reads
Sampling belongs inside a monitoring process, not instead of one. See the Consumer Duty outcomes monitoring hub for the process in full, outcomes monitoring in practice for what the scored cycle looks like end to end, and the MI dashboard for the trend view that makes the difference visible.
Frequently asked questions
Does the FCA ban file sampling?
No. Sampling can feed into monitoring as a documented, risk-based check — for example to test a specific area between full reviews. But a sample that finds no issues answers a different question from one that tracks whether customers are getting good outcomes over time. The FCA's publications library keeps that distinction in scope.
How many files should a firm review per quarter?
The cited FCA sources do not set a fixed number, and any fixed number would be arbitrary. The defensible version scores every new file against a fixed rubric and documents the rationale whenever a sample is selected, so the count is whatever the book produced — reviewable and repeatable.
Can consistent sampling count as monitoring?
A consistent, rubric-scored, trended sample that is board-reported is part of a monitoring process. An ad hoc 'no issues found' sample is not — it has no trend, no rationale, and no response, so a reviewer cannot read outcomes from it.
Your next step
Convert the sample into a scored cycle: fix the rubric and its version, score each new file, log the reasons, and page the board with the trend. Proven Duty scores every review against a fixed, versioned rubric and builds the graded log and board summary — so the sample becomes a monitored book without an afternoon a quarter. Start a free trial or see pricing.