Three FCA reviews landed in 2026, in March, April and July, and read together they say one thing. Firms have built the governance the Consumer Duty asked for. Very few can yet prove, file by file, that it is producing the outcomes it was built to produce. As firms prepare their third annual board reports, that is the gap a compliance file check either closes or quietly leaves open.
Start with the sequence, and notice what each review actually tests. These are not general governance health checks. Each one assesses firms directly against one or more of the four Consumer Duty outcomes, products and services, price and value, consumer understanding, and consumer support, and marks firms down against that specific outcome, not against “Consumer Duty” as an undifferentiated whole.
On 13 March 2026 the FCA published its review of the consumer understanding outcome specifically, finding that the strongest firms test communications with real customers before and after changes, rather than assuming clarity from a compliance sign-off. On 16 April 2026 it published its observations on firms’ second annual Consumer Duty board reports, assessing how firms evidence all four outcomes together, and flagging a specific and repeated weakness: firms present extensive management information without drawing conclusions from it against any one outcome, document board approval without documenting board challenge, and monitor their own book more closely than the parts of the distribution chain they rely on. On 27 July 2026 it published its review of outcomes monitoring, testing firms against consumer support and price and value in particular: firms need to show that monitoring is being used, that it identifies issues against a named outcome, informs decisions and improves outcomes in practice, not merely that a monitoring framework exists on paper.
None of these three reviews introduced a new rule. Each one tests the same thing a file check has to test: not whether a policy exists, but whether a specific file, against a specific one of the four outcomes, shows the outcome was actually delivered.
The Year 3 effect
Call it the Year 3 effect, because that is roughly what it is. Board reports written in Year 1 were judged, reasonably, on whether the framework existed at all. Year 2 reports were judged on whether the framework had matured, whether data had broadened, whether ownership had sharpened. By Year 3, a report that still reads as a well-organised description of the framework, with data presented but not interpreted, is no longer read as an early-stage report catching up. It is read as a firm that has had two full cycles to close the gap between process and outcome and has not done so. Firms that took the first two cycles seriously are, on the FCA’s own framing, increasingly distinguishable from firms that treated them as documentary exercises, and the distinguishing evidence is not the policy document. It is what the underlying files can show when someone goes and looks.
That is the part of this that belongs in a file review conversation rather than a board pack conversation, and it is where the FCA’s own list of common Year 3 weaknesses reads almost like a file-checking specification, and one that sorts neatly against the four outcomes rather than sitting outside them. Input metrics presented as outcome metrics: sales volumes and satisfaction scores standing in for evidence that a customer actually got a good result under products and services or price and value. Closed product books given a paragraph of reassurance on price and value while live business gets the real analysis. Vulnerable customer reporting under consumer support that describes a policy rather than an outcome. Action plans with no named owner or completion measure, whichever outcome they sit under. A distribution chain dimension, cutting across all four outcomes, that is acknowledged in principle and absent in substance. Every one of those is a pattern that shows up, or fails to show up, against a specific outcome at the level of an individual file, long before it reaches a board pack in aggregate.
What a board report actually needs underneath it
A Consumer Duty board report is, structurally, a roll-up. It is only as good as the file-level evidence feeding it, and the FCA’s 2026 reviews have been explicit that “we reviewed a sample and it looked fine” is no longer the standard the roll-up needs to meet. The July 2026 outcomes monitoring review is direct on this point: a firm needs to show that its monitoring changes something, which means a monitoring exercise needs to be able to point to the specific finding, the specific file, and the specific action that followed, in both directions, not just an aggregate pass rate.
This is the layer Frank is built to sit at, because it is the layer the FCA keeps saying is missing. What it produces, for every file, is far more granular than a single pass or fail: a category-by-category review of the file, client circumstances, existing plans and guarantees, attitude to risk, capacity for loss, suitability of the recommendation, costs and charges, disclosure, vulnerability, each rated independently, alongside a dedicated Consumer Duty section that puts a discrete Yes, No or Unclear against named outcome statements: was a good outcome delivered, was foreseeable harm avoided, are communications clear enough to support an informed decision, does the file evidence that the client actually understood the recommendation, and so on, each with the reasoning behind it drawn from the specific document it came from.
The pay-off is bigger than any one file. Because every review answers the same categories and the same outcome statements in the same structured way, a firm running a meaningful volume of files through Frank has, without any extra work, a genuinely comparable dataset: whether Amber findings on foreseeable harm cluster around one adviser, one product line, or one stage of the advice process; whether a category that was clean six months ago is starting to drift; whether new business and back-book files behave differently against the same tests. A stack of free-text file notes cannot be filtered, sorted or trended that way. A set of files each answered against the same fixed categories and the same outcome statements can, and that is what turns a pile of individual reviews into a picture of the whole book.
That granularity is worth more to a firm than a bare grade, and it is exactly the level of evidence the FCA’s 2026 reviews reward. A file can pass every other test and still fail on a single outcome statement: a suitability report that is otherwise well constructed but silent on a guaranteed annuity rate the client is giving up, say, fails “avoided foreseeable harm” and “communications clear” specifically, while every other outcome on the same file reads Yes. Multiply that across a hundred files and a firm is not looking at one grade repeated a hundred times; it is looking at a hundred structured, comparable answers it can filter and trend by adviser, by product, or by outcome, to see precisely where its book is strong and where it is not.
That last point is where the CP26/23 consultation on scope and proportionality, running to 18 September 2026, connects back to file-level work rather than away from it. Much of that consultation is about narrowing where the Duty applies, particularly for wholesale and cross-border business. None of the narrowing touches retail advice to UK clients, and the FCA’s own commentary alongside it makes clear that reducing scope at the edges is explicitly meant to let supervisory and firm attention concentrate more, not less, on the outcomes retail customers actually receive. A firm reading CP26/23 as a general signal to relax is reading it backwards, and a firm whose file reviews already carry outcome-specific commentary and reasoning is better placed than most to show it has done the opposite.
What this does not mean
None of this means every firm needs to review every file to produce a defensible board report; the Duty has always been proportionate, and a well-designed sample, properly stratified across product, adviser and vulnerability characteristics, remains a legitimate way to test a book. What it does mean is that whatever sample a firm draws on has to be capable of producing the kind of individual-file evidence described above, evidence that either confirms or contradicts whatever a firm’s own aggregate reporting says, and a firm needs to be able to show that reconciliation happened rather than assume the two agree. A board report that states a headline percentage with no visible route back to the files behind it is, on the FCA’s 2026 findings, precisely the kind of report that no longer clears the bar.
None of this means Frank writes the board report itself, and it shouldn’t: deciding what a pattern across many files means for the firm, and what to do about it, is exactly the judgement a board exists to exercise. What Frank changes is how much is actually available for that judgement to work with. Instead of starting from a bare pass rate, a firm starts from category ratings and outcome-by-outcome answers on every file it has reviewed, structured consistently enough to be pulled straight into whatever trend view, adviser breakdown, or period comparison the board actually wants to see, without anyone having to re-read the files to build it.
A note on this piece
This is commentary, not regulatory advice, and it does not constitute compliance sign-off for any firm or file. Descriptions of the FCA’s 2026 Consumer Duty reviews and CP26/23 are our reading of what has been published; firms should refer to the FCA’s own materials and take their own advice before acting. If you would like to talk about how Frank’s granular, per-outcome data across your files could feed the trends and evidence behind this year’s board report, we are happy to have that conversation.