A firm can count complaints, monitor abandonment, circulate a board pack and still know remarkably little about whether its customers are receiving good outcomes.
That is the uncomfortable finding behind the FCA’s latest review of Consumer Duty outcomes monitoring. Published on 27 July 2026, it draws on board reports, information requests and a survey of 56 firms across different sectors, sizes and business models. Many firms had plenty of data but could not explain what it meant, which risk it revealed, what they did in response or whether the intervention worked.
The publication does not create a new Consumer Duty requirement. It shows how the FCA is assessing firms against obligations already contained in the Handbook. PRIN 2A.9 requires firms to monitor the outcomes customers receive from products, communications and customer support, identify root causes when those outcomes fall short and take appropriate action. The consumer understanding rules also require firms, where appropriate, to test communications before use and monitor their impact afterwards.
Taken together, the review describes a chain from customer signal to interpretation, decision, action and a measured result. Most firms can produce evidence from somewhere along that chain; the regulatory test is whether they can connect it.
A dashboard is not evidence of a good outcome
Outcome monitoring has often been approached as a data problem: identify the available measures, establish tolerances and report the results to a committee.
In the examples the FCA regarded as effective, firms first defined what a good outcome should look like at particular stages of a customer journey, then selected information capable of showing whether customers were receiving it. One firm set different expectations for customers joining, using and leaving each of its principal products instead of relying on a broad statement that covered everything. Others linked known risks of harm to indicators such as complaints, attrition and compliance-monitoring results. Without that definition of the intended outcome, a metric is merely a number that moves.
The review found firms using operational measures such as conversion rates and review completion as proxies for customer outcomes without defining what good or poor outcomes looked like at the relevant stage. Some had thresholds for complaints, file-review pass rates, client retention and mortgage-review engagement but could not consistently explain why those levels were meaningful.
Conversion, retention and processing time may all reveal something useful, but none carries an inherent Consumer Duty meaning. A high conversion rate could reflect a clear, efficient journey or one that makes an important limitation too easy to overlook, while low contact volumes could indicate either customer understanding or difficulty reaching support.
A firm therefore needs an account of why each measure was chosen, which risk it can expose and what should happen when it moves outside tolerance. Additional management information cannot compensate for a missing definition and may even conceal it.
One FCA example involved a firm monitoring an onboarding redesign, the clarity of payment instructions and high-cash or orphaned clients. Although the subjects were relevant, the firm could not consistently show which information it had considered, where it had been discussed or challenged, what decisions followed, when action was taken or how the impact was assessed. Its own reporting acknowledged the need for better records connecting trends, actions and customer outcomes, leaving a collection of monitoring activities rather than a coherent decision record.
Other examples make the distinction clear: customer feedback and operational information led one firm to introduce in-app chat, then use AI-based routing and keyword recognition to identify possible signs of vulnerability. Over six months, its average first-response time fell from 22 hours to under two minutes and average resolution time fell from four days to under three hours.
Another firm traced withdrawal delays to the process for submitting identity documents. After piloting a different verification approach, it reported a 5 per cent improvement in anti-money-laundering pass rates and a 20 per cent improvement in bank-verification pass rates.
Neither example means every improvement can be attributed neatly to a single intervention because customer journeys rarely offer laboratory conditions. They do show what the FCA wants firms to evidence: a defined problem, a reasoned response and information capable of testing whether the position improved.
The review shows how frequently that final step is lost. One firm identified unclear live-chat interactions, verification delays, repeat contact and inconsistent first-contact resolution, then introduced clearer expectations, clearer ownership, improved escalation and staff training. Later evidence showed that customers were still being passed between agents and complex issues were not always resolved at the first attempt.
The actions had been completed, but the problem remained. A new checklist, training programme or workflow may be sensible; its implementation proves only that the firm changed a process, whereas outcomes monitoring asks what changed for the customer.
Marketing data belongs in Consumer Duty outcomes monitoring
One of the review’s most useful examples begins with acquisition data. A firm examined rejected applicants to determine whether its distribution channels were reaching the intended target market and found that some were producing high volumes of unsuitable applicants without sufficient income or savings. It ended two paid affiliate relationships.
The rejection rate was a signal, not a verdict. It prompted the firm to investigate the source of the applications, consider the risk and change its distribution activity.
That sequence should interest marketing leaders because a commercially efficient campaign can still produce poor outcomes. Cheap applications are of limited value when they repeatedly come from people outside the target market, while strong engagement may reflect persuasive creative that masks confusion about eligibility, cost or product limitations.
The Consumer Duty rules place communications inside the monitoring obligation. Firms must regularly monitor outcomes arising from their customer communications and, where appropriate, test material before publication and monitor its impact afterwards. If problems emerge, the rules require the firm to investigate and correct deficiencies in the communication or, where wording alone is insufficient, the underlying product or process.
This creates a shared agenda for marketing, product, customer support and compliance. Campaign data shows who entered the journey; application and abandonment data shows where they left; customer testing, repeat queries, complaints and service interactions indicate what they failed to understand; and segmentation reveals whether particular groups, including customers with characteristics of vulnerability, experienced a different result.
Kept in separate systems, each source offers a partial view. Connected around the customer journey, they begin to explain the outcome.
Communications approval is commonly treated as an endpoint: the final version is stored, the reviewer’s name is recorded and the asset goes live. Consumer Duty outcomes monitoring makes it closer to a forecast that can be tested against what customers subsequently do and understand.
A reviewer is making a judgement about how the intended audience is likely to interpret the communication, whether the important claims are properly supported and whether the presentation helps customers make an informed decision. Post-publication evidence can confirm that judgement or expose something the review process missed.
The FCA does not prescribe a new approval template in this publication. A practical response would be to preserve the reasoning behind material decisions: the intended audience, the principal claims and evidence, known risks of misunderstanding, conditions attached to approval and the signals that would justify a second look.
That record gives later monitoring something to test. Did customers understand the eligibility condition that attracted the most debate during review? Did one distribution channel produce a different application profile? Did support contacts fall after a difficult passage was rewritten? Were the results consistent for customers with different needs?
Without the original reasoning, outcome data arrives detached from the decision it ought to challenge. The firm may recognise that something went wrong but struggle to establish which assumption failed.
Preserving the judgement also improves future work. A reviewer who encounters a similar claim can see how it was assessed before, which evidence mattered and what customers subsequently did. Previous decisions should inform rather than dictate the next one, but forcing every team to reconstruct the same judgement from first principles wastes expertise and invites inconsistency.
Board governance must show how decisions changed
The FCA found that many firms could describe committees, reporting lines and escalation routes, although it was often less clear how those arrangements operated in practice. In some examples, boards received regular customer-outcome reports but appeared to review and approve them without challenging the analysis or pressing for further action. A completed report shows that information reached the board, not what the board did with it.
Firms that demonstrated active governance maintained central action records covering matters raised through risk events, product governance, committee discussions, management information and file reviews. Actions carried named owners, target dates and status updates, with oversight from senior management, the Consumer Duty Champion and the board. Other firms used triggers and thresholds to determine when an issue required escalation.
Board challenge should leave a mark: a revised threshold, a request for deeper analysis, a named decision, an owner or a change in business activity. The trail should continue until the firm has evidence about the result.
This applies beyond the firm’s own operations. The FCA’s review includes positive examples of firms using information from brokers, distributors, outsourced providers and other partners to identify outcome problems within their area of responsibility. In some cases, firms asked partners to change communications that could confuse or mislead customers.
The FCA also makes clear that smaller firms can take a proportionate approach, using a focused set of meaningful indicators and straightforward records instead of complex systems or large teams. A simple log of the issue, responsible person, agreed action, deadline and follow-up may be sufficient where it matches the business model and risk.
Effective governance moves evidence through the relevant forums and changes what the organisation does; the number of committees involved is beside the point.
AI can challenge communications, not certify them
The FCA’s review includes several examples of firms using AI and automation, but none supports handing regulatory judgement to a model.
One firm piloted synthetic testing to identify communications that might confuse or mislead different customer groups. The tool considered themes including vulnerability, fair treatment and the post-advertising customer experience, producing a comprehension score, engagement heatmap, risk matrix and recommendations. The firm applied it to a handful of communications that had already undergone human testing and reported broad alignment between the two sets of findings.
The regulator frames this as an additional source of challenge. It suggests AI may help firms identify potential comprehension risks, particularly where large customer panels or specialist research teams are unavailable, while emphasising that it should support human judgement with appropriate guardrails. Within that boundary, AI can compare a communication with known risk patterns, retrieve relevant requirements, surface similar decisions and make review records more consistent. It may identify an issue that deserves human attention or help specialists see connections across volumes of material that no individual could hold in memory.
AI cannot decide what level of residual risk the firm should accept, and a comprehension score cannot certify that real customers understood a communication in context. Its role in a regulated review workflow is to widen the expert’s field of view and preserve the basis of the eventual judgement, while responsibility remains with people able to interpret the evidence, challenge its limitations and act on the result.
An organisation can resolve a customer problem and still fail to learn from it: the complaint is closed, the wording changed and the committee action marked green, only for another team six months later to make a similar claim, use the same channel or introduce the same point of friction into a different journey.
The FCA’s findings on root-cause analysis, action tracking and follow-up point towards a model in which outcome evidence flows back into the organisation’s working knowledge: campaign briefs, review standards, customer-testing plans, product design, partner oversight, staff guidance and future approval decisions. The Handbook already requires firms to identify root causes when customers do not receive the required outcomes, and using the resulting evidence in future work turns individual regulatory decisions into institutional knowledge.
Marketing then gains a clearer understanding of which audiences, claims and channels create difficulty; compliance can apply previous judgement with context instead of relying on memory; product and service teams can see where communications expose a deeper flaw in the journey; and boards receive evidence of how the organisation is changing, not simply how much monitoring it performs.
The audit trail has to reach further than approval. It should show what the firm expected, what customers experienced, how specialists interpreted the difference and whether the response worked.
Taken together, this provides evidence for a regulator and a record of how the organisation is improving its decisions.
What teams need to know
What did the FCA say about Consumer Duty outcomes monitoring?
The FCA said that collecting data, listing metrics and reporting management information do not, by themselves, demonstrate good customer outcomes. Firms should be able to explain what the information shows, how it identifies risks or issues, what action follows and how they assess whether the action improved outcomes.
Is the July 2026 review a new Consumer Duty rule?
No. It is a good-and-poor-practice publication illustrating how the FCA has seen firms apply existing Consumer Duty obligations. The main requirements on monitoring, root-cause analysis and action sit in PRIN 2A.9, while PRIN 2A.5 contains specific rules on testing and monitoring customer communications.
What is a Consumer Duty outcomes-monitoring evidence chain?
It is a traceable record connecting a customer signal to the firm’s interpretation, decision, action and assessment of the result. The FCA does not use this as a prescribed template, but its review repeatedly distinguishes strong approaches with clear audit trails from weaker ones that cannot show how information led to improvement.
What should marketing teams monitor under the Consumer Duty?
Relevant evidence may include audience and channel data, rejected applications, journey abandonment, customer testing, repeat queries, complaints and support interactions. The appropriate measures depend on the product, target market, communication, distribution model and potential harm. Firms should be able to explain how each measure relates to an expected customer outcome.
Do smaller firms need complex Consumer Duty dashboards?
No. The FCA says smaller firms can take a proportionate approach using a limited set of meaningful indicators and straightforward records. What matters is whether the firm can identify a possible poor outcome, assign responsibility, act and check whether the position improved.
Can AI support Consumer Duty outcomes monitoring?
Yes, as a supporting tool. The FCA gives examples involving synthetic communications testing, customer-support routing and improved customer records. It presents AI as a supplement to human review and judgement, with appropriate controls, rather than proof of compliance or a good customer outcome.