Consumer Duty has been in force since July 2023. The compliance question shifted on that day from "did you follow the process" to "can you evidence the outcome was good for the customer". For firms whose primary record of customer interaction is a call recording, that has profound implications. This piece is a checklist of the seven things the FCA expects to be evidenced on those recordings, plus the question they will ask if you cannot.
What changed in 2023
Before Consumer Duty, the regulator's interest in a recorded advice call was largely procedural. Did the adviser introduce themselves and confirm authorisation? Did they capture ATR and capacity for loss? Were charges disclosed at the point of recommendation? These are still required (COBS 9.2 has not gone away), but they are the floor.
Consumer Duty added an outcomes layer on top. The four outcomes (products and services, price and value, consumer understanding, consumer support) are not boxes to tick. They are tests the regulator will apply when reviewing your records. The recordings now have to evidence not just that you did the procedural thing, but that what you did was suitable, fair and understood.
That is a higher bar, and it falls on each call recording rather than on the file as a whole. Below is the checklist, with the procedural element underneath each outcome that customers must be evidenced for.
1. Target market fit
Outcome: products and services. What the recording must evidence: the customer falls within the target market the firm has defined for the recommended product, and the adviser made that link explicitly.
The regulator's question: "How did the adviser confirm this customer matched the target market?" If the recording shows the adviser walking through risk profile, time horizon, capital protection requirements and product features against the firm's documented target market for that product, you have evidence. If the recording shows the adviser jumping from "you said you wanted growth" to "I'm recommending product X", you have a procedural compliance file with no Consumer Duty evidence.
This is the most common gap we see in scored recordings. Advisers are confident the recommendation is right but do not articulate why on the call.
2. Vulnerability awareness
Outcome: consumer support. What the recording must evidence: the adviser screened for vulnerability indicators (per FG21/1's four drivers: health, life event, resilience, capability) and adapted the call where indicators were disclosed.
The regulator's question: "Did the adviser identify and accommodate any vulnerability characteristic the customer disclosed?" If the customer mentioned recent bereavement, redundancy, ill-health or financial stress and the adviser carried on at the same pace using the same language, that is a Consumer Duty fail even if the recommendation was technically suitable.
This is hard to score by sample because vulnerability indicators are rare per call but cumulative across a portfolio. Systematic scoring on 100% of calls catches the patterns; sampled scoring does not. The same evidencing standard applies to CONC-regulated debt conversations, where affordability and forbearance discussions carry the same burden of proof — see our debt collections compliance use case for how that plays out.
3. Fair value, justified out loud
Outcome: price and value. What the recording must evidence: charges (initial and ongoing) were disclosed clearly, and the value the customer receives in exchange was articulated against those charges.
The regulator's question: "If the customer asked at the end of this call why your fee is £X and a competitor's is £Y, did the adviser answer that question on the recording?" Disclosure of charges is the floor. Justification of charges against the ongoing service the customer receives is the Consumer Duty bar.
Common gap: charges disclosed quickly at the end of the call, fair-value articulation skipped because the adviser assumes the customer already understands. Customers do not.
4. Consumer understanding, verified rather than assumed
Outcome: consumer understanding. What the recording must evidence: the customer demonstrated understanding of the recommendation in their own words. Not the adviser saying "does that all make sense?" and the customer saying "yes". The customer paraphrasing the recommendation back.
The regulator's question: "How do you know the customer understood, rather than felt obliged to nod along?" If your recording has the adviser explaining at length followed by a "yes" from the customer, that is not evidence of understanding. If the customer rephrases the recommendation in their own words, however imperfectly, that is.
This is the outcome most firms underweight. Understanding is asymmetric: advisers are domain experts, customers are not. Consumer Duty expects firms to bridge that asymmetry actively.
5. Charges disclosed clearly, in plain English
Outcome: consumer understanding plus price and value. What the recording must evidence: the charges were stated as numbers the customer can compare against alternatives (percentages alone are not enough; the pound-and-pence equivalent over the customer's term is the standard).
The regulator's question: "Did the customer leave the call knowing what they would actually pay?" "0.85% per annum on £200,000" is procedurally compliant but cognitively opaque to most customers. "£1,700 a year, or £8,500 over a five-year horizon" is what Consumer Duty expects.
Easily fixable, but only if you can identify the calls where it is happening. Sample reviews catch a small fraction.
6. Outcome confirmation
Outcome: consumer support. What the recording must evidence: the adviser confirmed at the end of the call what the customer agreed to, what would happen next, and gave the customer the chance to question or pause.
The regulator's question: "Did the customer have a clear picture of the next steps before the call ended, and were they given space to back out?" A confident close where the adviser races through paperwork and the customer signs off without articulating their own understanding fails this test.
Best-practice closing scripts cover this. Live mid-call coaching can prompt advisers when the engineered close is missing key elements. Our financial services use-case page goes deeper on what gets scored against this outcome.
7. Cancellation rights and complaint route, in plain language
Outcome: consumer support. What the recording must evidence: cancellation and cooling-off rights stated in plain English, complaint route mentioned, and contact information given for follow-up questions.
The regulator's question: "If the customer wants to cancel or complain after this call, did your adviser equip them to do so?" The procedural minimum (mentioning the 14-day cooling-off period) is not enough. Consumer Duty expects the adviser to have made the cancellation route easy to use, not just legally available.
This is one of the easiest items to score consistently because the language is fairly standardised, which makes it a useful first scorecard item to build out when implementing AI scoring in a financial planning firm.
Why a 5% sample no longer cuts it
The seven items above are testable on every call. The question is whether you are testing every call or 5% of calls. The FCA does not specifically mandate 100% review, but the supervisory direction since 2023 has consistently been "systematic monitoring" rather than "statistical sampling". The line manager evidence form on a sampled call is a much harder document to defend than a per-call breach register sitting underneath an automated 100% scoring engine.
This is the leverage that AI scoring brings to compliance work in financial services. We covered the operational case for 100% coverage in a separate post; the regulatory case is the same one expressed in compliance language.
What to do next week
Three concrete steps that work in the order listed.
Audit your scorecard against the seven items above. Pull your existing call-checking scorecard and lay it next to this list. Where are the gaps? Most firms find target market fit (#1), consumer understanding verification (#4), and outcome confirmation (#6) are weaker than the procedural items. Start by adding scorecard criteria for those three.
Score 100 recent calls against the updated scorecard. Do this manually if you have to, but ideally on an automated scoring engine so you can do it quickly. The goal is not to find disasters; the goal is to find the systematic patterns. The same Consumer Duty failures show up across many advisers, which means they are scorecard or training issues rather than individual issues.
Build a remediation plan tied to those patterns. Update training scripts. Add coaching memory so advisers who flagged this month are coached on it next month. Track the trend. The supervisory question stops being "are you compliant on this call" and starts being "are you systematically improving against the four outcomes", which is the question Consumer Duty actually asks.
If you want help with #2
If "score 100 recent calls" is the part where this gets stuck because you do not have the QA capacity, that is exactly the gap AI call scoring is built to close. We run a 15-minute demo against five of your own recordings. You see your scorecard, your breaches, and your coaching drafts on the call. Email hello@callguardai.co.uk and we will set it up.