How banks can evidence customer outcomes under FCA Consumer Duty

A conversation with Sean Kenny, Client Director.

Regulation in banking has shifted. It is no longer enough to show that policies and controls exist. Firms are now expected to evidence the outcomes those controls deliver for customers. 

We spoke to Sean Kenny at Dufrain about where banks are struggling, what is driving those challenges, and how organisations can move towards a more proactive, outcome-led approach. 

(The Financial Conduct Authority Consumer Duty requires firms to demonstrate “good outcomes” across products, price, understanding and support – FCA, 2023) Consumer Duty implementation: good practice and areas for improvement | FCA 

Question: Where are banks still struggling to evidence FCA expectations on customer outcomes?

Answer:

In my experience, the biggest challenge is the gap between policy and execution. 

Most banks can show strong frameworks. They have governance, controls and documentation in place. But applying those consistently across live customer journeys is where things start to break down.  

That becomes a real issue when outcomes need to be explained or defended. 

Ownership is often unclear.
Decisions sit across risk, data, product and operations. When something is challenged, there is no single view or accountable owner, just different interpretations from different teams. 

Legacy systems make things harder.
Customer journeys and calculations are spread across multiple platforms built over time. That makes it difficult to show, end-to-end, how a specific outcome was reached. 

Scale introduces inconsistency.
Processes that worked for smaller volumes, often manual or spreadsheet-driven, start to break down. Variations creep in, and outcomes are no longer consistent across customers. 

And in most organisations, there is still a heavy reliance on SMEs.
That creates key person dependency, which is difficult to evidence and even harder to scale. 

The result is not just inefficiency.
It is a real challenge when it comes to proving that outcomes are fair, consistent and defensible under scrutiny. 

Question: What data, reporting or governance gaps make this harder across the customer journey?

Answer:

A lot of the challenge sits in the data layer, but it is not just data quality in isolation. It is the impact of it. 

One of the biggest gaps is traceability. Being able to move from an outcome back to the underlying data, logic and decisions is still difficult in many organisations.  

When that traceability is missing, it becomes much harder to stand behind decisions with confidence. 

There are also inconsistent definitions and metrics. Different teams measure the same thing in different ways, which leads to conflicting answers. That creates confusion internally and weakens the ability to evidence a clear, unified position externally. 

Another issue is that reporting is still largely process-driven. Banks track volumes and SLAs, but not always: 

• whether outcomes are fair across customer groups  
• whether certain segments are experiencing worse outcomes  
• whether there are unintended consequences  

That means organisations can show activity, but not necessarily whether that activity is delivering the right result. 

Data quality is also often not linked to decision impact. Without clear metrics and thresholds, it is difficult to judge whether outputs are reliable enough to support regulatory reporting. 

And finally, reporting is siloed. Risk, compliance, operations and data teams all have their own views, with no single version of the truth.

The overall effect is a view that is: 

1. fragmented  
2. backward-looking  
3. difficult to defend  

Which is exactly where regulatory pressure is increasing. 

Question: How can banks move from reactive FCA compliance to a proactive, outcome-led approach?

Answer:

The shift is about moving from explaining what happened to understanding what is happening and what might happen next. 

The starting point is still and always will be data foundations: 

• consistent definitions  
• governed models  
• clear lineage  

Without that, it is difficult to scale anything else. 

Next is reducing key person dependency. Logic and decisioning need to move out of spreadsheets and into controlled, documented systems that can be audited and reused. 

Clear ownership of outcomes is also critical. Not just ownership of processes, but accountability for what those processes deliver for customers. 

From there, organisations can start to introduce outcome-based monitoring: 

• tracking real customer experience  
• identifying variation across segments  
• spotting unexpected patterns early  

This is what allows firms to move from, reactive reporting to proactive identification of potential customer harm  

“Organisations must be able to demonstrate compliance not just at a single point in time, for a single obligation, but continuously, as AI systems and the regulations governing them operate and evolve.”

Embedding governance directly into processes is also key: 

• versioning  
• auditability  
• automated controls  

So that evidence is created as part of the process, not reconstructed afterwards. 

Question: How can banks move from reactive FCA compliance to a proactive, outcome-led approach?

Answer: 

AI can play a role, but it reinforces the same point rather than replacing it. 

One of the clearest themes from our banking AI Beyond Blind Trust, Responsible AI discussion was that organisations do not struggle to build AI use cases. They struggle to take them into production safely and consistently. 

That is where governance becomes critical. 

A useful way to frame it is: 

1. Can we build it?  
2. Should we build it?  
3. Is it working as intended?  
4. Does it continue to deliver over time?  

That second question “should we?” is often the one that comes too late.

As Isobel Daley, Head of AI at Dufrain, highlighted in our AI Beyond Blind Trust webinar:

“Lots of things are technically possible… people spend a lot of time developing something and then get to a point where they actually want to do something with it. And then suddenly someone asks the question, ‘should we actually do this? Is this the right thing to do?’ And actually that question should be asked at the beginning and throughout, continually questioning, should we?”

In regulated environments, this matters even more. AI needs to be: 

• explainable  
• monitored
• governed  

And most importantly: The model is never accountable. People are.

“The model is never accountable. Humans are always accountable, and the buck will always stop with us.”

That means AI should sit within existing: 

• risk frameworks  
• governance structures  
• accountability models  

Not outside them. 

Used properly, AI can support: 

• earlier identification of anomalies or bias  
• more proactive monitoring of outcomes  
• better decision-making at scale  

But without the right foundations, it introduces more risk, not less. 

What does “outcome-led” actually mean in practice? 

At its core, it starts with the customer. 

What are they experiencing? Is it fair, consistent and easy to navigate? 

From there, organisations work backwards: 

1. what decisions drive that outcome?  
2. what data supports it?  
3. how is it evidenced?  

And importantly, it is continuous. 

It requires: 

1. monitoring  
2. feedback loops  
3. ongoing adjustment  

So that outcomes improve over time, not just get reported after the fact. 

Q: Final thoughts from you Sean?

Answer:

The shift to outcome-based regulation is a fundamental change. 

Banks that rely on policies and retrospective reporting will find it increasingly difficult to evidence compliance. 

Those that invest in: 

• strong data foundations  
• clear governance  
• outcome-led thinking  

will be in a much stronger position, not just to meet regulatory expectations, but to respond to them with confidence.