The latest figures from the Financial Conduct Authority should serve as a wakeup call for the banking industry. In the first half of 2025, UK financial services firms received 1.85 million complaints and paid out £283m in redress, a 20% increase on the previous six months.

These numbers are not an anomaly. Complaint volumes have remained consistently high for several years, and there is little to suggest that trend will reverse in the near term. What is changing is the pressure this volume is placing on banks’ ability to respond effectively. As volumes hold steady or increase, the gap between what customers expect and what institutions can deliver is becoming more visible.

Complaints as a signal, not the cause

It is easy to attribute rising complaints to external drivers. Fraud is one. UK Finance reports £629.3m lost to fraud in the first half of 2025 across 2.09 million cases. Disputes around reimbursement follow naturally.

There are also ongoing remediation programmes, including motor finance, alongside increased scrutiny under Consumer Duty.

However, complaints themselves are not the root issue. They reflect how effectively banks are operating and communicating with customers. When issues are resolved late or inconsistently, they surface as complaints. In many cases, the complaint is simply the first formal moment a breakdown becomes visible.

Where legacy processes break down

Most complaint handling models still rely heavily on manual work. Cases move across teams, systems, and formats. Key information sits in emails, call recordings, and documents that need to be reviewed individually.

This creates delays before an investigation even begins. It also introduces variation. Different handlers can reach different conclusions based on the same information.

Around 57% of complaints continue to be upheld, according to FCA data. That figure points to a wider issue. Many complaints are not edge cases. They are valid concerns that could have been addressed earlier in the customer journey.

At scale, the challenge becomes more pronounced. Large banks operate across multiple product lines and jurisdictions, often on infrastructure that has evolved over decades. Processes that were once workable at lower volumes are now stretched, making consistency difficult to achieve.

Rising expectations from regulators

At the same time, regulatory expectations have shifted. Consumer Duty requires firms to demonstrate fair outcomes, supported by clear reasoning and consistent treatment of customers.

This is a higher standard than many existing processes were designed to meet. It is no longer enough to resolve complaints within a timeframe. Firms need to show how decisions were made and ensure similar cases are handled in similar ways.

That level of traceability is difficult to deliver using manual workflows alone, particularly when decisions rely on fragmented data and individual judgement.

What changes with AI

The role of AI in complaint handling is often misunderstood. It is not about removing human involvement. It is about creating a consistent foundation for decision-making.

One of the most immediate benefits is the ability to structure unstructured data. Complaints involve large amounts of narrative information. AI can process this quickly, extracting timelines, identifying relevant details, and presenting them in a standardised format.

This reduces the time spent gathering information and allows investigators to focus on the substance of the case. It also shortens resolution timelines, which has a direct impact on customer experience.

Consistency is another area where technology can have a direct impact. By applying policies and regulatory guidance in a systematic way, variation between similar cases can be reduced.

There is also a scale advantage. Systems can process large volumes of complaints at once, which becomes particularly important during spikes driven by fraud or remediation programmes.

From resolution to insight

There is a broader shift that comes with this. Complaint handling can move beyond resolving individual cases to identifying patterns across them.

When data is analysed at scale, recurring issues become visible much earlier. Product design flaws, communication gaps, and operational weaknesses can be identified and addressed.

This creates a feedback loop into the wider organisation. Complaints start to inform change rather than simply reflect it. Over time, this can reduce repeat issues and improve overall customer outcomes.

An operational issue, not a complaints issue

Complaint volumes are unlikely to fall in the near term. Redress costs are rising. Regulatory scrutiny continues to increase.

The underlying issue is not the number of complaints. It is how they are handled.

Until those changes, the figures reported by the FCA will continue to tell the same story and banks will continue to pay the price.

João Pedro Almeida, Co-Founder of Noxus AI