In our latest research, Care by Design, we found that nearly three-quarters of financially vulnerable customers have felt like giving up on getting help because customer service was too confusing to navigate. James Towner, Chief Growth Officer, explains the problem usually isn’t from bad intentions, it’s that services get built around the technology rather than the person who has to use it. So how can organisations fix that?
For someone in financial difficulty, contacting their bank or insurer should be the beginning of getting support. Too often it marks the start of a “doom loop”: navigating complex automated menus, repeating information across multiple channels, being redirected between teams, and enduring long waits – only to abandon the process before receiving the help they need. This isn’t just a tech issue. It also reflects a broader industry focus on cost-cutting and operational efficiency, often at the expense of customer experience, leaving vulnerable customers caught in systems that aren’t designed to support them when it matters most.
The problem isn’t AI itself. In fact, technology has enormous potential to improve outcomes for vulnerable customers. The challenge is that many organisations are deploying AI and redesigning customer journeys without fully considering the needs of those who are most likely to struggle. The result is that technology intended to make services more efficient can inadvertently create new barriers for these people.
The support gap vulnerable customers are facing
Our latest research, which includes a survey of 1,000 senior decision-makers across UK financial services firms as well as qualitative interviews with senior leaders, finds that access to support for vulnerable customers is not keeping pace with rising demand, and the rapid adoption of AI-enabled customer journeys is a significant factor.
The findings reveal a troubling disconnect. While 88% of financial services organisations have increased their use of AI in customer-facing operations over the past year, only 27% test those systems using real vulnerable customer scenarios before deployment. Technology is advancing faster than the safeguards designed to ensure it works for everyone.
Perhaps most strikingly, 77% of organisations are concerned that their own AI strategy could pose a risk to vulnerable customers.
What “good” looks like under Consumer Duty
For many industry leaders, the problem starts with a lack of guidance. The data reveals that 85% of firms want greater clarity from the Financial Conduct Authority (FCA) on how Consumer Duty should specifically apply to AI-enabled customer journeys. The FCA, on the other hand, has reiterated that it will not publish any new regulations for AI, but will instead rely on existing frameworks.
So where does that leave financial services firms who are under pressure from the government to move faster on AI but also conscious of not leaving vulnerable customers behind?
Organisations that start with the customer rather than the technology are already well-positioned. Consumer Duty is outcomes-based regulation, meaning firms should focus less on whether AI is being used and more on whether it delivers fair, effective and accessible support for every customer, including those in vulnerable circumstances.
A firm, or its outsourced contact centre partner, should be able to demonstrate:
- Clear, accessible human-escalation routes that do not require significant effort to navigate.
- Services designed and tested against real vulnerable-customer scenarios, not just edge cases in a lab.
- Outcome monitoring: not just whether the call was handled, but whether the customer’s situation improved.
- Board-level oversight of vulnerability risk and AI deployment, with named accountability.
- Signposting and referral capability, connecting customers in difficulty to the right external support.
- Regular identification of where AI is and isn’t serving vulnerable customers, with mechanisms to adapt.
At ArvatoConnect, this is evidenced through outcome-led service design, where every AI-enabled interaction is continuously monitored against vulnerability indicators and real-world customer outcomes, ensuring compliance is demonstrable rather than assumed.
Specialist support for customers in financial difficulty
Supporting customers in vulnerable circumstances effectively requires specialist knowledge and empathy. Agents need to understand the wider support ecosystem, including organisations such as StepChange, National Debtline and government-backed debt advice services. They need to know when a customer’s situation requires a specialist callback rather than an immediate resolution attempt and avoid language that shames or rushes someone already under financial pressure.
At ArvatoConnect, every agent receives specialist vulnerability training and our Quality Assurance process is focused on outcomes rather than procedure. When we assess the work completed by agents, we’re looking to ensure they’ve understood the customer’s circumstances and needs and delivered the right outcome to them, not whether they’ve followed every process to the letter.
Scaling support when demand strikes, without dropping the ball
Periods of economic uncertainty inevitably increase the number of customers requiring additional support. Cost of living pressures, rising debt and changing personal circumstances all place greater demands on customer service operations at precisely the moment quality becomes most important.
Technology has an important role to play here.
Conversational analytics that identify vulnerability earlier in an interaction. Digital agents that assist with form-filling and reduce friction for customers with low digital literacy. Intelligent triage that gets customers to the right person faster. Targeted, considerate technology deployment means scaling never has to come at the expense of quality.
In practice, organisations that combine early vulnerability detection with intelligent triage consistently reduce repeat contact and abandonment, improving both operational efficiency and customer outcomes. The majority of leaders (88%) believe AI can improve support, and they have already deployed tools to help vulnerable customers.
Building trust into AI-augmented services
The most effective implementations start with the decision-making process and build safeguards into customer journeys from the outset.
ArvatoConnect’s partnership with XpertRule combines ArvatoConnect’s technology-enabled customer experience expertise with XpertRule’s deterministic AI built for high-trust decision environments. Rather than replacing the judgement of trained, expert agents, the technology combines business rules, AI and governance to support more consistent, transparent and explainable decision-making.
For vulnerable customer interactions, this kind of approach can help organisations:
- identify indicators of vulnerability earlier;
- ensure customers are routed to the most appropriate support pathway;
- maintain consistency across complex decisions;
- provide clear audit trails and governance for regulatory purposes; and
- adapt customer journeys as new evidence emerges about what delivers better outcomes.
Importantly, the technology is designed to keep humans in control of critical decisions while giving organisations the confidence to scale support responsibly.
Designing AI around outcomes, not the other way round
Too many organisations are leading with the technology and retrofitting the customer outcome. Consumer Duty, as a regulatory framework, demands the reverse: start with what good looks like for the customer, and build, with or without AI, to deliver it.
AI deployment in financial services must be anchored to existing Consumer Duty expectations. That means clear accountability for AI-driven outcomes, testing against real vulnerable-customer scenarios before deployment, and putting the governance structures in place to monitor and adapt when the evidence shows something isn’t working.
