OPINION How machine learning could offer inclusive digital experiences to vulnerable customers

Ade Webb, Chairman at LAB Group, shares the findings of new studies on techniques that can identify vulnerable customers and offer them a better experience.

Digital transformation has a dark side. Vulnerable customers are at risk of being collateral damage as firms drive to digitise products that used to be sold and serviced face-to-face, where people can pick up other people’s potential vulnerabilities quickly and intuitively.

The Financial Conduct Authority’s (FCA) quantification of the more than 10 million vulnerable customers in the UK (pre-Covid) includes people whose needs range from transient vulnerability or low financial competence to psychiatric conditions. It covers people who suddenly find themselves as carers, those with poor numeracy and low reading ages through to people with bipolar disorder or suicidality.

Without appropriate digital journeys to guide, signpost and respond appropriately, those at risk of harm through their circumstances or health are at risk of exclusion, confusion, poor service or – worst of all – ending up with products inappropriate to their needs.

Digital services are super-efficient because they remove human interaction. People are expensive, so insourcing jobs to web pages yields huge financial gains. Unfortunately, in the process, such services fall prey to performance marketing and inevitably become optimised funnels that only work well for un-exceptional customers.

Using psychology to offer a more human experience

In 2018, LAB Group in London started to look for digital techniques being used in practice to help vulnerable customers within the UK financial services market. It was a disappointing search. Monzo has made a credible effort with its introduction of positive friction for gambling, but most providers exhibit meagre attempts to identify and differentiate journeys appropriately when humans are not involved.
Good efforts abound where people are involved, and this is where the majority of the PR output from companies has focused. Initiatives such as the BRUCE protocol from the Money Advice Trust show how appropriate care is being taken very seriously during transactions that involve a human mediator.

When no such mediator is involved, the problems arise and become even worse when full scale optimisation is underway because the process inevitably removes questions that might point towards the presence of vulnerability.

Now, cutting-edge psychological modelling is offering ways to prevent these vulnerable customers being left behind by optimisation. One of the most notable developments is a digital environment that can absorb some of the cues that we naturally process in human-to-human transactions. In effect, these are digital ways to spot discomfort, anxiety, confusion and impulsivity through the subtle medium of kinetic interactions.

In January 2019, LAB Group proposed that a psychological model used previously as an ecommerce optimisation tool could be turned on its head to better identify and serve vulnerable customers. The richest results came by bridging the ‘observation gap’: those tics and biases of human behaviour that a human can assess face-to-face, but a digital journey cannot.

A huge academic study involving qualitative and quantitative tests run through machine learning kicked off, funded by a UKRI grant. LAB set a challenge to its psychologists and neuroscientists to make the browser ‘see’ the customer. If we can reliably detect vulnerability based on kinetic patterns using machine learning, then we can also deliver hyper-personalised digital interventions (e.g. subtle UX changes), in real time, to help protect different types of vulnerable customers going through financial journeys online.

Early results show huge promise, but digital vulnerability requires not only scientific rigour but a willingness to sacrifice small degrees of optimisation efficiency in order to prioritise the needs of vulnerable customers.

Being the change

A variation of the lightbulb class of jokes goes: How many psychologists does it take to change a light bulb? The answer: only one – but the light bulb has to really want to change. We must ask if financial services firms really want to change.

The economics of pure competition – red in tooth and claw – are not in favour of vulnerable customers. That won’t change and some companies may view the prospect of future regulatory fines as collateral damage outweighed by the raw efficiency of optimised digital journeys where detecting vulnerability is near impossible.

A win:win situation can occur if vulnerability is detectable very early in user journeys – well before the prejudice of big data look-ups – and journeys adapted accordingly. The vulnerable customer can be appropriately guided and served with the right information and signposts, while the non-vulnerable customer served the standard journey.

The next five years is the time for firms to find this courage and embrace the opportunity that vulnerable customers could present to them as well as understanding the risks. It is a step that millions will thank them for.

You can read more on the studies here:

LAB Whitepaper – Consumer Vulnerability Financial Services.

LAB Qual Report on Vulnerable Customers.