Print this article

Tech Traps: Beware Style Over Substance - And Siloing - In Risk-Profiling

Greg Davies

1 April 2020

“Stupidity well packaged can sound like wisdom,” wrote Burton Malkiel in A Random Walk Down Wall Street.  It is a lesson well-learnt, and well-used throughout financial services, albeit not always for well-meaning purposes.

Common approaches to risk-profiling have quickly gone from nowhere to temptingly well-packaged triviality, but all too often have forgotten to stop and pick up a sound scientific methodology along the way.

There are plenty of ways to do risk-profiling poorly. One of the latest is potentially the most dangerous, because on the surface, it looks like a great idea.

Playtime is over
“Gamification – increasing user engagement by improving the user experience, specifically by incorporating techniques from games – is rightly a mainstay of behaviour-change protocols. In a complex and, to many, mundane field like financial suitability, not using some gamification techniques feels like an unforgivable oversight.

However, such temptation should signal caution. Gamification is great for engagement, but the techniques alone are not enough.

At Oxford Risk, we embrace the techniques of gamification wherever we can – particularly in the design of user interfaces to enhance client engagement and experience. However, we never gamify at the expense of accuracy. The game is to enhance engagement, not sell snake oil. Gimmicky games trivialise risk tolerance, they do not test it.

There is a time for simplifying, and a time for science. For example, some “tests” favour using the sort of individually intriguing but scientifically vacuous influence of a user’s past investment actions, or even a self-assessment of their risk tolerance. Users like this because it attaches a psychologically meaningful narrative to their past actions, but academics dislike it because it adds nothing, while taking away validity, integrity, and relevance – and it can end up “optimising” for precisely the behaviours we want to guard against.

Form should follow function, not replace it; if you are not measuring what you’re supposed to be measuring, the playfulness of your polish doesn’t matter

Pretty vacant
A focus on a stylish front-end at the expense of the sort of scientific robust substance on which any psychometric assessment must be grounded creates a Potemkin village of a process – great for show, but ultimately not fit for purpose. Capturing clicks is no use without capturing valuable, usable, client insights.

Einstein’s famous (though possibly misattributed) entreaty to make everything “as simple as possible, but no simpler” applies here. When technology tackles complexity, it tends to err on one side or the other: either technically optimal solutions with no care for user experience, or solutions simplified so far that anyone can use them, but where no one learns anything useful from doing so.

A sufficiently optimal solution sits in a sweet-spot that has a deep understanding of both the textbook solution and the behavioural traits and tendencies of its users.

Customer understanding is crucial. But helping clients navigate complexity is better than pretending that it can be cost-effectively avoided. The real returns from an understanding of the customer are preferable to an artificial understanding by the customer.

Being able to trust the outputs of a profiling process means being able to trust both the user’s inputs (of which their engagement with and understanding of what they’re doing are elements) and the methodologies that underpin the design of the assessment and its subsequent creation of the output. Trust needs to be earned with expertise, not masked with marketing.

Behavioural science has a crucial role to play in each of these steps, in ensuring correct functioning, and displaying it in a form fit for easy consumption. But the science must come first.

The quality of a psychometric test is a question of validity and reliability – that it measures what it claims to measure and that when inputs are consistent, so are outputs. Testing the tests requires a silent sophistication: complexity beneath the surface that is not necessarily evident on the surface.

An effective question set is like a team, or an orchestra: more than a mere collection of individual parts, the correlations between them count too. Picking the best team requires trials to see which elements work best together

Suitability shouldn’t stop at the start line
The complexity investors need to navigate is a function of the number of moving parts involved. Because investment markets move around more than an investor’s relatively stable willingness to trade off the chance of bad outcomes for good ones (i.e. their risk tolerance), traditionally most attention is paid to risk tolerance and the wider suitability process at the outset.

However, over the course of an investment journey, it is the moving human parts – a panoply of behavioural reactions – that are arguably more worthy of attention. Suitability is dynamic; it suffers when seen as a snapshot.

Many aspects of technology are insufficiently creative because the client “profile” is considered as only an onboarding issue, segregated from the reporting and relationship management that influence investor-investment interactions through changing circumstances.

Humans do not turn into robots when they start to own investments. Siloing risk tolerance into a bucket of onboarding chores leads to suitability and client-satisfaction risks, and lost opportunities in sales and engagement because of a rushed, and incomplete, approach to client attitudes. Just because a transient behaviour shouldn’t be baked into an investment solution, it shouldn’t be ignored in deciding how that solution should be presented and managed over time.

The person who makes a plan is rarely the person a plan is made for, whether that’s the alert and inspired future gym-goer of New Year’s Eve turning into the tired and emotional duvet-hugger of New Year’s Day, or the calm investor sitting with an advisor for an hour turning into the confused one reading the news six weeks later.

Simple, but not simplistic
Too often, to borrow a phrase from historian Will Durant, “The fertility of simplicity defeats the activity of intelligence.” Engaging investors with the profiling process is vital, but if it’s done at the expense of competently measuring what you need to measure, then it’s both dumb and potentially dangerous. Regulatory risks rise as the seriousness of testing investor attitudes to risk falls.

Forgetting what you’re trying to do and why, in favour of how you’re doing it, is a common error when designing shiny new technological toys.

Humans and tech perform best when they play together. Managing moving financial and emotional parts benefits from blending human and technological qualities. Humans are good at some parts of the suitability process. Tech is good at others. They each have distinct, complementary, roles to play. Tech should be leveraged to help humans navigate complexity, not add another layer of it, or become an end in itself. As simple as possible, but no simpler; beware both the simplistic and the over-engineered.

Technology offers the opportunity to produce rich and accurate financial-personality assessments at scale, that in turn can be built in to hyper-personalised approaches to engagement, communication, portfolio construction and reporting.

Well-designed digital platforms deliver personalised, easy-to-use information to clients which is shaped by their behaviours. By taking the legwork out of the risk-profiling process, technology can save human energy for appreciating the ambiguity inherent in its interpretation.

But it can do this only when being good is followed by looking good, not replaced by it.

This forms part of this publication’s latest research report, Technology Traps Wealth Managers Must Avoid. Download your free copy by completing the form below.