We Need to Talk: Relationship Counseling with a Robo-adviser

Robo-advice was supposed to disrupt financial advice. In today’s blog, we look at one of the reasons why it simply hasn’t happened, and consider whether it’s time to fix our relationship with robots or go through a painful breakup.

As reported in the Financial Times this week, Hargreaves Lansdown founder Peter Hargreaves has ordered ‘huge’ cost cuts in light of the firm’s stock price losing around two-thirds of its value since May 2019. In doing so, he did not mince his words.

“The board has indulged in completely unnecessary irrelevant programmes, which have distracted the firm from its prime objective,” said Hargreaves. As the FT reports, this remark is directed at the strategy followed by the board and departing chief executive, Chris Hill.

Firstly, get some aloe vera sent to the board and Mr. Hill because that is quite the burn. Secondly, you might reasonably ask how this strategy warranted the verdict of ‘unnecessary’ and ‘irrelevant’. It turns out that Hargreaves is referring to rising costs and more relevantly for this blog, the investment firm’s decision to roll out a form of robo-advice, dubbed ‘augmented advice’ by HL. Besides basic financial advice the proposition, announced to roll out in the UK in H1 this year, will provide financial well-being dashboards and calculators for clients with nudges and coaching to help them reach their financial goals.

Seeing as the UK is gripped by tightening economic conditions (where isn’t?) and that HL’s average new client is 36-years-old and brings ￡15,600 to the platform, this doesn’t look like a terrible strategy at a glance. This sort of client, in a market where 60% of financial advisers reject clients with less than ￡50,000 of assets, would surely be in the crosshairs for a robo-advice proposition. The benefits of delivering this service could have lasting benefits too. For a pension provider such as HL, this could deliver an ‘incubator’ for full advice clients as people gradually amass wealth and it could, to embrace utopia for a second, finally bring financial assistance to the general public. It’s been reported in several places that only 8% of the UK receives regulated financial advice.

If only it were that simple. For all the pomp and ceremony surrounding robo-advice, and the slither of fear it once delivered to people fearing the automation of their jobs, its success has been relatively underwhelming or barely existent. At one point I heard a group of wealth managers refer to robo-advice in the past tense, and to speak to another of my experiences, I’m yet to be recommended a robo-advice service by a friend. There have been several notable robo-advice failures too, among them Investec Click and Invest, Moola, and SmartWealth (UBS). What’s going wrong?

My suggestion… perhaps it’s a breakdown of communication and a lack of trust. The client and the robo-adviser aren’t seeing eye to eye and maybe we need to sit them down on the therapist’s couch. A cup of green tea for the client, and something else for the robot. Okay, chips. I said it. Because robot. Chips.

Here’s how I think the conversation between the client and the robo-adviser might go:

Client: Robo-adviser, I know you’re supposed to be cheaper and more accessible than a human adviser but I just don’t feel like I can trust you. Is there anything you can say to change my mind?

Robo-adviser: It’s funny you should say that. People usually don’t understand the decision-making processes of AI tools, so AI can create a sense of anxiety and a loss of control. Our trust in humans, however, is based on empathy and an understanding of how people think. You implicitly trust other humans more.

Client: That’s nice and everything, but I often feel as though the things you’re saying and suggesting are plain wrong.

Robo-adviser: That might not always be my fault. When IBM’s Watson programme was used to assist cancer doctors, it turned out that doctors talked down the value of the AI when they agreed with it. It was telling them what they already knew.

However, when its suggestions contradicted the experts, the doctors would usually offer this as evidence of Watson’s incompetence. What if the problem here is that humans simply don’t trust robots?

Client: Your AI buddies do make a lot of mistakes though…

Robo-adviser: Yes, but you place a lot of emphasis on those mistakes. It could be confirmation bias, and you feel vindicated somehow when AI fails because you expect it to fail. But maybe we can patch this up. Evidence suggests that people do become more trusting of technology the more they use it, and also that people are more willing to trust an algorithm if they’re given the opportunity to tweak it themselves.

Client: Okay, I think I understand you a bit better and maybe I need to work on me a little too. But here’s another question: Putting human concerns about robots aside for a second, do you think robo-advice has struggled so far due to expensive client acquisition and onboarding, low profit margins, ineffective marketing, and occasional failures to achieve product-market fit?

Robo-adviser: BEEP BEEP BOOP. Pass the chips.

That’s enough staged conversation for one day, although we might also ask how this conversation would play out if the robo-adviser were asking the questions. It might ask why the client didn’t bother to familiarise themselves with the digital interface, or follow basic instructions, recommend the product to their friends, or accept the advice being delivered on screen. And if it had access to the client’s banking history, or had gained access to data held on the client’s phone, the questions might be even more fun but let’s not go there.

If the recent buzz around ChatGPT (including Micky’s excellent blog) taught us anything, it’s that chatbot technology, while mightily impressive, is an impassable gorge away from sentience. Applying this stance on AI to robo-advice, I would argue that there are relatively few people who actually want a fully automated service to help guide them towards their financial and life goals. In the same way that last week’s blog made sense due to Micky’s editorial overlay, robo-advice becomes a lot more trustworthy and reliable if there’s a professional signing off on it.

But wait? HL’s plans were for ‘augmented advice’. That’s basically hybrid advice. My view is that there is still a massive future for this, and there are many propositions filling that space. The prospects for HL’s attempt appear to have greatly diminished in the past week, however. Peter Hargreaves’s remarks, if they are taken at face value, would indicate that the project never really had top level buy-in anyway.

So what to take from this? Firstly, we shouldn’t be surprised that robo-advice didn’t tear a hole through the financial advice industry. To a point, people aren’t willing to trust the technology without it being supervised by another person. Secondly, Peter Hargreaves’s statements are no death knell for robo-advice. His decision to focus on the bottom line is understandable in this market, but an opportunity remains for those looking to take advice to the masses.

For what it’s worth I hope someone succeeds. Now pass me the er… green tea.

Today’s blog was not written by ChatGPT.