There is much talk in the wealth management industry about the benefits brought about through advances in artificial intelligence (AI).
A possible next step for robo-advice will be to make operations more efficient by automating customer acquisition – the onboarding process. This will build upon the existing robo-advice model from what is essentially "digital wealth management," as one provider described itself, into a more sophisticated machine capable of gathering detailed know-your-customer (KYC) data to help firms broaden their advice and service offerings.
The robo-advice model has been shown to cut fees and charges and, as a result, has been held up as having the potential to plug the advice gap. However, it is worth taking a look at what robo-advice has achieved so far to determine whether it has brought value and benefits to customers as well as firms.
Like traditional wealth management firms, robo-advisers have varying business models, but all have the same remit: risk-profiling customers and aligning them to model portfolios. In most robo-advice scenarios, the customer must answer a set of questions about risk, much as they would do in a traditional advice process. This helps assess their attitude to risk, knowledge and experience and capacity for loss. Having completed the questionnaire, the customer is typically matched to a model portfolio or might be warned that their answers mean their chosen risk profile is not suitable.
Key points
- A possible next step for robo-advice will be automation of the customer acquisition process.
- Robo-advice has brought perceived value to customers in reduced charges.
- Robo-advisers' systems can communicate with and inform customers at the right time.
The biggest risk with this is that a systemic issue has been overlooked in the design of robo-advice models, which could result in mis-selling. In previous instances of mis-selling, the firm might have failed to have oversight of a few unscrupulous advisers or the odd rogue trader. With robo-advice, if the system has been incorrectly configured without appropriate monitoring, there could be widespread issues to the detriment of customers.
The risk profiler is a first step in the process to establish the risk a customer is willing and able to take. The question is, can the robot effectively take on the next step of intervention and questioning, or should this pass to a human?
Robo-advice has developed in the post-financial crisis world and has brought perceived value to customers in reduced charges. However, history has shown us that the true test of a firm’s proposition and quality of its controls is how well they hold up in a falling market. With political and economic uncertainties in the news and testing market confidence, record highs have given way to volatility. When correction comes – sooner or later – it will provide the first true test for robo-advice. Any correction could show us how well the systems hold up and how well customers have understood the offerings.
Some models include a hybrid approach, meaning that, when the circumstances dictate, the customer is passed on to a human adviser to introduce emotional intelligence and pragmatic intervention, but even this relies on the programming to be right in the first place.
How well are firms monitoring their systems? Even within the hybrid approach there is a risk that systems will not redirect, or expel, customers at the appropriate time.
Take a closer look at the risk-profiling techniques these firms employ and more risks begin to appear. The risk-profiling tools employed in the robo-advice world are not new nor fundamentally different from those employed by traditional wealth managers, and these have been shown to carry risk.
The challenge robo-advice firms face is how they manage those risks. Well thought out, rigorous and continuous outcomes testing can demonstrate that systems are fit for purpose; identifying potential limitations and assisting in their mitigation.
Firms using robo-advice should have first considered their target market. Given the journey is automated, the typical customer of these firms should have a good grasp of financial services and an understanding of their own needs and objectives. The onboarding processes these firms use need to draw out and identify when a customer does not fit their target market or needs further support.
One of the great strengths of robo-advisers is that their systems can communicate with customers at the right time. All of the firms Huntswood contacted for research provided illustrations of potential returns and gave information about the likelihood of achieving returns as well as providing key features documents and/or key investor information documents (Kiids) at the appropriate time. That said, the research conducted by Huntswood did highlight some risks.
When testing these tools, one provider highlighted that there were inconsistencies with answers given by a tester to the risk profile questionnaire and their chosen risk profile, but did not highlight what these were. It also did not prevent the tester from setting up the portfolio, but asked for confirmation that they understood the risks. In the absence of any explanation of what these risks were, it is questionable whether this approach goes far enough and provides sufficient protection to customers.
Typically, in this scenario, I would expect a firm to highlight the specific inconsistencies in a customer’s answers and, rather than asking the customer to confirm they understood the risks, actually call out those risks. In some cases, there should be a point where the customer is unable to proceed due to a clear misalignment between their risk appetite and their investment mandate.
Enhancing these controls seems an obvious next step for robo-advice. If robo-advice AIs can be developed so that they understand the environment and recognise customer responses to the point they can identify levels of unsuitability, that would be a great leap forward. This would not only help robo-advisers manage their risks and provide protection to customers, but could also be applied to more traditional wealth management businesses and even financial planning. Automation in this area would bring major improvements in processing and undoubtedly reduce compliance and front-line checking costs in the longer term.
In the meantime, firms should have alternative controls in place to monitor their systems and the effectiveness of their risk profiling tools. Being an automated process, robo-advisers should be capable of producing management information that captures anomalies in customers’ responses and customers with potentially unsuitable portfolios. After obtaining that data, firms need to act upon it and ensure customers have understood the proposition and are aligned to their target market.
If firms are not already doing this, we may find that when a market correction comes, there is a lot for the robots to learn.
James Kelly is a senior consultant at Huntswood
