As wealth managers look to AI for cost or UX benefits, what do they need to know about robo-advice, behavioral analytics or the value of clean data?
Artificial intelligence (AI) is increasingly being used in wealth management as the catch-all term for next generation capabilities to attract and retain customers.
From personalized portfolio management to customer behavior analytics, the potential for significant UX enhancements and cost reduction is vast. But progress is only possible if there’s clean, well-organized data to start with.
We take a look at the technological initiatives behind AI and how they apply to the wealth management industry.
The components of Artificial Intelligence
Natural Language Processing concerns itself with the creation of systems that can understand language such as identifying Company A as being a supplier of Company B, identifying entities or sensing sentiment within passages of unstructured text.
Machine Learning involves training computers to make predictions resulting in specific behaviors or answers.
Training seeks to identity the most relevant features and to optimize performance improvements given new training sets and new data. It does this by using supervised learning (predicting Y given X) or unsupervised learning (either clustering problems, such as grouping customers based on their attributes, or association rules, such as investors who own X tend to own Y).
This leads to trade-offs between prediction accuracy (precision) and result completeness (recall).
While Machine Learning looks at feature combination (combining the attributes of the data), Neural Networks and Deep Learning aim to uncover the features that best represent a problem at a more granular level. For example, facial recognition using machine learning involves identifying facial features, while Deep Learning performs the analysis at a pixel level.
Inference Engines enable computers to identify contextual triggers, typically through the use of sensors, such as going to sleep, going for a run or landing in a foreign country. Today’s smartphones include around 10 sensors that generate data for attributes such as location, surrounding brightness and user activity.
Finally Knowledge Representation covers the activity of representing data in such a way that computers can make use of it.
In practice, this is one of the most complex and important facets of artificial intelligence as it determines whether the computer can have a useful dialog in natural language or make an appropriate prediction or recommendation successfully.Discover how to optimize your productivity and engage your client with Thomson Reuters Wealth management solutions
Applications in wealth management
The surge in fintech has been driven by the attention of incumbents being diverted towards regulatory compliance, leaving new entrants to take advantage of cloud computation and storage offerings backed by investors seeking higher yields in a low interest environment.
For Wealth Management, AI is being focused on today’s biggest drivers behind customer acquisition and retention: user experience and cost.
Content personalization will be a focus in Europe where the advent of the Payment Services Directive 2 (PSD2) will mean that financial Institutions need to make APIs available. Aggregators will spill over into wealth management and differentiation will be created by delivering content with superior relevancy to users.
Customer behavior analytics is the focus of start-ups like DataRobot, Sift Science and BehavioSec who aim to reduce operational and business risk by targeting fraud and money-laundering activities by identifying suspicious patterns early.
Customer service will see firms like NokNok Labs complement operations by performing user authentication via voice recognition and evolve into full conversational platforms like those of Clinc, Kasisto, and Penny.
Customer behavior analytics will contribute to opportunity management to predict client attrition, and identify opportunities for upselling and cross-selling products, but also by pushing clients to digital-only channels or even closing accounts.
Major life events are common sales triggers, but data generated by inference engines such as CleverNudge will enable new interactions based on lifestyle choices.
Behavior analytics will also apply to the firm’s advisors. It’s entirely feasible that advisor allocation will be based on the data-driven features of the advisors rather than the gut feel of the management. This will likely become particularly transparent as people seek ways to compare themselves with their computer counterparts.
Robo-advisory offerings have targeted the portfolio management space using learning-based advisory models to personalize and optimize asset allocations.
This will spill over into more rigorous market analysis and new investment strategies.
The clean data requirement
While AI will lead to significant user experience improvements and cost reduction, it relies heavily on having clean well-organized data.
Unclean data leading to prediction accuracies below 50% mean you would have been better off just flipping a coin.
So do you clean your existing data, enhance your legacy systems or move to new systems?
There is no right answer, but tomorrow’s winning firms will have used investments in GDPR to modernize their infrastructure in order to solve the root problems around management of customer data. And they might even have leveraged the change in customer data ownership coming from PSD2 as a way to get customers to clean their data for them.