Harnessing the power of big data is enabling financial institutions to drive real change in predictive analytics.
It’s a data-hungry world out there. And one of the most valuable ways in which data is being put to use is its ability to predict the future. Welcome to the world of predictive analytics – an exciting and fast-evolving environment which could have a major impact on every area of financial services.
Predictive analytics market grows
Take the example of Earnest, a start-up financial company which offers a different way of making loans. It uses predictive analytics when credit scoring customers for loans. Rather than just credit history they look at factors such as employment history, career trajectory, savings history and growth potential to draw up a unique financial profile of each customer. They aim to score customers based on what they think they will do rather than what they have done.
Risks assessment is a rapidly growing area of analytics. It’s already being used in the insurance sector. Admiral Insurance, for example, use a little black box to analyse driving style. It can then produce a more accurate view of a customer’s driving style to assess the likelihood of them suffering an accident in the future.
The technology is developing rapidly and though it remains unproven the potential is enormous. Imagine a future were machine learning can take cues from your past, present and even behaviour. For example, it might look at your handwriting on an application form, how long you spend reading the terms and conditions or whether you capitalise correctly, to judge how good a borrower you would be.
Elsewhere banks are using AI and predictive analytics to deepen their relationships with customers. Data about call transcriptions, helpdesk messages, social media interactions and complaints can identify those customers which are most at risk of leaving and help operatives address their concerns.
It can improve cross selling. This is always a risky moment for a bank. If a valued customer is not happy with a product they have been given it could impact on the relationship with the bank. Machine learning can look at what products a customer uses, their buying behaviour, personal situation and predict what they might want in the future – sometimes even before the customer cottons on, themselves.
By analysing buying behaviour they can also improve the detection of fraud. Algorithms can develop a picture of a customer’s buying habits and quickly issue an alert when it detects suspicious behaviour.
One of the most ambitious areas though is in stock markets. Predictive analytics has been a holy grail for analysts for many years. Algorithms can analyse vastly more data and spot patterns that humans miss. What’s been missing until now has been the human element, which is where AI comes in. IBM’s AI technology Watson Super Computer was recently hired to run an ETF by ATF Managers Group. It will actively trade based on an AI model developed by Equbot and Watson. Advanced AI can bring learning, and active judgement to the table in a way that traditional algorithmic trading cannot.
Questions surrounding analytics
The implementation of analytics raises enormous ethical and operational questions. Predictive analytics in credit scoring, could lead to banks running afoul of anti-discrimination laws. Imagine if an algorithm determined that background, ethnicity, or clothing style had a significant impact on credit risk. It raises both regulatory and reputational issues for a bank. Customers are slowly coming around the idea of their data being stored if it provides them a better service. Even so, the concept of banks tracking behaviour personal information appears Orwellian and, for many, will not sit comfortably.
In addition, the innovative nature of these technologies means they are unproven and it’s not easy to demonstrate a clear business case. Furthermore, although they have machine learning and automation at their core they owe much to a very human element. Any algorithm will make its decisions and learnings based on the information programmed in by their human data scientists. They must determine what information it should take note of and what conclusions it should draw.
The potential, though, is so vast that banks almost feel compelled to increase their focus. It’s an opportunity to deepen their relationship with customers and improve the revenue generate per person. It can reduce risks, analyse business performance and put managers in a position to make more informed and better strategic decisions.