Picture: 123RF/RAWPIXEL
Picture: 123RF/RAWPIXEL

Capitec is increasing its investment in new technologies such as machine learning and artificial intelligence (AI) as a key enabler in improving its credit risk and its customer experience. This comes ahead of the onslaught of new competition expected in the banking industry.

“We intend to be a world leader in the application of AI to consumer banking. Banks generally, and Capitec specifically, have large quantities of quality data. We protect, manage and govern that data very well and that gives us the best raw materials to start with,” says Graham Lee, head of data and digital solutions at Capitec Bank.

The bank was itself a disruptor to the traditional “big four”  — Standard Bank, Nedbank, FNB and Absa — when it launched  18 years ago with its low-cost and simple retail offering that has made it a darling of consumers and investors.

This has seen it leapfrog competitors in terms of the number of retail clients, with the bank recently tweeting that it had added 266,000 new clients in January 2019 alone, indicating that its client acquisition may be accelerating.

The intention to employ more cutting-edge technology indicates Capitec is preparing for a fight against a host of new entrants that include Michael Jordaan’s Bank Zero, Patrice Motsepe’s TymeBank and Adrian Gore’s Discovery Bank — all expected to launch in 2019.All three have previously stated that technology will be a key feature of their competitive offering.

Lee thinks machine learning can benefit Capitec in three key areas. “Firstly, we improve our understanding of clients so we can better tailor solutions for them. Secondly, we can improve our existing models by replacing rules with machine learning to be more flexible in the way we price credit. Thirdly, we can automate manual processes saving employees time that they can spend better helping the client.”

Machine learning is a sub-discipline of AI that uses large amounts of data to model or predict future behaviour, and it is dynamic, so that “rules” can just as easily be replaced as they can be determined by the system.

In the context of banking, conventional credit extension has often been based on hard or soft rules. For example, a bank would in the past not have extended a loan greater than three times a clients’ monthly salary. It is these rules that can become more flexible based on data gathered from clients, says Lee.

One of the biggest challenges to employing more of this technology is the expertise required to implement it. It was for this reason that Capitec, together with Praelexis, a machine-learning company based in Stellenbosch, brought the Machine Learning Summer School to SA in December to facilitate “the exchange of knowledge and ideas” in the era of machine learning.