Which customers are likely to default on personal loans? Which will probably become profitable, long-term customers? How do you assess the risk of someone who lives in an informal settlement?
For South African banks wrestling with challenges of reaching millions of underbanked people, the answers to these questions will have a direct effect on the bottom line. 

That’s why they are turning increasingly to new-generation technology, known as predictive analytics, to speed their penetration of the vast underbanked market, take new product ranges to high-risk sectors, raise profitability and measurably reduce risk in a competitive marketplace.
Predictive analytics is a dimension of business intelligence that allows organisations to assess both risks and opportunities.
Cliff Court, Chief Technology Officer of Cape Town-based Grapevine Interactive, says it allows banks to safely enter new markets more rapidly and maintain acceptable risk levels in existing markets.
Compared with existing methods, Austin Logistics’ predictive analytics software solutions provide very early-stage accurate detection of potential risk and fraud behaviour, literally from the first transaction.
This contrasts with traditional methods, which can take six months to generate accurate risk assessments and means that, during the first six months, accounts are at risk for fraud losses.
“Using an extended set of measurements, companies that extend credit are now able to make decisions about opening new accounts, issuing loans or marketing new products based on expected consumer behaviour,” says Court.
Court says that in the South African context, assessing risk may be more difficult, and at the same time, more important.
Although software-based decision-making is sometimes seen as mechanical, Court says such systems are subtler than a human-based evaluation, since people cannot always take in all the factors in making an informed decision. Realtime predictions give organisations a significant edge.
“Systems can be set up to scan data daily to look for a customer that a risk pattern. Alerts can then be sent to customer-facing personnel to take action. Call centre representatives can even check on those indicators while talking to a customer and see opportunities in realtime,” says Court.
He stresses that this kind of technology is not only about limiting risk, but promoting additional customer satisfaction and comfort, thereby rewarding good customers.
“There is no doubt that predictive analytics can provide added business value and business benefit to a range of businesses,” says Court.
“Most organisations have good controls, and even some analytical capacity. A products like Austin Logistics doesn’t just give you a statistical score based on segmentation, but provides you with a mechanism that includes business rules, workflow and compliance, removing the need for any manual processes or interventions.
“Importantly, it allows you to make a decision at an individual accountholder level – effectively, it’s what CRM promised but never delivered.”