Transforming Credit and Collection with Predictive Analytics
According to a Federal Bank report, more than $600 billion of household debt in the U.S. is delinquent as of June 30th, 2017. Out of which, $400 billion is delinquent for more than 90 days. This shows the need to rectify this issue before it becomes a crisis. A crisis that not only poses a serious threat to consumers, as debt accumulates interest, but may also cause trouble for businesses, as there is a chance that it may eat into their revenues. And for these companies, it all boils down to collection of as much debt as possible, as quickly as possible in order to avert any further issues. And for that, they are looking up to new-age technologies.
Today, it’s no secret that most forward-thinking businesses are keenly following the latest developments on big data, artificial intelligence, machine learning, and predictive analytics. And they are even willing to dig deep into their pockets to implement some of these technologies to increase revenue, lower costs and improve their business processes.
Cracking the Debt Collection Process
It is estimated that one in three Americans has some type of debt in collection. And to add to it, the credit industry is facing several uphill challenges, with the revenue per transaction slowly decreasing. All of this has led to the establishment of several bodies and regulations focused on collection methodologies such as the Consumer Financial Protection Board and the Fair Debt Collection Practices Act (FDCPA). The current debt collection landscape hints that technology’s intervention can harness a lot of positive impact on improving collections.
It all boils down to how we negotiate with the customer. Of course, good negotiation techniques are beneficial to everyone – the consumer, the creditor and the economy. In several instances, the consumer is presented with an opportunity to improve his/ her credit history and future creditworthiness. Thereby pumping money back into the economy.
Technology is already proving its mettle in improving the debt collection scenario. One such technology is Artificial Intelligence. Machines have developed the ability to learn things that will empower humans and businesses the world over. Using algorithms, AI is now able to store data before making a prediction about something – such as when a debtor is likely to pay. And this data is crucial in taking the necessary steps to ensure successful debt collection.
Another such technology is Big Data Analytics, which helps in acquiring the most crucial information about a debtor. Information like demographic data or behavioral aspects, such as the time of the day when the debtor is likely to respond to a call, makes a great difference. With Big Data, it is possible to acquire and segregate data with laser sharp focus with respect to one singular debtor.
Predictive Analytics, a form of advanced analytics is also making great breakthroughs in the solving the debt collection problem. By clubbing various techniques like data mining, machine learning, artificial intelligence and statistical modelling, it makes predictions about events in the future. One such interesting case study is WNS.
WNS is a leading utility company, which put predictive analytics to use. An improvement of 50% in debt collection was seen in just 3 months time, that too without any loss on customer interaction. This goes on to show the potential it holds.
Enabling Our Clients to Improve Their Collections Using Technology
One of our clients was facing a similar issue. Their business offers its customers two types of loans: Short-term unsecured loans and installment loans. And as a protocol, their collection process begins with getting in touch with their customers and seek assurance from them that they would make their payment on a specific date or follow up on an earlier decided promise or a scheduled payment.
Their issue at hand was to decrease the Portfolio at Risk. With an increase in bad debts, we were approached to step in and work with the collection team to come up with better collection strategies and provide better insights on the installment loan portfolio. Along with that, we were also assigned to develop a system that gets the dialer data into the clients’ database instead of the dialer systems proprietary DB. The data included all the available ‘Return Items’, which is the expected payments, fees and interest owed in the upcoming days.
We then identified that, by analyzing the days-past-due data that we have to come up with better strategies to ‘cure’ (make delinquent ones to current) Installment loans. And for that, it was important to a 30-day threshold mark.
Also, we noticed that, earlier, the data consisted of only a few metrics, based on which the team classified its customers. We thought giving a better insight into the customer’s credit history and income modes would help. So, we updated the data provided into the dialer to provide a better classification of customers and come up with strategies to handle them.
We have also observed that a majority of the numbers provided were invalid or not existing. So, we revamped and cleaned the data to provide a cleaner list of phone numbers, which in turn decreased the idle time of agents.
Then, we held meetings with the BI team to conceptualize how we can execute these new strategies and report back to the client about the daily outputs of various collection campaigns. Additionally, we discussed intensively with the IT team about the possible Data transfer and connection to the DB.
This way, we put together a whole new system in place that is making the process of debt collection much easier and more efficient for our clients.