Debt Collector Predicts Payment Outcomes with TIBCO Data Science
New information to drive the process, time and cost savings
Debt collection is tricky. Businesses can't survive if they allow customers to get away with not paying their bills. "But no one wants to lose customers if they can help it," explains the call center manager. "Even people who have a late payment now could become a great customer later. If we do our job well, a customer with an unpaid bill could go from feeling anxiety at the sound of your business's name, to settling the debt and singing your business’s praises."
The agency had been using an internal solution to obtain information about its clients' customers, but was finding it insufficient. "All of our analytical work was based on gut instinct and individual experience," says the call center manager.
Any new data mining and analysis solution needed to support decisionmaking by scoring debts and predicting the likelihood of success for each step of the collection process. The solution also needed to integrate easily with the company's existing solution.
The agency deployed the TIBCO® Data Science analytics platform to help build the models that estimate the probability of whether a debtor will pay or not and specify an optimal collection strategy for each debtor. The solution also cleans and prepares data. TIBCO provided staff trainings on the TIBCO Data Science platform, the theory behind model building, and how models could benefit the agency.
"We were able to implement the TIBCO Data Science solution quickly and easily, using our own employees. And our users find the TIBCO Data Science environment intuitive and simple, even for non-analysts,” says the call center manager.
For tracking a client's past-due accounts, TIBCO Professional Services analyzed the company's previous system and helped the team select data to focus on. They advised the team on client segmentation and helped with model development, implementation, monitoring, and customization of the software for creating daily reports of call center productivity.
New Information that Drives the Process
With the TIBCO Data Science solution, the agency has access to advanced, user-friendly tools for creating data models that depict the various companies, regions, and circumstances across its diverse operations. Using information such as type of client, nature of debt, region, and available contact information, these models help predict whether certain debt collection processes will be successful or how they should be modified to improve success. For example, based on analysis, some debtors may receive a phone call when a bill is past due, others may receive dunning letters.
Model generation in the TIBCO Data Science platform is also completely automatic, with workflows that run the models every night in batch mode to compute probabilities for new debts. These predictions can be used the next day to assign steps and a timeline to the collection process for each debtor.
Time and Cost Savings
"The TIBCO Data Science system allows us to automate analytics and transform our processes—and these changes are now mission-critical within our business," says the call center manager. "Our innovative use of analytics has saved time and money and increased collection of bad debts."
Instead of an intensive collection process focusing on all unpaid debts, the company can now zero in on debts most likely to be paid, greatly reducing call center and administrative costs.
Specialists now easily connect and access data using the TIBCO Data Science query functionality without needing help from IT. By automating certain tasks, the agency saves 52 hours of administrative time each month, one-third of an employee's workload. Automating call center report generation for management review saves team members an hour each day, and report generation now takes only one minute.
Now that analytics is being used within a specific segment of the debt portfolio, the agency intends to expand use of automated analytics within other segments of the debt portfolio and the business.