How Toyota Revamped Its Collections Biz with Big Data Analytics

Toyota wasn’t the only automaker to suffer during the Great Recession. But when the volume of Toyota customers behind on car payments spiked to record levels in 2009, the company decided to overhaul its approach to collections using big data analytics.

If you go into a Toyota dealership today to buy a car, it’s likely that you’ll be introduced to Toyota Financial Services, the auto giant’s banking arm. With more than 4 million customers and an $80-billion portfolio of car loans and leases, the company is sizable in its own right.

When the financial crisis of 2008 gave way to the Great Recession of 2009, the number of new car purchases plummeted and the number of people more than 60 days behind on their loans hit a new record. In 2009, the number of delinquent loans on Toyota’s book spiked by 25 percent compared to 2006, reflecting the broad job losses and significant financial distress occurring at that time.

“That was the first time we ever saw 100,000 people per day behind in their car payment,” says Jim Bander, national manager for decision science at Toyota Financial Services. “Continuing to do things the way we had already done it didn’t make sense anymore.”

Up to that point, TFS followed the textbook “bucket-based” approach to collections. A customer at the earliest stages of delinquency would first receive a robo-call. That would be followed by a call by a customer service agent using a predictive dialer system. Finally, the case would be assigned to a collections agent, who would use a variety of means to resolve the situation.

“That’s the basic text book approach to collections and it just wasn’t working anymore,” he says. “We could no longer treat every customer with a one-size-fits-all approach.”

Three-Pronged Strategy

As the economy recovered, Bander oversaw a three-stage process to overhaul the decision-making systems at TFS. The strategy collectively is referred to as the Collections Treatment Optimization (CTO) program.

The first process involved implementing a new decision engine from Experian (LON: EXPN), the credit bureau and analytic tool provider. The firm’s second stage broached the realm of predictive analytics, and involved building custom scorecards that allowed TFS to assess the risk level of each invidious costumer based on various data points.

Toyota sought ways to keep people in their vehicles when their car loans started to sour (Kurhan/Shutterstock.com)

The final phase of the CTO project began in 2011 and involved deploying an advanced prescriptive analytic solution. Bander says the solution, which was developed and implemented by FICO (NYSE: FICO), a provider of credit scoring tools and other data products, enabled TFS to employ a statistically and empirically sound approach to analyzing millions of records.

“What FICO brought was bringing our predictive analytics to the next level, which involves micro-segmentation and combining multiple models to get one result,” Bander tells Datanami. “It involves action-effect modeling, which allowed us to figure out, when we take an action, what are the individual customer microsegments likely to do in response to that action.”

Crafting Outreach

The most important decision made in the collections process is the outreach strategy, Bander says. “We can’t help them until we speak with them.” By calculating how a given customer is likely to respond to a robo-call or a call from a human, versus receiving an email or a text, TFS is able to accelerate the collections process and arrive at a resolution that benefits everybody involved.

The FICO solution (built on FICO Xpress Optimization Suite and FICO Model Builder) doesn’t dictate which communication strategy collections managers must use. Rather, it gives them a choice of relatively good options, and allows them to select the one that best fits the situation, Bander says.

In a big batch job that runs every day on its IBM (NYSE: IBM) Netezza data warehouse, TFS analyzes all of its customer accounts to build “what-if” models to determine responses to customer situation.

“We crunched the numbers the night before to say ‘If a customer calls and says they’re in a financial bind and would like to skip a payment or two, does that make sense for the individual customer? Is it likely to be a temporary situation or is it more likely that they have a serious shock in their life and we need to get them into a less expensive vehicle?'”

Real World Results

The results from the CTO program are in. According to TFS, it has helped 1,600 customers avoid having their car repossessed. And it’s helped 10,000 customers avoid reaching delinquency, which is defined as being 60-days late in their payment.

What’s more, TFS was able to grow its portfolio by 9 percent without adding headcount under the CTO program. That all contributed to TFS winning the 2015 FICO Decision Management Award for Debt Management in November.

While FICO did the “heavy lifting” on the implementation of the prescriptive analytics and the data integration routines it relies on, there were a number of other analytic vendors also involved in the project. In addition to IBM and Experian, TFS relied on products from SAS, Oracle (NYSE: ORCL), and Tableau Software (NYSE: DATA) to make the TCO project a success.

Careful vetting of the latest analytic tools and capabilities was a big requirement in this project, Bander says. “There’s no way we could have done this type of predict modeling, this type of visualization and this type of optimization, using technology that was available 10-12 years ago,” he says.

Overseeing a project like this that takes a “best of breed” approach requires not only stitching together disparate products and data, but deft navigation of political turf. On top of all of this is Toyota’s corporate culture, which leans heavily on nemawashi, a Japanese word that essentially means gaining consensus across a wide swath of stakeholders.

“I’m a proponent of fitting your analytics into the corporate culture,” Bander says. “Toyota’s corporate culture [is] based on continuous improvement and respect for people. So an awful lot of that involves getting people from multiple departments and giving them all a voice.” In this case, Bander consulted his colleagues in the IT, business intelligence, legal, finance, accounting, and training departments.

Banks came under a tremendous amount of pressure when loans and mortgages started souring during the Great Recession. Toyota Financial Services was subject to that pressure too. But thanks to shrewd application of big data technology, the company found an innovative solution that benefited Toyota, as well as its customers.

“At the end of the day, we’re a mobility company,” Bander concludes. “Financial services is only one part of what Toyota is about. If we can keep customers in cars, we’re fulfilling our core mission.”

Related Items:

How Big Data Analytics Is Shaking Up the Insurance Business

How Deloitte Transforms Clients into Insight-Driven Organizations

Automakers Embrace Predictive Analytics to Boost Sales

Editor’s Note: The article was corrected. Thanks to the CTO program, Toyota was able to help 1,600 customers avoided repossession, not the 6,000 we previously reported. The program also helped 10,000 customers avoid reaching delinquency, rather than the 50,000 originally stated. Datanami regrets the error.