Analytics Offer A Smarter Approach To Collections And Underwriting

A new approach to debt collection at WSECU is cutting contact center costs by $8,000 a month and identifying borrowers that need early intervention.


With COVID-19 turning the economy on its head, credit unions are well advised to identify and support struggling borrowers sooner rather than later. For the past couple of years, the analytics team at WSECU has helped the credit union determine which borrowers are at risk of defaulting and who might just need a push to pay. Borrower data also helps the cooperative determine how to reach out and via what channel.


Washington State Employees Credit Union
Data as of 03.31.19

HQ: Olympia, WA
MEMBERS: 276,121
ROA: 0.90%

It costs approximately $1 a minute for the average call center to serve a caller. That’s according to callback solutions provider VHT. As such, reducing call volume by shifting calls to automated phone systems, online banking, and mobile apps can translate into big savings.

For credit unions, outbound calling is an important channel for borrower communication, debt collection, and charge-off reduction. However, how can a credit union determine which borrowers to call? How soon and how often should the credit union call? And, which slow payers are at risk of becoming delinquent borrowers?

Analytics and member data are helping to answer those questions at Washington State Employees Credit Union ($3.0B, Olympia, WA), which has doubled in size in the past decade and today has branches stretching from Seattle to Spokane. 

WSECU is employing analytics and machine learning to make smarter, more efficient use of member solutions department resources, reducing costs related to its third-party collections partner and improving collections results across its loan portfolio in the process.

Bill Peterson, vice president of analytics and insights at WSECU, says his team has gained numerous insights for predicting borrower behavior and prioritizing collection attempts. But to get here, the credit union had to take several swipes at analyzing data and fine-tune its methodology.

Tell Us Your Story

How is your analytics department using data to help members navigate the new economic realities brought on by COVID-19? Tell us your story.


Initially, the team at WSECU looked at whether a pre-built collection score from a credit bureau was useful in predicting a loan charge-off. Early on, the team discovered the collection score wasn’t a meaningfully better predictor than the member’s last reported FICO score, which the credit union already collects for risk purposes.

So, the credit union combined the FICO score with other member data and created its own custom collection scoring system.

“We used machine learning algorithms to generate the scores and found they worked really well across a range of products,” says Peterson. “While the overall methodology employed is proprietary, I can share that it includes dimensions that weigh the strength of the member’s relationship with WSECU, as it includes an understanding of the member’s previous history of default. This gives us the ability to better identify members who will ‘self-cure’ [repay] based on less-expensive contact methods.”

Armed with the custom collection score, WESCU can identify members who need fewer contacts or who are just as likely to respond with automated phone reminders. That’s helped WSECU significantly reduce call volume handled by its collections vendor and has saved the credit union nearly $8,000 per month.

“We’re also improving member experience by using contact methods that are less intrusive than a live person calling,” Peterson says. “This more impersonal touch by automated phone messaging can serve as a nudge with results that are just as successful.”

On the flipside, WSECU also can identify members it needs to contact earlier because they have a higher risk of default. 

“We have more meaningful conversations with the people who need our support sooner in the process,” the VP says. 

No Credit Check Necessary

WSECU’s analytics department is a central support group and aids in the success of QCash and QCash Plus, an instant loan program with a low fee and no credit check. QCash grants loans between $50 and $700, while QCash Plus grants loans of $700 to $3,000. WSECU uses a statistical modeling approach and internal scoring to underwrite the loans.

“The model has been in production for a little over a year and we’ve seen great results in charge-off reductions,” says Bill Peterson, vice president of analytics and insights at WSECU.

And what about mailed notices and physical communications? Peterson says the organization as a whole is looking for ways to communicate with members in the channels they prefer. 

“That work is more complicated and would likely be less impactful than our initial gains,” Peterson says. “But, it’s the next obvious step in extending the models into different areas of the credit union.”

WSECU’s analytics department is a central support group for the Evergreen State cooperative. As such, the department also is using analytics to improve results for other parts for the credit union such as marketing and risk.

“One area we hope to get more traction in is product recommendations,” Peterson says. “This area lends itself to the type of automation that data science works best with, and we think it will be a useful growth tool.”

WSECU has been working on the product recommendations it makes through its new online banking platform. However, the biggest technology challenge there lies in how to automate the delivery of the recommendations rather than what model or analysis to deploy.

“Going forward, we feel like this is going to be a major use of data science within the organization as it lends itself well to automation and machine learning,” Peterson predicts. “There are a lot of promising opportunities for using analytics.”

Want more credit union strategies? Sign up for the free newsletter.