Article

Data Analytics: More Than Just Numbers

jumbled pile of shiny metal numbers on a black surface
Contributing Writer

10 minutes

With analytics foundations in place, credit unions can expand their data sources and business intelligence reach.

As more credit unions firm up their data analytic foundations, excitement over early wins may fuel speculation about what else is possible on the business intelligence front. What other organizational issues can be identified and addressed through these evolving tools, and what other data sources are available? 

Weekly sessions on BI planning at $1.5 billion TwinStar Credit Union often “end up going so far down rabbit holes about what we could do” with existing and future data sources, such as the ability to tap into social media, says CUES member Elkan Wollenberg, VP/digital transformation of the Olympia, Washington, credit union. 

“Google Analytics will give us a ton of information about website usage, but if we could compare that with other data sources, such as what members are saying when they call us or come into the branch, that would really expand our centralized view,” Wollenberg says.

Credit union leaders should let strategy guide the identification of new avenues to track down information and deploy data analytics, recommends Emily Engstrom, director of client relations with CUNA Mutual Group’s AdvantEdge Analytics, Madison, Wisconsin, a CUESolutions provider.

Otherwise, she notes, “when credit unions start thinking about what type of data they need to execute their data analytics strategy, they can easily go into analysis paralysis because there are so many sources.”

Start With a Strong Use Case

Information stored in internal systems grew exponentially during the COVID-19 pandemic as members relied increasingly on digital channels, applications and payment methods. And credit unions can further augment their in-house data stores with information from third-party, social media and vendor sources. 

“Access to data is not really the problem. Where a lot of credit unions start to struggle is that they don’t know what they’re trying to solve for. The data itself just doesn’t rain insights,” Engstrom says. “They need to decide what use case they’re trying to tackle and then figure out what data is needed.”

That strategy-first approach offers three benefits, she explains: It’s action-oriented. It will provide some quick wins in solving key problems. And it can be executed relatively quickly and with less cost because the data analytics team won’t need to devote time, energy and resources to gather unnecessary data. 

Emily Engstrom
Director of Client Relations
AdvantEdge Analytics
When credit unions start thinking about what type of data they need to execute their data analytics strategy, they can easily go into analysis paralysis because there are so many sources.

TwinStar CU has four overarching and intertwined strategies—data, digital evolution, member experience and culture—that guide its organizational structure, the allocation of resources for initiatives and the identification of key performance indicators to measure the return on those investments. 

“Data analytics and BI have been on our radar for years now, and along with a lot of other credit unions, we weren’t exactly sure how to execute well,” Wollenberg says. “There’s a big difference between consolidating all your data into a warehouse versus actually consuming it in a meaningful fashion.”

Still, that consolidation process—bringing together information into “a single source of the truth”—is an essential foundation to facilitate data gathering and analysis, he says. TwinStar CU worked with the Knowlton Group, Sandwich, Massachusetts, to export, transform and load data into its warehouse. The credit union hired a director of business intelligence in the summer of 2020 to manage data analytics and work with colleagues across departments on meeting their BI needs.

Over time, the data analytics initiative at TwinStar CU will phase out the “ton of spreadsheets” operations, accounting and other managers rely on to generate reports—with the goal of moving from “I think I know something” to being able to back up (or refute) those assumptions with evidence in near-real time. 

For example, assembling all member data in one place can facilitate a more accurate view of demand for personal finance management tools or card controls rather than deciding whether to offer those services based on managers’ perceptions, Wollenberg says.

Another example is testing internal opinions that TwinStar CU’s online and mobile banking channels are subpar compared to the offerings of megabanks and fintechs. “But if we really look at the data, according to our membership, it’s good enough for them,” he notes. “However, our digital and experience strategies dictate that good enough isn’t acceptable. We want our online and mobile offerings to be a reason for people to join the credit union. We achieve this by leveraging data analytics to hyper-focus our efforts on removing friction points within our online and mobile member experiences.” 

Tap Into an Internal Wealth of Data

TwinStar CU’s primary sources of data are its Symitar core system, MeridianLink online consumer lending platform, and CUES Supplier member Q2’s online and mobile banking channels. It is also working to integrate information generated by a new mortgage origination system and its CRM software. A voice-to-text project is underway to capture the content of member service calls, and the credit union also plans to incorporate member input from its online chat system as part of its three-year vision for building BI capabilities.

With an analytics strategy in place, the starting point for most CUs launching their data analytics journey is to ensure ready access to core system and other internal data, says Vicki Potter, senior analytics performance team lead with AdvantEdge Analytics. From that foundation, she recommends moving on to ACH data, though she cautions that it can be haphazard. For instance, trying to figure out which members use their credit union accounts to pay for Amazon purchases can be challenging when those transactions are posted with multiple descriptors, like Amazon, AMZ and other variations.  

Vendors that offer to conduct those searches have sprung up, but their services can be pricey. “Doing your own searching on keywords won’t be perfect, but it will give you some insights on money that’s leaving the credit union,” she says. 

Credit unions can leverage their trove of internal data holistically by identifying use cases to improve operational efficiency, risk management, and employee recruitment and retention, to name a few examples, Engstrom recommends. By convening a “center of excellence” or cross-functional teams to brainstorm possible business intelligence projects, credit unions can gain an enterprise-wide view of how to make the biggest impact with their investment in analytics. 

For example, organizations that had their data organized and accessible when the pandemic halted normal operations last spring “could quickly jump in and assess which branches to close down and which ones to keep open in some fashion and then alert their members to go to those branches,” she explains. “They could look at ACH data to identify which members had direct deposits for payroll stop and which members were receiving new government employee benefits through unemployment coverage, so they could do some proactive outreach to those members. Other credit unions had to spend weeks up front getting that data together and organized before they could take action.”

Aim to Solve Big Problems 

A major challenge for credit unions in 2021 will be proactive credit risk management by projecting which loans are likely to become delinquent, says Suchit Shah, founder and chief operating officer of CUES Supplier member CU Rise Analytics, based in Vienna, Virginia. 

“Every credit union and bank in the United States has granted payment deferments, so members who’ve suffered pandemic-related income loss could skip payments,” Shah says. “Those missed payments are not being reported to credit bureaus to protect members’ credit standing, making the aggregated credit scores they produce essentially irrelevant for the time being.”

To fill that information gap about members’ current credit standing, CU Rise Analytics has developed a risk score based on behavioral attributes from credit unions’ internal account data that goes well beyond loan delinquency status. Have members’ payroll deposits stopped? Are they making lower payments on their credit card balances? Have they begun tapping into home equity lines of credit and credit cards that previously didn’t carry balances? 

“We’ve created a statistical algorithm that gives higher weight to recent events as the basis for a robust internal credit risk score,” he notes. “It utilizes the credit union’s view of changes in income and changes in payment behavior to present a realistic, real-time view of credit risk, which is much more relevant in the current situation.” 

Equipped with that information, credit unions can guide portfolio management and outreach to members who may be struggling financially to offer assistance with personal financial management and options for loan payments. 

Elkan Wollenberg
VP/Digital Transformation
TwinStar Credit Union
There’s a big difference between consolidating all your data into a warehouse versus actually consuming it in a meaningful fashion.

Tackle Business Challenges

Data analytics can also guide lending leaders in product development, Potter says. She offers a use case of gathering information about the credit utilization of younger members who may be in the market for lending but need to build their credit standing. By enrolling in a credit builder product, those members can then qualify not only for loans but for apartment leases and better rates on insurance coverage. That commitment to helping young people achieve their financial goals can solidify the “members-first” promise behind the credit union brand. 

At TwinStar CU, “we want to get to a point where our data warehouse is self-service, so if a branch manager wants to know more information about a member we’re helping, they can use a web-based interface and get the data themselves versus having to submit a ticket to IT,” Wollenberg notes. “We’re not at the point of self-consumption yet, but we’re building it out as we develop our data governance.” 

TwinStar CU has adapted Tableau, a Salesforce company based in Seattle, as its primary data analytics and visualization platform, and its BI team is working with lending, finance and operations managers on how to use that tool. Marketing currently relies on an internal data analyst to generate reports.

Beyond establishing a shared compendium of data, the credit union’s developing data analytics system also offers a more efficient means to crunch the numbers, Wollenberg says. One of its key successes has been in moving away from using Excel spreadsheets to manage participation loans.

With all the information required to balance these loans, accounting analysts “were pretty much overwhelming their PCs every time they would try to run something, because Excel was never designed to deal with that much data,” he notes. “We were able to streamline the entire process. What used to take days now takes hours. That was a big win for us.”

Dig Into Predictive Analytics

The next big data frontier for credit unions may be predictive analytics, Shah suggests. This tool can be used to create attrition models that assess the likelihood members will move their accounts to another financial institution in the near future and to suggest the “next-best product” in which members might be interested, as just two examples. 

CU Rise Analytics’ attrition scores are generated by a machine learning model that identifies behavioral changes such as year-over-year spending with debit and credit cards and shifts in spending patterns such as declining recurring utility payments or bill pay entries. Credit unions can use those scores to target their member retention and cross-selling strategies, Shah says.

Next-best product algorithms used to be fairly basic, he notes: If a member signs on for X, cross-sell Y. However, modeling has become more sophisticated. For example, CU Rise amasses transactional data from debit and credit card, ACH and P2P payment systems plus financial behaviors over time based on year-over-year trends in spend and deposits. That data aligns with members’ life stages and their likely affinity for financial services much more closely than simply relying on their date of birth. 

Changes in financial behavior signal when members move from college into their first jobs, when they marry and become parents, and when they buy their first homes, Shah says. “Credit unions can’t keep track of those events on an individual member basis, but data can define those life stages” and guide credit unions’ interactions and marketing offers.  cues icon 

Karen Bankston is a long-time contributor to Credit Union Management and writes about membership growth, operations, technology and governance. She is the proprietor of Precision Prose, Eugene, Oregon.

Earn an Executive Certificate in Digital Marketing!

Digital marketing is one of the most effective ways to build member relationships and promote your credit union’s products or services.

Register now for the online Strategy and Digital Marketing Cornell Certificate Program, running April 21-June 15, 2021.
Register Today
Compass Subscription