Serving Cardholders With Predictive Analytics

Businessman reviewing analytics report on a tablet
Contributing Writer
Fab Prose & Professional Writing

3 minutes

Today’s card campaigns require a depth and breadth of predictive analytics and a deep understanding of data.

Gone are the days of relying on traditional query methods to bolster card portfolios. Today’s campaigns require a depth and breadth of predictive analytics and an understanding of data from a credit union’s perspective.

“Building stronger card portfolios requires the coalescing of all data so you can paint a picture of your true member,” explains Jeff Carelli, SVP/data and analytics for CUES Supplier member PSCU, St. Petersburg, Fla. “By observing an array of correlations from all available data, including those between debit and credit card data, loan transactions, purchasing and payment patterns as well as channel usage, we find meaningful correlations and clusters of cardholders who behave similarly.”

Predictive analytics not only helps us understand buying behaviors more deeply, adds Carelli, but it also reveals unexpected usage patterns and smaller subsets of people to provide offers to (new micro-segments), at all ends of the spectrum. “Those smaller subsets enable us to execute targeted campaigns with fewer false positives, which increases effectiveness and reduces cost.”

Carelli shares an example from a recent back-to-school spend campaign. “Using predictive analytics, we analyzed cardholders with and without children and their spending habits over the back-to-school shopping season. We examined clusters of cardholders of two to four years with children, cardholders of two to four years without children, and new cardholders with and without children. 

“Predictive models identify that when cardholders of two to four years with children have lower year-over-year spend during the back-to-school shopping season, they become much more likely to attrite in the near future. The other clusters show no such change in attrition risk, even when they have spent less year-over-year during the back-to-school shopping season,” Carelli says.

The CU can then focus retention efforts and offers on this specific at risk segment of cardholders. “A key benefit of this targeted approach is lower retention offer cost,” he adds. “The predictive model used data to determine relevant variables, as opposed to a human deciding to focus on certain assumed or arbitrary values.”

Used wisely, predictive analytics can boost card portfolios in a variety of ways. 

“Analyzing behaviors of the past and using predictive models going forward can reveal non-traditional patterns as well as useful assumptions,” concludes Carelli. “Ultimately, this can help CUs increase their card transactions, card spend and even recognize the risk for attrition, for greater campaign effectiveness.”

As part of its comprehensive data and analytics offerings, PSCU’s Member Insight enables credit unions to query data, view structured reports, and create their own queries. Carelli notes that about 80 percent of CUs tap into the system’s collective knowledge of ready-made queries. But for the 20 percent of CUs that have resources to devote to query analysis, the system offers a robust solution. “It’s a win-win. Eighty percent benefit from the collaboration while the other 20 percent drive us to be more innovative and find new, exciting and relevant solutions.”  

Stephanie Schwenn Sebring established and managed the marketing departments for three CUs before launching her business. As owner of Fab Prose & Professional Writing, she assists CUs, industry suppliers and any company wanting great content and a clear brand voice. Follow her on Twitter@fabprose.

CUES Learning Portal