Why FI’s need Predictive Modeling
Not all attrition models are created equal. Some focus on identifying product closures - yet, by that point it might be too late. Others focus on identifying traditionally slow-moving data, such as branch visits to predict churn - introducing a roller coaster of variables, fluctuating often, and providing a false sense of churn indication.
A better approach is to look at the full universe of customer data - both product data as well their everyday purchase spend data - providing a more holistic view and helping an FI identify the early warning signs that a customer is reducing their commitment to the institution. Models that leverage a customer’s spend transactions, held-away payment activity, banking behaviors, and product mix consider the full 360-degree view of the customer in evaluating their likelihood to attrit. It’s the uniqueness of the transaction data - such as identifying customers making micro-deposits into Chime, or a drop off in automatic withdrawals of car insurance payments, or an increase in the number of monthly payments made to competing institutions - combined with product data that is the predictive modeling approach FIs need now.