What does Netflix have to do with online banking?
Banks and credit unions of all sizes face the challenge of finding ways to reconnect with customers, sell more products and services, and remain vital to their communities. This is especially true as online and mobile banking become the norm.
The Netflix recommendation model highlights machine learning in action and offers a best-practices approach to help address the challenge.
Netflix, without a doubt, is a major success in tailoring entertainment in line with customers’ tastes. Curating a dizzying array of available choices into a custom set of offerings for its viewers, the Netflix model could be utilized by financial institutions (FIs) to truly understand its customers’ needs and to predict what products a customer is most likely to adopt.
Collaborative filtering at work for FIs
Q2, in an extensive research effort, sought to identify those customers most likely to benefit from and be interested in a banking product based on an overall analysis of the products customers hold across a base. After spending significant time gathering and studying online and mobile data, we applied collaborative filtering-like Netflix uses-to make predictions about end user preferences using the collected information across the set of users.
In the case of Netflix, there is a set of data that characterizes customers’ preferences such as movies viewed or products owned. Netflix then identifies the best choices for each customer based on the choices made by similar customers. In Q2’s experimental exercise:
- We identified customers who would be the most likely to prefer a new product or service-in our exercise it was an auto loan.
- Then, we measured whether the customers in the selected group opened new auto loans.
- We found that identified users from our collaborative filtering model were three times as likely to open the new product as a user from a larger, more general group, indicating the algorithm for selection has strong predictive power.
The way Q2 approached machine learning and predictive behavior is discussed in the whitepaper “The Data Dilemma: Unlocking Customer Insights with Machine Learning.” Written by Q2 CTO Adam (Anderson) Blue, the whitepaper takes a deeper dive into the important lessons gained in tackling machine learning and behavioral modeling, with a special section on collaborative filtering.
Q2 SMART: Driving Insights for Banks and Credit Unions
Much of what we have discovered in our data research has become the foundation of Q2 SMART that delivers the insights banks and credit unions need to grow their products per household. Non-IT staff can use Q2 SMART to tailor specific marketing messages and offers to the right account holders at the right time to drive product and services adoption and, more importantly, build relationships in the digital banking age.