How can banks and credit unions thrive in an environment of increased competition, especially from nontraditional sources? Beat the competitors at their own game. Use predictive analytics -- a key weapon competitors wield -- to drive operational efficiencies and strengthen client relationships.
Predictive analytics use historical patterns in data to predict future behavior. They use statistical models to generate a score or classification that represents the likelihood of a specific future event taking place. A credit score is the cardinal example of predictive analytic result. Analytics work best when they are applied across mountains of electronic information, something financial institutions have at their ready disposal. Predictive analytics is usually divided into two categories, each with a different purpose and approach.
Predictive modeling: Predictive modeling seeks to predict the future behavior of an individual person or thing based on the past behavior of similar persons or things. The key here is that it is individual-focused: What is the likelihood of person X doing Y? As a result, predictive models are especially useful for making real-time decisions.
Descriptive modeling: Descriptive modeling uses data about a collection of people or things to identify natural groupings in that collection. Comparison of various data items within and across groups is then used to characterize each group. This underlying data can also be used to assign new people or things to one of the existing groups, as in customer segmentation.
Applications: Banks and credit unions who want to leverage predictive analytics likely already have the data needed to see benefits in three major areas: growth, productivity, and risk reduction.
Growth: Predictive analytics can play a huge role in acquiring new customers, as well as retaining and cross-selling to existing customers. Predictive and descriptive models may be combined with other data to help target marketing to those high-value customers and prospects who are most likely to respond.
Predictive and descriptive modeling of purchase and banking activity data may be combined with descriptive modeling to predict whether an existing customer is likely to purchase a new home or car and generate marketing materials accordingly.
Productivity: Because they are focused on individual persons or things and can be applied in real time, predictive models are well-suited to streamlining processes and increasing productivity. Rather than have staff review each loan application, a predictive model based on historical loan application data may be used to screen incoming applications, redirecting those that fit a specific pattern for further review/processing. Predictive analytics of branch and ATM transaction data may be used to forecast cash needs. Caller ID and customer data may be used to route support calls to the department/area most likely to be able help that person.
Risk reduction: Predictive analytics can help reduce risk. A predictive analytics approach to fraud detection takes a holistic look, going beyond individual transactions to identify patterns throughout the data that may point to fraudulent activity. Predictive models may also be able to identify customers who are at risk for late payments, allowing proactive intervention.
Next steps: Predictive analytics have the power to radically change how banks and credit unions leverage their data to improve services and their bottom line, all without significant new investment. It is important to have a knowledgeable guide to help ensure that models are accurate and effective, without unintended real-world effects.
A powerful analytic solution is one that is well thought out, accurate, and customized to a specific data set. To get started, reach out to an expert to evaluate your data and help identify where predictive analytics can generate the greatest and most immediate impact on your business.
• Tim Connell is a data scientist and solution architect at RedMane Technology in Chicago.