BankFinancial F.S.B. has a roughly 1.5 percent rate of customers who terminate their bank relationships for unknown reasons. Eager to combat the trend of customer defection, BankFinancial is moving off of its first-generation marketing database and adding a new analysis tool that eventually will merge disparate data sources, support a complete view of each customer, and provide the knowledge necessary to retain and improve customer relationships.
Chicago-based BankFinancial has $1.6 billion in total assets and operates 16 locations. The bank relied on a first-generation marketing customer information file (MCIF), or marketing database, and an application called Clementine, from predictive-analytics software provider SPSS (Chicago), to conduct predictive analysis and customer scoring. While the software did provide successful marketing research and campaigns, BankFinancial decided to add a more dynamic tool that could pinpoint customers at risk and help the bank to stop customer defection.
"The marketing database is very vertical and there is not a lot of data within the system," explained William J. Connerty, assistant vice president, market research, BankFinancial. "Data mining is a time-consuming process and we needed a way to increase how we manage and access data. We wanted to better manage our churn rates."
BankFinancial found the solution that fit its needs from SPSS, as well. SPSS features analytical applications that analyze data and produce conclusions about current and future events. BankFinancial chose SPSS' PredictiveMarketing, a Web-based GUI application that integrates into an organization's campaign management, sales force automation or CRM system. For BankFinancial, PredictiveMarketing will sit on the marketing department's Microsoft SQL server and will pull data located in the centralized SQL database. The application's pre-built, customizable templates will help Connerty's two-person marketing team analyze customer data to anticipate which customers are likely to sever their relationship with the bank.
"The tool will pull all transaction data from a dedicated datamart then apply different churn predictors-including the number of checks specific customers have written, their total deposits, even withdrawal volume-to determine account usage patterns," Connerty said. "This monitoring will produce a list of likely churn candidates. We will segment these customers based on their originating branch, and send lists to bank managers. Now managers can connect with these customers via a telephone call. They can proactively intervene and re-establish the customer's relationship with the bank."
BankFinancial began customizing the PredictiveMarketing application in November and, at press time, expected to begin using the tool in July. At that time, Connerty expects the tool to help increase retention between two and three percent.
"The solution automates a lot of the data manipulation and data flow, and the pre-built models and easy-to-use interface will enable us to shorten the time it takes to develop predictive models," said Connerty.
The application has an $80,000 price tag, according to John Held, senior product manager for SPSS. The price increases based on number of users and the customer's existing hardware. BankFinancial is currently creating a metric that will measure its return on investment.
"The fact that there will be fewer customer touches and that the bank will gain a targeted pool of customers will keep messages more focused and less clouded. In the long term, targeted messages will positively impact the bottom line," noted Held.
BankFinancial regards PredictiveMarketing as the catalyst "that will eventually provide a 360-degree view of our customer base," Connerty related. "We chose to focus on retention first because it would provide the biggest bang in the shortest timeframe. Then we will begin to focus on other prediction models."
Predicting The Future
Next, the plan is to tap PredictiveMarketing to conduct cross-selling analysis. "We also plan to use the application to analyze customer lifecycle value, profitability, and likely-to-buy data models," Connerty added.
In addition to transaction data, BankFinancial also plans to merge other data sources into its centralized repository, including telephone survey results, demographic data, geographic information systems data and event data from its Web and call-center channels.
"We want to leverage more information to increase our predictive capabilities within the application," Connerty concluded. He did not reveal a specific time frame when BankFinancial would launch these data models or supplement data streams.
INSTITUTION: BankFinancial F.S.B. (Chicago)
ASSETS: $1.6 billion in total assets
BUSINESS CHALLENGE: Reduce rate of customer defections with a new tool that ties in campaign management, sales force automation and CRM.
SOLUTION: Chicago-based SPSS' PredictiveMarketing.
KEY QUOTE: "The solution automates a lot of the data manipulation and data flow, and the pre-built models and easy-to-use interface will enable us to shorten the time it takes to develop predictive models." -William J. Connerty, Assistant VP, Market Research BankFinancial