Find a Professional

Find a Professional

Case Study: Regional Bank


Regional Bank: Customer Analysis, Predicting Customer Behavior, Loyalty Program Design

An interview with Stefan Boedeker

Every business needs to acquire more customers than it loses. While market research can generally sum up reasons for customer (dis)loyalty, careful econometric analysis of data around customer departures and arrivals can help identify what customers really value. Mr. Boedeker discusses his previous experience with putting companies on track to positive customer growth and retention.

What was the opportunity or challenge? 

I was retained by a regional bank that is part of a national holding company. It had experienced a drop in the ratio of new acquisitions over attrition and was losing more customers than it was able to acquire. The new executive vice president of marketing approached about whether I could advise on the situation from a proof-of-consumer-behavior standpoint, leveraging a wealth of data to which they had access.

What did your work involve?

Two separate approaches were developed. One focused on the attrition situation and the other on the acquisition opportunity. In the final stage of the project, there was a marriage of those two approaches. 

For the attrition, econometric models were used to process a 48-month dataset that tracked customer behavior in terms of deposits and other activities of past clients. A warning system was used to model the specifics of clients that left, and that was compared to the base of clients that had stayed. Current clients that showed the same statistical properties as clients that had left were flagged so the bank could anticipate the situation and enter client-retention mode.

This was aided by a second model: the next-best-product model. I looked at existing customers that had up to three products from the bank and measured what would be profitable for the bank to offer as the next product. For clients that may have been in danger of leaving, I worked with the marketing department to individualize their approaches to identify potential next-best products.

I utilized the bank's data and acquired outside vendor data from Claritas and Acxiom. These sources contain data about consumer preferences, socioeconomic and demographic properties, and characteristics of households at a detailed zip-code level. These data were matched into existing customer address information, which enabled the bank to send out more targeted mailers, product offers, etc. 

So, rather than stopping at just identifying the leading causes of client attrition, I developed a second dimension in the shape of a customer-acquisition model.

What were the results? 

Lasting value took many forms. First, routines were written to extract the right data from the data warehouse and the corporate holding company. These routines were updated on a quarterly basis.

Most measurable was the change in the customer-acquisition ratio. When the work started, the ratio of new acquisitions over attrition was approximately 0.95: for every 100 accounts the bank lost, it only acquired 95. After 18 months, the score moved up to 1.03—customer acquisition was then outpacing attrition. It took about 12 to 18 months to implement the model and run a few marketing campaigns to see measurable improvement.