Challenge
A major bank wanted to increase response and purchase rates for direct mail cross-sell/up-sell campaigns to existing customers.
Actions
An advanced tree-based algorithm -employing information theory and fuzzy logic- was developed to analyze existing customer data and build a predictive model that rank ordered the bank's customer list according to the likelihood to respond to the mailing. Starting at the top of the list, the bank selected a predetermined number of customers to whom mailings were sent.
Result
Use of the predictive model tripled response rates and brought in more than $9 million of new business with one mailing campaign.
