-
An almost overwhelming amount of customer data can now be mined and used to improve marketing and product design efforts. But not all data needs to be stored and analyzed.
June 1
Banks have access to vast new troves of data, such as social media feeds, that they have to learn how to use in their marketing and sales efforts or risk losing share to other institutions that are catching on faster.
One bank that's trying to get a head start, Georgia-based United Community Banks, believes the "big" data that really matters is customer transaction information. The bank teamed with several other banks to build algorithms based on pooled, anonymized customer data. Then it applied the new algorithms to its own customer databases. The anonymous customer data powers predictive analysis in a manner similar to credit scoring — other customers very similar to customer "x" behaved in a certain manner. As a result, the response rate for debit card marketing pitches increased to 23 percent from 14 percent, with average spending per card increasing to $483 per month from $233.
"That's a couple of hundred thousand dollars in revenue…Not only did we achieve more efficiency and cost effectiveness, we also improved customer satisfaction scores to more than 95 percent [of customers reporting a positive experience]," says Craig Metz, an executive vice president who oversees all marketing activities at United Community Bank. "The program works kind of like Amazon.com, where you know that similar people made certain purchases."
The Blairsville, Ga.-based United Community Bank, which has about $7 billion in assets, is one of about a dozen banks that are participating as subscribers in a one-year "proof of concept" project with Fiserv. The tech firm is accumulating data via an integration with the banks' core processing systems, covering a range of transactions and activities such as account processing, online banking, bill payment, and point of sale card transactions. This data is then run through an analytics engine internally developed by Fiserv to determine the best prospects for loan products or other financial services that the bank is pushing at a particular time.
"Most predictive models have been built around a limited data set, but if you look at the wide array of behavior and transaction data, there's a richness when you look at new sources," Metz says.
If the results that United Community Banks says it's enjoyed so far hold up, cooperation of this nature pays off.
