BROOKFIELD, Wis.-The ability to decipher member information into actionable intelligence is a goal many credit unions executives would like to achieve. This all-important concept was covered, for example, during this year's NAFCU annual conference.
"Business analytics brings together the best of data management, analytic methods and the presentation of results-all in a closed-loop cycle for continuous learning and improvement," noted presenter David Wallace, Global Financial Services Marketing manager for the SAS Institute.
With the goal of enhancing marketing and performance management strategies-taking business analytics one step further-one financial services technology firm is offering a predictive analytics solution it says will help turn member data into actionable intelligence
In early 2010, Fiserv's CEO had a "wild idea" to bring together data for clients in a predictive model at a low cost, said Don Hopper, the Brookfield, Wis.-based company's director of predictive analytics. "This is compared to hiring a data scientist or purchasing expensive statistical software."
By 2011, Fiserv built on its proof of concept and began a beta program that included in market tests for various channels such as auto loans. This service was then free of charge for early adopters testing the solution. "We wanted to solve the puzzle and provide meaningful insight that would offer members of credit unions a solution-a best next action," said Hopper.
Predicting Sales
Last year, the beta test came to an end with the launch of Predictive Scores. The platform harnesses multi-channel transaction data from account processing, debit and credit card processing, online and mobile banking and electronic bill pay systems. Hopper said this represents billions of transactions and millions of banking behavior observations.
"We have seen a ton of interest across the board. During the first year of beta testing, we presented Predictive Scores at our client conference. There were maybe 10 or 20 people in the room kind of scratching their heads as we talked about big data," said Hopper. "During our conference this year, the room was packed for the same presentation."
To date, Fiserv has 34 financial institutions using Predictive Scores, including nine credit unions. Since the solution is new and results are still being aggregated, Hopper said he could only share results from a bank client.
The $2.4 billion-asset retail bank, with 55,000 customers, realized a 244% increase in funded auto loans, according to Hopper. This was accomplished by a customized direct mail campaign to individuals that Predictive Scores determined were likely to purchase autos in the near future. "This increased applications by three fold and the campaign netted the bank a 200% return on investment," said Hopper,
To realize these types of high return rates, a deeper dive is required. To this end, Predictive Scores used client automatic bill pay histories accessed through the bank's core system and was able to pinpoint those customers who had recently paid off auto loans or would be making final payments in the near future. With this information, the campaign was born.
While Hopper said the solution is scalable, with one credit union client having 13,000 members, he noted that the majority of clients using the system have roughly $500 million, or less, in assets. The service is offered as a 12- or 24-month contract. While costs can vary depending on the complexity of the organization and determined goals, Hopper said credit unions should expect to pay roughly one dollar per member for the service.
Tapping Into Member Desires
The solution not only helps in predicting sales, it also taps into member wants and needs. One credit union beta tester, for example, determined that it wasn't prudent to continue to invest large sums of money in its digital channel because a large percentage of the member base preferred banking at branches or ATMs, Hopper explained.
The analytic process is designed to "minimize fatigue" and is based on an algorithm that is refreshed at the end of each 30-day campaign. "The lengthy piece is the first data pull, which can take one week to one month," said Hopper. "Once that is in place, the marketing department is brought on board and we have phone and webinar training. It's a simple approach, like the Amazon.com model."
