TAMPA, Fla.-The ever-growing file of aggregated, disparate information known as "big data" is leaving many credit unions with more questions than answers.
A new report, however, shines some light on how CUs can use predicative analysis to increase revenues.
"Big data is something we have recently begun to focus on but we are new to it," said Robert Keats, vice president of information technology for Grow Financial FCU. "I believe there probably have been some missed opportunities in the past, but if we had all the data in front of us, and understood it, better decisions could be made."
Grow Financial FCU is not alone in this quandary. In June, Filene Research Institute released the report, "Big Data and Credit Unions: Machine Learning in Member Transactions," authored by Philipp Kallerhoff, PhD. The report anonymously profiled information and account records from five CUs in the U.S. and Canada.
Kallerhoff explained that the data encompassed some 500,000 credit union members and 250 million transactions over a five-year time frame. Among variable correlations studied were transaction amounts, account balances, credit scores, income and gender.
For example, the "cluster analysis" applied predicted the next best product with 30% accuracy. He added that the best predictors are the balance-to-income ratio, expense-to-income ratio and the balance of loans to savings.
"The financial products, combined with transactional data can give a good picture of future needs of the customer and also a good estimation of future delinquencies," said Kallerhoff. "We've already proven that big data, if used and interpreted properly, can cut costs while rising revenues-sometimes dramatically."
Taking Proactive Measures
Grow Financial Federal Credit Union is taking proactive measures with the goal of better understanding the collected data from its approximate 160,000 members. "We have just established a data warehouse, which is a new department with four employees dedicated to learning how to best use our data," said Keats.
Along with the new department, Keats said the CU's 2014 budget will include funds for hiring "big data" consultants. "We hope to start new data-based initiatives in 2014. Our first goal is to understand the data and the consultants will help quicken the pace. Then, over the long terms we will focus on revenue generation and member satisfaction."
Since there is a seemingly never-ending stream of data available to credit unions, it's often difficult to determine where to begin the data mining process. Kallerhoff said it's best to look at personal information, transactional data and then the information about the products that the member is using.
While an important first step, this initial process is considered superficial. As such, a deeper dive is required such as understanding how members use the products, both in the past and presently.
"The combination really is a wealth of data that remains mainly unused in the credit unions. The main reason is that the data is very large and you need specialized personnel or expensive software to use the data," said Kallerhoff. "Therefore we designed a research project with Filene that was very lean and also hands-on for the credit union to use the results."
He noted that the five participating CUs in the report wanted different insights gleaned from big data-from a better understanding on how members cycle through different products at different stages of their membership to transactional data to improve underwriting.
"From these results and some very simple additions to their current infrastructure the credit unions were able to improve their credit scoring by 15%, increase number of loans by 1% and recommend products 10% more successfully," said Kallerhoff.
Kallerhoff, along with Eli Mohamad, co-founded Walkmore and announced an open registration pilot project that will combine big data analytics and predictive powers to increase revenues and open new market opportunities.
Where The Benefit Is
"The participating credit unions will benefit from insights based on their members' data and will be able to attract new and reward current customers through a novel health-based approach whereby customers improve health from a baseline and are rewarded with increasingly better financial products," said Mohamed. He added that pilot participants are provided with web, iPhone and Android-based tools.
Kallerhoff explained that Walkmore is using models that harvest a vast field of data to gain quick, actionable insights producing Amazon-like recommendations for CUs and their members. As the data analytics industry becomes refined, even smaller CUs that are without dedicated data mining resources can now benefit from member information collected.
"You do need a certain level of resources in order to internally gather and analyze big sets of data," said Mohamad. "Smaller credit unions with limited budgets could benefit from asking their CUSOs or local/regional credit union associations to do pool-in resources to do this."
For Keats, looking to 2014 is an exciting proposition complete with a few big data-based campaigns. "To start, we are expecting small, measurable results. Once we have pulled more data in and are able to analyze it, we can start to ask the right business questions of our credit union. There is no question that big data is a hot-button issue."
