August 9, 2017 by W.B. King
Aggregating data is one thing, but effectively analyzing it or predicting member behavior is quite another. And, many times well-intended executives make mistakes in this all-important pursuit.
So what are the biggest mistakes CUs make when it comes to data analytics? Credit Union
Journal queried a panel of experts to find out.
“Our primary methodology was too focused on transactional data from within the core at first.
The challenge is that, like many credit unions, some transactional data, such as credit card and
mortgage servicing, is housed elsewhere,” said Redwood Credit Union’s CIO Tony Hildesheim.
“Without all of your loan data, your reporting engine is not all that effective.”
Symitar Solutions Team Business Consultant Patty Moore added that a common mistake she
sees CU executives making when collecting data is “failing” to view performance measures in
context.
“In talking with credit unions who are interested in or pursuing predictive analytics, they have
more certainty about what they want to predict,” said Moore, “such as which loans have a high
probability of charging off [or] which members are at risk for closing, but they do not always
have an associated action plan to go along with the predictions.”
Asking the right questions
For the $1 billion Santa Rosa, Calif.-based Redwood CU, which supports 285,000 members at
19 branches, data gathering must answer at least one of the following five important questions,
explained Hildesheim. Does the analytics report:
1. Allow you to take an action?
2. Allow you to make a decision?
3. Does the data drive profitability?
4. Does the report drive staff engagement?
5. Most importantly, does the report drive member engagement?
“Our primary methodology is to focus on bringing the data together and establishing
normalization and definitions,” said Hildesheim. “The actual analytics is not all that difficult, once both the meaning and question are established.”
In many cases, credit unions executives make data-related mistakes by looking at data in
isolation, said Tim Keith, co-founder and chief strategy officer of Infusion, a Little Rock, Ark.based firm specializing in in data analysis for marketing campaigns.
“Credit unions need to seek partners with a proven track record who are willing to devote
qualitative time to the partnership,” said Keith. “The unfortunate reality is that there are many excellent vendors who focus solely on larger institutions and simply do not have a servicing model designed to work with the vast majority of credit unions.”
Keith further explained that most credit unions operate as two to three member bases within a
member base. These include single-service savings members, indirect loan members (with only
a nominal savings balance) and relationship members. Without data integrated across products,
he said, there is “no way” of distinguishing these groups, which is “fundamental to
understanding” the member base.
“Single service members typically make up 15 to 50 percent of an institution’s member base.
Without a household-level view that brings together accounts within a relationship, the institution has no way of distinguishing the member who has five different account types with the institution versus the member that only has savings,” he said. “Identifying single service members allows the institution to begin the process to assess how much potential those members have to bring additional business.”
Among Infusion’s clients is the $550 million Bartlett, Tenn.-based First South Financial Credit
Union. A recent data-gathering pain point centered on analyzing credit card data and signing
members up for cards during the onboarding process, explained Senior Vice President of
Marketing Delynn Byars.
“Marketing to indirect loan members has been a huge success,” said Byars, whose CU supports
approximately 56,000 members at 16 branches. “Many credit unions face a challenge with these
members because they don’t have the same level of awareness or attachment as those who
opened their accounts via other avenues.”
Byars explained that the CU worked with Infusion to develop an onboarding strategy, which
included an onboarding process with offers geared toward indirect members. At the beginning of
2016, she said, the CU’s product and service penetration was 2.89 for indirect members. By the
end of June 2017 that figure had increased to 3.35.
“Now that may not sound like a lot, but these members have basically no loyalty or affinity for
us,” said Byars. “They’re more like prospects.”
Data warehouse woes
One way to avoid data missteps is by understanding what data should be stored where,
explained Hildesheim. It doesn’t make sense, he said, to “copy data from one database to the
warehouse” if the analytics engine can access that data directly without performance of that
system.
“Transactional and financial data are two types of data that we determined needed to be stored
in the warehouse versus, for instance, web and online banking use that is directly accessible,”
said Hildesheim. “We are fortunate that our core provider [Symitar] also has a data solution
called Advanced Reporting for Credit Unions.”
Moore explained that Symitar’s ARCU has 315 pre-built standard reports. ARCU clients range
from $52 million in assets to $8 billion, with the average asset size for an ARCU client being
$1.1 billion.
“The larger-than $1 billion in assets credit unions are typically the ones with the in-house
resources to pursue predictive analytics on their own,” he said. “Trend reports include year
over-year, month-over-month, day-over-day comparisons so that anomalies can be detected,
patterns identified and data viewed in historical context along with goals and peer comparisons.”
Recently, Hildesheim employed ARCU to answer data-related questions, such as “Why do
members leave” and “How to detect signs of potential mobile fraud.” For the latter question, five critical elements were identified, including deposit velocity across the entire member base and geolocation data.
“Leadership needs to focus on making sure the right questions are being asked for the results
that they trying to achieve,” said Hildesheim. “Otherwise, you risk just ending up with a report
that doesn’t add value.”
Original Source: https://www.cujournal.com/news/how-credit-unions-are-avoiding-data-analytics-mistakes?brief=00000158-73f9-d502-a5fd-7bfbd3d10000