NEW ORLEANS—Facebook “knows” of at least three-million couples currently engaged to be married. Candian Tire “knows” people who buy Mobil1 oil are significantly better credit risks than those that bought generic motor oil. Why don’t credit unions “know” such information?
The answer is the former two companies are better at Big Data, according to one person, who suggested credit unions have much of the same actionable information, but just don’t “know” it.
Speaking to the CUNA CFO Council meeting here, Bill Goedken, president/CEO of Omaha, Neb.-based idea5, said one part of the challenge every CU must address is the “Big” part of Big Data.
Bill Goedken speaking to CUNA CFO Council meeting in New Orleans.
“Ninety-eight percent of the big data out there you can ignore,” he said. “It’s the 2% that’s the gold. A skateboard and the space shuttle are both methods of transportation, which is why it’s critical to be able to sort through data.”
Just what is Big Data? According to Goedken, the term refers to extremely large data sets that may be analyzed computationally to reveal patterns, trends and association, especially relating to human behavior and interactions. That can include the weather, phone use, Twitter feeds, card purchases, ATM statistics, etc. He noted that Big data can be numerical or text or documents, can be public or internal, can be structured or unstructured, and is is usually stored in the cloud.
Marriage Proposal
“The question is how can you marry the outside information with the internal information you have,” he said. “Banks and credit unions have massive amounts of internal data; getting to it is a task, but finding what’s right is pure gold.”
The most common ways credit unions and FIs use Big Data include:
- Fraud detection
- Trends in delivery channels
- Marketing and CRM trends
- Comparing to peer groups
- Predictive modeling in the near future
“The next step is to go beyond the common uses and combine it with external data,” said Goedken.
Goedken offered a couple of case studies for how Big Data has been used by FIs.
Case Study #1: Internal Pattern Discovery.
Goedken shared the story of one bank that found there was a “major blip” in its home banking account activity around 2 a.m. every night. The bank finally deduced the blip stemmed from customers who couldn’t sleep, worried over whether a check had cleared and they were going to be overdrawn. In response, it target-marketed a specific web page and instant message ads about overdraft protection during the time in question. The result was 400-plus sign-ups for overdraft protection in a 90-day span. Fee income was up by $28,000 in the first year. (The information was discovered by a Gen X employee as a tangent to another project.
Case Study #2: External to Internal
Goedken shared the story of one credit union management team that was being pressured by the board to open a brick-and-mortar branch in a certain community, while NOT closing any branches. He said the management team had a “few” statistics to back up their claim that more resources should be devoted to electronic delivery channels AND to support a decision close the one “dead-weight” branch. The management team gathered internal stats (Big Data) on delivery channel changes, and using national statistics and external data to talk to the board about delivery channel changes. The result: the credit union is opening a much smaller branch while also closing one in 2015, with an annual savings projected to be $300,000 in the first year alone, and $1 million over three years. Part of the savings is going to electronic channels.
Case Study #3. Internal to External Discovery
Goedken shared that one Tampa Bay CU used local market unemployment data and overlaid its own delinquency data. Not surprisingly, the two largely tracked each other. A loan demand chart, also not surprisingly, was the reverse. But what both data sets had in common was they were lagging the unemployment rate. Watching the data the credit union could see a decline in unemployment, and was able to work with both the lending and marketing functions to let them know they had approximately three months to gear up for new demand so they could be prepared.
“Using unemployment forecasts, can also project out loan demand in the future,” added Goedken.
But with all that information, he reminded his audience of CFOs to “remember that 95% of external Big Data is basically useless to the credit union and does not change your strategic thinking. Another 3% may be useful, but it is very difficult to get, is expensive, and difficult to make sense of. So concentrate on the remaining 2% and ask yourself the top 10 questions you are trying to answer. Involve your management team, ask what keeps you everyone up at night, and remember it’s often the questions you don’t ask that may get you into trouble.”
As an example of understanding Big Data, Goedken said research done by idea5 has shown that the branch is declining in popularity among all age groups, but it remains a very solid #2. “Most respondents (53%) indicate they will NOT be doing banking with any FI without some type of a physical branch purpose,” he said of the surveys his company has conducted.
The number-one method of interaction, of course, has become PC/Internet/home banking, while mobile banking has, not surprisingly, been adopted most by the 18-34-year-old age group. Mail and telephone use, on the other hand, has risen among people 55 and older.
“When it comes to delivery channels, remote banking is taking over, but most consumers still want a physical branch network of some type,” he said. “A very large discussion/strategy rethink is happening around the country. But I can tell you closing a branch is a very traumatic.”
Idea5 has also done quite a bit of analysis around Internet/online banking.
“You may think it’s crazy, but your web page says a lot about you,” said Goedken.
He said his company’s research has found that the average FI website now has twice as many “visits” as a branch. He asked CFOs if they were tracking visits to their own CUs site (few had the information), including more detailed Big Data, such as who was visiting and how long they stayed.
“We discovered that credit unions are better at websites than most community banks and it’s something I don’t’ think we’re taking advantage of,” he said.
Over a six-year period idea5 looked at 500+websites, mostly Midwest institutions. In 2009, 63% were found to contain errors on the “flow,” with pages “under construction” or sending visitors to the wrong screen, etc.
“That’s an absolute no-no, yet it’s almost two-thirds. By 2013 that had dropped to 35%,” he said.
In 2013 idea5 looked at the same websites and found a number of improvements, he said. But Goedken still has a pet peeve.
“A big bugaboo of mine, and banks are more terrible at it than credit unions, is the percentage that had no indication of community involvement: 58%,” he said. “There is a pretty good pocket of people who will choose you over any other FI based on what you do in the community. Giving back to the community, having your people involved, is important.”
In 2014, idea5 conducted another study of more than 200 websites to calculate their usage and effectiveness using an internally developed metric that involved traffic, time on the site, pages that are viewed, etc.
“What we wanted to do was say, ‘How much do your members use your site, are they in it for long?’ Think about it from a retail standpoint. If I’m in Dillard’s or Macy’s and I’m the store manager, what do I want? I want people in the store longer; chances are higher they’re going to buy something. The same is true of your website. But there’s a caveat: if there in there a long time it may indicate your website absolutely sucks, so you need to take a look at it.”
Goedken said that nine of the 10 websites that were “power-rated” by his companies were credit unions, and of the top 10 fie were located around major universities, including University of Wisconsin CU and University FCU in Austin, Texas.