SALT LAKE CITY (4/6/16)--The age of data analytics has raised the bar for credit unions, whose members expect a celebrity-like experience, explained Peter Nohelty, chief technology officer, Royal CU, Eau Claire, Wis., during his presentation at the 2016 Credit Union National Association Data Analytics Roundtable.
Nohelty described a scenario in which a member is casually shopping for a car online during the lunch hour. Without the ability to offer that member a loan in real-time, the credit union is in danger of losing the loan--and the member--to another financial institution.
The expectation, implied or otherwise, is that providers of products and services use member characteristics and behaviors to craft offers detected in a real-time context.
“If you can’t run at the speed of the member, eventually you will fail,” Nohelty said.
Real-time, Nohelty said, is “something less than the attention span of the member.”
Accordingly, credit unions must adapt their architecture and systems to respond to data as they receive it, processing information in the present rather than the future.
“Most of our systems were built around transactional architecture, not data architecture,” Nohelty said.
Traditionally, credit unions processed data in batches, which provides a historical perspective. Real-time processing requires they add layers of stream processing, which offers context and personalization, and predictive analytics, which provides a future perspective.
Doing things in real-time often eliminates people from the process, he said.
“If you’re going to do things in milliseconds can you insert people in those processes? Probably not,” he said.
He called data analytics an era of machines learning to think and respond like humans. “You have to understand what humans do in those processes, what decisions they’re making and then be able to automate that,” he explained.
Credit unions can’t accomplish the transition to a data-driven architecture overnight, but should instead expect to embark on three- to five-year journey.
The first priority, Nohelty said, is to organize, format and time the collection of the data.
“The key is collecting the data,” he said. “You can spend the money and download the best tools, but if the data’s not there, it’s all for naught.”