Community banks get on board with bots
Large banks have avidly adopted robotic process automation, technology that lets people set up small pieces of software called bots to handle simple tasks normally done by humans. It's something they don’t like to talk about it because it’s perceived as a job killer.
It's tougher for smaller banks to follow suit. They often lack large technology staffs to create and manage bots, and sometimes legacy systems don’t play well with RPA software.
Yet community banks — Carter Bank & Trust in Martinsville, Va., and Extraco Banks in Waco, Texas — have embraced RPA and are deploying it in a growing number of ways throughout their organizations.
Carter Bank & Trust
About three years ago, the $4 billion-asset Carter began a core conversion and a harsh reality set in.
“We had 40 years of, I would say, lack of data discipline,” Chief Information Officer Matt Speare said. “We saw we had massive amounts of corrections to do.”
Most banks experience data nightmares when they try to pour data from existing applications into a brand-new core system. Format, software and inputting glitches can mess up thousands or even millions of records.
A date-of-birth glitch from a previous core conversion, for instance, gave all customers the birth date of Jan. 1, 1900, setting the average customer age at close to 118.
Having humans manually fix these kinds of problems takes an inordinate amount of time.
“We needed a way to be able to automate that, which took us down the path of RPA,” Speare said. “RPA is great at being able to take an input and put it in one or more systems and do that as a repeatable process over and over and over again.”
Speare and his nine-person team then started finding other opportunities to automate similar tasks. For instance, the bank offers so-called positive pay, where business customers provide a list to the bank each day of the checks they send to it in order to prevent fraudulent ones from being honored. The bank compares the list against the actual checks received, flags any that do not match and notifies the company.
Some commercial clients have questions throughout the day — did the bank receive the file, did it process the file, and so on.
“If we were to handle that manually, we would have to have somebody watching this file transfer protocol server, grabbing the file, uploading it into the positive pay system, looking for any errors, sending an email back to that customer that says, got it, we processed it, then at the end of the day, doing a summary document,” Speare said.
RPA can handle all the steps like clockwork, he said.
“The more we work with the tools, the more ideas we come up with,” Speare said.
Today, RPA software from WorkFusion takes care of 211 processes throughout the bank. Speare estimates the bots save the bank $2.5 million to $3 million a year in labor costs. They also can easily work at night, which is often hard for humans to do.
For instance, the Federal Reserve sends the bank an email each evening with advance notice of what checks will be returned the next day. The RPA tool reads the emails and places holds on those funds so the customer doesn't accidentally or purposely overdraw on the account. Without the RPA, someone would have to work the night shift to take care of that.
RPA requires some training and a degree of tech savviness, Speare said. Three people at Carter Bank are dedicated to building bots, he said. A broader group of nine people, including Speare, decides what to automate and how.
That said, “we have enough enterprise-level visibility on it that the lines of business have had somebody coming to us with opportunities to automate because they definitely see the value,” he said.
No one at the bank has lost their job because of bots, Speare said.
“It does allows you to take some things off their plate or reallocate them for something else,” he said.
WorkFusion software runs on a server and provides management tools that let Speare and his team see and control what all the bots are doing.
The software can also blend bot and human actions. For instance, in anti-money-laundering work, bots can do certain kinds of steps, such as Google searches on new customers, overnight. In the morning, Carter's Bank Secrecy Act analysts can review the information the bots have come up with. Over time, with machine learning, the bots learn from the articles the humans accept and reject and become more accurate in the information they present to analysts.
At Extraco Banks, which has more than $1.5 billion in assets, 15 branches and 120,000 customers, a team of two people has begun deploying RPA software, starting with vetting the use of mobile check deposits.
“We know how much check fraud we're probably going to get every year,” said Cal Stevens, vice president and business process improvement manager at Extraco. “We’re always looking at newer technology, whether it be with our core or with an RPA tool, to help us manage the risk.”
The bank analyzes each customer and check according to several criteria and comes up with a score to determine which pass the test.
Then RPA software, which used to be called Foxtrot and is now called Nintex RPA, takes the score and uses predefined rules to determine how many checks a customer is allowed to deposit per day and per month.
“RPA does that for us so that we don't have a human being here at the bank doing that monotonous work, where they go into a portal and they look at all the customers who submitted a request for remote deposit capture and they have to qualify them and adjust their limits,” said Stevens. “And then of course that opens up the possibility of human error, when you've got a human going through one customer request after another and their eyes start to cross.”
The same bot can also determine whether someone should be allowed to deposit checks at the ATM. It will look at the customer’s qualifying score and how many debit cards are on the account, and turn that functionality on or off.
“Believe it or not," Stevens said, "since all this COVID stuff has popped up, one of the top 10 requests we get is customers calling saying: ‘I have a debit card and I really don't want to go through the drive-through and touch a deposit slip, touch a withdrawal slip, write a check out and have you touch it and send it back to me through the tube. I really just want to go to the ATM machine real quick and deposit a check, whether it be after hours or before hours of operation.’ And since not all of our customers have that ability, it's got to be requested by the customer and then we approve it. We get fewer phone calls in our call center because they're not having to call in and ask for that functionality.”
Another area where Stevens and his colleague are deploying RPA is in automatically detecting people who signed up for bill pay but haven’t used it in six months, and deactivating it. This will save the bank money, because it pays per account for the bill-pay service.
Stevens has been in the business process improvement department since last November. The bank originally learned about RPA technology from another bank that, like Extraco, uses core software from Jack Henry. The other bank was using Nintex RPA to automate processes across applications outside of the core.
“Every year we have a goal to save a certain amount of hard-dollar savings and soft-dollar savings,” Stevens said. This year he and his colleague have a long list of processes they want to automate using enterprise workflow and RPA.
One idea is to use bots to handle the forgiveness aspect of Paycheck Protection Program loans the bank made this summer. Another is to automate some of the bank’s regular reports through RPA.
“The RPA tool works with Excel really efficiently,” Stevens said. “There are a handful of reports that are pretty cumbersome that take quite a bit of time to develop.”
New uses for the technology will keep coming.
“It's going to be a long time before we run out of good ideas,” Stevens said.