You may also be interested in:


Failing Upward: Knowing When to Pull the Plug on Robotics

  • By Andrew Deichler
  • Published: 5/2/2018

SAN DIEGO – We’ve been hearing a lot about the benefits of robotic process automation (RPA), and for good reason. But what if you were trying to implement RPA for process improvement and found that it didn’t work? Should you disband the project, or keep pumping money and time into it to make it work?

Tuesday morning at NACHA’s Faster Payments 2018 conference, attendees received an inside view of an organization that attempted to implement RPA and ultimately had to let it go. Although the organization is a bank, many of the lessons learned can apply to corporate treasury and finance.


To be clear, it was not the intention of this session to scare people off of using RPA; far from it. Instead, the goal was to provide financial professionals with advice on what to do when they attempt to adopt this innovative technology.

Karen Buck, executive vice president, commercial, retail and payment operations for TD Bank, began by attempting to dispel the fears that many people have about RPA and artificial intelligence (AI). There is a concern among treasury and finance professionals that their jobs could be automated away. However, she argued that as RPA works its way further into the corporate and banking world, jobs are going to evolve.

“When you look at technology, there are more roles today than there were 10 to 15 years ago in banking,” she said. “So 10 to 15 years down the road, there will absolutely be more opportunities available—they’ll just be different.”


When you’re considering implementing RPA for process improvement, it is important that you’re not trying to automate a bad process. Obviously the goal is to make your process more efficient, but you also don’t want to use RPA to simply replicate a process that doesn’t work. Buck recommends following a four-step pre-implementation method.

  1. Lean out the process. Step back and ask yourself if there are steps in the process that no longer make sense.
  2. Look for opportunities to break down process silos. It’s important that the process runs as smoothly as possible and to do that, all components of it must work together.
  3. Ensure that you have standardized processes across the organization. As more people get involved, ideas may conflict. It’s important to make sure everyone is following a standard when it comes to RPA implementation.
  4. Measure the metrics. You want to make sure you have a benchmark, so that once that process is automated, you can demonstrate the value you’re bring to your organization.


TD Bank saw an opportunity to improve its wire payment operations. TD was having an issue of wires that weren’t processing straight through, which then required an employee to go into the system and fix the problem. “When you’re doing thousands of wire a day, even if you have single digit manual intervention, that’s a lot of people and a lot of processes that involve fixing those exceptions,” she said. “So as we were looking at this opportunity, we said, ‘We have a fair number of wires that are dropping for manual intervention.’ We brought in our process improvement team, and said, ‘Help us understand why these wires are dropping and what we can do to leverage RPA to find some improvements.’”

The group identified a number of areas where RPA could make an improvement. The most glaring one it found was employees cutting and pasting information from one system to another, allowing the wire to process straight through. “This was not a great employee experience to have to cut and paste data, and not a great customer experience because the wire is dropping for manual intervention,” she said. “Perhaps it only takes a couple of minutes to fix, but imagine the experience of a customer sitting at a closing table.”

For RPA to be successful, it is imperative to engage all of the right parties at the right time. Buck and her team were quick to involve their technology and risk partners, because it didn’t want to start an initiative that would create more risk for the organizations. “We partnered with the right people; we have a number of vendors with whom we partner across North America, and we chose the partners that we felt made sense to help us on this journey to implement our solution,” she said. “And we communicated very effectively to our business partners; we gave demonstrations of what the bots could do.” 


Ultimately, TD Bank determined that RPA wasn’t appropriate for improving wire payments. Therefore, it proved critical that Buck’s team was ready to “fail fast” from the start of the project. Her team quickly recognized that RPA technology didn’t play well with TD’s existing systems, and so they let the initiative go.

“We could have spent a lot of time working to fix that, but we have a whole pipeline of initiatives that we want to get done,” she said. Is it really worth the effort to spend lots of time trying to fix this, when we could shelf it for the time being and focus on another initiative? So we opted to redeploy those resources and focus on ongoing initiatives.”

So while the project ultimately failed, Buck still views the outcome as positive. “What we saw is that as you put the right governance in place, as you learn, as you work with partners, there are a lot of things you can do to implement robotics as yet another tool in your toolbox to drive efficiency and effectiveness and improve employee and customer experience,” she said.

Don’t miss the educational session, How to Successfully Incorporate Robotics into Your Treasury, at AFP 2018.

Optimize your cash flow.

Our treasury management and payment services can help you manage cash flow, control risk and become more efficient. Member FDIC.

Let’s get started.

Copyright © 2021 Association for Financial Professionals, Inc.
All rights reserved.