RPA in Banking Case Study: How Carter Bank & Trust Achieved 3,800% ROI with RPA

A complete transcript of the webinar is below.

Paul Necklen: 00:01: Okay. Good morning, everyone. This is Paul Necklen with HelpSystems. We're going to get started. Today's webinar will be an RPA Success Story: How Carter Bank & Trust Achieved 3,800% Return on Investment with RPA. This webinar is being recorded, and you will be sent a link by email after. So, you don't have to furiously take notes. We're going to provide this to you in a link, and as you know, it is now being recorded.

Let's start with some introductions. Today's presenters, again, I'm Paul Necklen, the Director of RPA Technology at HelpSystems. We are honored to have Matt Speare, Chief Information Officer of Carter Bank & Trust doing the presentation today. Matt Speare has been the Chief Information Officer at Carter Bank & Trust since 2017. As CIO, Matt provides executive oversight for the bank's technology infrastructure and assures that the technology platforms effectively support the bank's strategic objectives. In addition, he assures that the bank's information risk management practices are consistent with industry best practice.

Matt, welcome, and again, I just want to, on behalf of HelpSystems, give you a very sincere thanks for your time today and presenting on this webinar.

Matt Speare: 01:25: Pleasure to do so, Paul.

Paul Necklen: 01:27: Thanks, Matt. We are going to cover our broad solution in just a couple of slides, and then we're going to have some interactivity with the audience. We've got three polling questions we're going to intersperse throughout the presentation to give you an opportunity to interact with us, as well as we can get a pulse of where you're at in terms of your RPA journey.

For those on the line that aren't familiar with HelpSystems, we offer a broad solution set of IT software. We have a tagline here, called We Make IT Better. We do that through three key areas: Automation, security, and information. Notice we said we make IT better. We don't make IT best, because best almost insinuates that there is an endgame that you can actually finish. We believe this is a journey, and we clearly are going to start with the automation in the upper left here, the robotic process automation, today.

But as you can see from the list of solutions we offer, very comprehensive on the automation side of our business, and then complimentary to that, we also offer a full suite of security products. As we know, specifically with banking, Matt can attest to this, being in a highly regulated industry, security is paramount and being able to demonstrate that, as well as do things like penetration testing and identity access management, et cetera. We'd be happy to entertain your organization for these areas, as well, but today, we're clearly going to focus on robotic process automation.

Then at the bottom, of course, information, we provide the ability to take all of HelpSystems' products, as well as third party applications, and combine that into a single pane of glass, and provide executive dashboards, and reporting, and analytics.

Today, we're going to talk about a product called Automate. Automate is a robotic process automation platform, designed to automate repetitive and manual processes across your organization. Matt is going to key in on three areas that you see bulleted there. Automate can handle everything from repetitive tasks to highly complex, mission-critical business operations, and he's going to talk about that. It also centralizes your automation efforts and does key integration of business applications, and obviously in banking, that's centered around your core system. Then last, of course, as we talked about, improving security and staying in compliant with your industry standards.

We're going to move ahead here with Matt through the presentation. First, we're going to send out a polling question. Now we're going to interact with you and ask a favor for your participation, so we can kind of get some engagement with you and Matt, so he gets a sense of where you're at in your RPA journey.

Let's start the polling question. Here we go. Question number one, I'm going to launch that. Go ahead. We really encourage participation, if possible, to make your selection. These results should pop in, and clearly, just getting started by researching RPA is number one. Number two, evaluating RPA solutions and vendors. Three, planning your RPA strategy and testing solutions, or we already have an RPA solution in place.

We're going to give that just a minute here to percolate, let everybody participate, and those that have already made your selection, thank you. This is very helpful, and I think once we post the results here, it'll be very interesting. We're going to just give that another maybe 30 seconds here if we've got some stragglers here that want to participate. There's still some results coming in. Just a brief pause here. Another four or five seconds. We've still got results coming in. I'm going to actually wait here. This is very, very good. Looks like the responses have slowed, so let's go ahead and close the poll.

Okay. Interestingly, we see that 33% of the responders said they are just getting started researching RPA. That is a third of the respondents. Coming in number one, though, which we find interesting is, over half, 52% of you had said that you already have an RPA solution in place. So, Matt, that's kind of interesting. Half the audience-

Matt Speare: 06:29: That's great.

Paul Necklen: 06:30: Yeah. They seem to be like you. They've got a solution in place, and they're actually practicing automation. At third place, we've got 9% showing that they're planning their RPA strategy and testing a solution. They're kind of early in their journey, and I think you can probably talk about some of that, what your process was like, Matt, through your presentation. Then bringing up the end here, 6% said that they're evaluating. 6% of the audience is actually looking to evaluate a solution and vendor for RPA. So, thank you. That was very interesting. Thank you for your participation, everybody.

At this point, I'm going to turn it over to you, Matt. It's all yours.

Matt Speare: 07:14: Great. Well, thanks very much. Why don't we go into the next slide? And appreciate everybody's input. It let me understand about where you're at. The reason that we went down this path in RPA is that it really came down to, it was somewhat a little bit of a crisis for us. We had made the decision in early 2017 that the banking systems that we were using were out of date, and it was that the length of time it would take to put in a new product or a service was just too lengthy. And so, we decided to kind of do a whole scale rip and replace across all of our banking systems and go to a new package of applications.

With that, unfortunately, what we have found prior to my joining is that there had not been much of a data discipline in the organization over time. And so, what we had was basically 40 years of really not much quality control in the world of our customer's data. I'll just use this as an example. If you believe the date of birth for our customers, the average age of our customers would've been about 118-years-old.

When it was time to get through and kind of get the data correct as part of the underlying core conversion, we realized we had a big problem and that we needed to fix that. So, with that, we realized there was going to have to be a lot of data correction that would occur. However, we were going to have to do most of that manually. And so, we realized that, one, it would not be feasible to do that in certain instances, such as incorrect data birth. Well, we had 44,000 customer records that were incorrect. We needed to come up with a way to be able to take the data and then get it into the core at that time so that when it was time for extraction, that the data would be correct, versus trying to deal with—Whenever you go through one of these conversions, the next six months afterwards, you are doing nothing but firefighting just to get used to working with the new core. Ours was a lot broader in that we saw that we had hundreds of different data elements and differing volumes that needed to be corrected along the way. I'm going to use an example of one of those a little bit later, as well as what we were also doing is, we were changing over from being a mainly on-premise organization in terms of where those applications resided, and we're going to be in a hosted environment. When you do that hosted environment, you do not want something to be broken in that hosted environment where a function within, I'll say like your online banking platform, that isn't working. The only way that you know about it is that you're getting calls from customers. That's not the way to be proactive around it.

We needed to build out the ability to be able to not only monitor performance, but also monitor functions, because they all tied back to service level agreements and get money back if they didn't hit those. Unfortunately, the reporting that you get from the providers in this space, they'll always tell you, "Everything is running along fine," unless they have some kind of system-wide outage, versus, "Hey, check processing is not working and has been out for four hours." You would have no visibility on that, and so you could be much faster to opening tickets, getting the items remediated by active SLA monitoring.

And so, the decision criteria for us and what we were doing was that, one, we needed a platform that really had robust capabilities, allowed us to do a lot of different things within one platform, rather than having to do several, and because we were under somewhat of a time crunch, I couldn't afford to need to take six months to get people trained up on how to do something properly, and most importantly for us, that we wanted to get immediate return on investment for doing it because we had so many things that needed to be done.

If you go to the next slide, Paul, that when you kind of look at it for us as a whole that, what were the key things that we were doing? One, we have these mass data corrections that we have to do, and we still do a lot of them today. I'll talk about a couple of those in a minute. I just, literally, I did not have the manpower or the space to do it. One, it was not just a cost play, but literally that I would never be able to get 40 people in and teach them how to use this system, and then having them working around the clock to do data corrections. Let's face it. Whenever you have humans involved in doing a repetitive task that requires getting data from one spot to another, that they are going to make mistakes.

And so, one of the key criteria for us, as well, is ensuring that we had a high level of accuracy, so that we didn't have to go back, and in the case of having to correct 44,000 customer data birth records, that, well, okay, first pass, you get half of them right. Then you have to go back in. Now you get a little bit better. You get 70% of the remaining 50% right, and then it just kind of goes on and on for a long period of time. So, what this allowed us to do was be able to go through and get a much higher accuracy rate. We run about 99.3%, overall. Then the rework ended up being much, much less. That was one of the key things that we wanted to accomplish, as well. So, more or as close as you can get to one it and done in having that completed.

The next slide, if you wouldn't mind, on it. So, what kind of benefits have we seen on it? We have been with HelpSystems and done this now for just a little over two years. What we do is we run a rolling kind of 12 month lookback. You'll see it in one of the upcoming screens. But in the last 12 months, as an example, we have saved 110,000 man hours. This is pretty simple to do and calculate in that, how long does it take the average human to do it 10 times? We always use kind of 10 times as the... That is the baseline for how much it takes them to do it consistently every time.

What we have seen in it is that it's a basic industrial engineering 101 time study. Give somebody the task to do 10 times. There's a stopwatch for when they start. Stop it when they get through the 10th one. Then we would also, as part of that analytics, take a look and see, okay, so if you give it to them 10 times, how many times can they do it correctly? What it allowed for us to do is take it from an 80% accuracy rate up to that 99% accuracy rate. We have had about an 85% input error reduction, which was important for us.

Then with that SLA monitoring, I can tell you that because of our speed of being able to know when a specific function with an application is not working correctly, that we've been able to the main time to repair has dropped by about 80% on our average component average. And so, it's been hundreds and hundreds of hours of downtime saved on it, overall.

Of course, it comes down to, are you happy with it? I think we've proven out that, one, the cost benefit is there, as well as we continue every day to find more and more opportunities to leverage the tool, to do tasks for us, and I'll use another one as an example.

One of the many things that auditors and regulators love for you to do is that you're doing periodic access reviews on your applications. Well, HelpSystems has made that very easy for us, and we're able to do it much more frequently. We've got a grouping of about 10 applications that we consider to be related to Sarbanes–Oxley or our critical applications. What we're able to do is, every Sunday, the RPA task goes in and extracts the user information, compares that to a daily file that is pulled from our HR system, and basically marries that up, saying, "Hey, the 400 users are in this system, and three of them do not exist in the HR system. So, we need to do something and make the determination on, do they stay? Or, hey, they're really not here. We take them out."

What we have found with that is, that would be something you do every couple of months, and it's scrub spreadsheets back and forth. We have automated that entire process so that it is able to occur on a weekly basis, and we're able to keep access much cleaner in those systems, overall. It's another good example of how we continually find new things to do with it.

Paul, if you want to go to the next slide. I do believe we have another poll.

Paul Necklen: 17:38: We do. Thank you, Matt. Let's bring that up. I'll just launch the poll. Again, we're going to ask some interaction from the audience, if you could be so kind. Here, we're going to look at a few selections around, what is your organization looking to solve with RPA? Go ahead and make your selections. You can read the questions there. I see the answers percolating in, so that's good. We're moving along here. Give it just a few more seconds here, Matt, if that's okay.

Matt Speare: 18:14: Absolutely.

Paul Necklen (18:16): Well, it's kind of like a horse race. You get to look and see. 

Matt Speare (18:19): It is fun.

Paul Necklen (18:21): but there's a clear leader here. We've got a clear leader, a close second, third, and then kind of distant as you kind of get into fourth and fifth. Looks like we've... Oh, they're still percolating. I'm going to give some time here. All right. There. Looks like we've stabilized, so I'm going to close that, and then I'm going to share that on the screen so everybody can kind of see the results.

Look at our first place winner there, Matt. Just what I think you highlighted was that efficiency and saving time. That's what we would expect, but that's a huge first place, that 94%. Second, obviously looking for that error reduction, which you, again, highlighted was critical in your business. That was one of your major drivers, so that's great to see that that aligns with the audience.

Third place, cost savings, which I think you talked about, labor savings, so that's good that they're in tune with that, and then we got a tie. At the end of the rainbow here, we've got two 19% responses, so, both, compliance and staying competitive, which I think that compliance is, as you said, very important in your highly-regulated industry. Let's go ahead and move to the next slide. You're up.

Matt Speare: 19:49: That's right. All right. I wanted to point out what you kind of look at in terms of... I just thought it would be good because a lot of people have, over the years, asked me about, "Well, how do you track what's the impact of doing RPA task?" This is just a screenshot of something that we use. Basically, it's an Excel spreadsheet. Every row constitutes a different bot task. Then what we do is, that when every time the bot would run or the RPA task run, there in the number of runs column, it would increment that by one.

And so, by building it in, it can calculate out in terms of what's the cost benefit of that particular RPA task? It really comes down. It's a numbers game. If it takes somebody 10 minutes to do something, and I'll just use that correcting the CIF validation where we ran it seven times. It takes about 20 minutes for somebody to do it, and then the number of total items, about 4,500. What we've calculated out, we're very conservative in our rates and saying that if I had to hire a clerk to do this, what would be the fully loaded rate for that? We haven't revamped this since last year, but know it's more than that, but we just say, "Look, if we had to hire a clerk to do it, then we'd be paying about $23.17 an hour for them to work and do these tasks." And so, it calculates it all the way across. This is a screenshot from a couple weeks ago, but we're up to about $2.6 million in savings, and about 110,000 man hours have been saved.

If you go down to the next slide, Paul, one of the things that you do have to consider when you're doing this is how you do RPA operations. This is one of the mechanisms that we use. We basically have a dashboard where we list out every single task and to which bot do we leverage it to do that. That's kind of important in segregating out your RPA test so that you don't have the same bot agent attempting to do two things at the same time, especially if it has to do something where it needs to pop something up on the screen, because they will lose focus. They will step on each other on those tasks. And so, we use it to keep the master overview of what is occurring, at what bot, at what time, what day.

For us, we happen to use five serve DNA for our core. A lot of things occur in that core, and so we want to add a date as to whether or not it uses that, because it also signals to us whether or not we need to create a separate ID because whenever an RPA tool interacts with other applications, it also has to have an ID. You try to limit what that ID has the authority to do down to, what is the task it's trying to accomplish? And so, this is kind of our master overview.

If you take a look at the next slide there, Paul, that we also have within the HelpSystems' tool is our master calendar. We know what is occurring with each bot all day long, and you can see there are certain times of the day where we have lots of things going on, generally at the top of certain hours that you'll have a bunch of things that are going on at the same time, and it helps you when you think through, okay, what is my opportunity to schedule so that I can de-conflict things from occurring?

If you go to the next slide, what do we like about Automate and enjoy for this that, one, the first two kind of go together, intuitive interface. Let's put it this way. I know that for years, everybody thought about putting these tools into the hands of an operations clerk, and somehow they'd be able to program that. I would say that is not a great idea. You don't have to be a programmer to be able to make these tools work effectively because you don't have to custom code things into Java.

However, you do need to have a programmer's logic because there's lots of prebuilt capabilities in it, and they'll get to about 90% of the way there, but what you'll actually do is spend, once you have built out the workflow on what you want the RPA task to do, then you certainly are going to want to go back and fine tune it. Then as you start putting volume to it, you're going to find issues that you did not anticipate while you were building it, and then you're going to need to go back and figure out how to do other error corrections in there, as well.

Of course, we love it that when you think about, what are the triggers that you're going to use to kick off an RPA task? It is basically going to be something along the lines like a schedule that every 15 minutes, I want it to go and check this email inbox to see if a certain email with a certain type of subject line comes through, and then I know, at that point, that I'm going to need to go ahead and kick off a task, or in the case, a lot of times, when we're doing mass data corrections, we'll want to run through a small subset once, and then we'll actually put it up on the big screen as an ad hoc task just so that folks can watch it and make sure that everything is according to the manner that we think it should.

To go to the next slide, all right, I know sometimes it's hard to visualize what's going on with your RPA task, but I thought this was a good way to do it. I'll explain what this task is doing. Basically, it is a spreadsheet on the left-hand side that has account numbers and then different fields that need to be corrected into the corset, which is on the far right side. This is an example of where we had to do mass data corrections. In this particular one, this was about 18,000 data corrections that needed to occur. And so, in the middle is the actual HelpSystems' interface and what it looks like when you are running it in ad hoc mode so you can watch it. It's an attended task at this point. This would be one of these things that we would have up kind of on the big screen and our IT operates in theory.

If something ran into an error, like on the far right side, that for some reason, the application got slowed and didn't respond fast enough, even with some of the timings that we put into the RPA task, and every once in a while, a web server would get hung, and it would be slow, and then people would respond to it. Hopefully it gives you a little idea on the visualization.

If you want to go down to the next slide, just in the big picture around RPA learnings, first, I think we got a poll.

Paul Necklen: 27:30: Yes, we do. This is our final poll. I'll post that on the screen now. Again, we'll look for a little interaction here. This is, what processes are you looking to or currently automating? Loan applications, account reconciliation, centralizing legacy systems, fraud detection, or account setup? Looks like the results are coming through, and then I'll share that when complete here, and then Matt, you can take a look at that in terms of, what are they looking to do?

Matt Speare: 28:04: Great.

Paul Necklen: 28:06: Okay. Still got some results posting, so it will be just a moment, and then we can post those results. This one is a little tighter. This one doesn't have a clear first place. This is much tighter in terms of spread across the different categories. I think we're done there. Let's close that out, and let's share that. Here we go, Matt. What processes are you looking to automate? It looks like we've got a slight edge to the centralizing legacy systems. That's certainly something you went through, of course.

Paul Necklen: 28:42: Account setup, you talked about that. That's kind of a-

Matt Speare: 28:46: Yeah, we certainly do a lot of that. This is great information because, one, the centralizing the legacy systems, really, when you think about it, it's swivel chair. You use RPA for kind of swivel chair where I need to integrate two systems that do not integrate on their own. Our RPA is a wonderful tool for that. Account reconciliation, absolutely. We have automated a lot of account reconciliation processes, and it works very, very well, overall, and account setup.

I would say out of the 37 different systems that we have, we have 26 of them that the entire on-boarding process moving at change is entirely automated through HelpSystems. We change out 250-300 people per year in our organization and have automated that whole thing. It makes it incredibly easy. Plus, you got a great.

Paul Necklen: 29:42: Great. I'll go ahead and close down the poll, and then this is the final slide. So, we'll wrap it after this and take questions from the audience.

Matt Speare: 29:52: That's great. One, we have done a lot of great things with RPA. At the same time, we know that there is still a lot of upside opportunity. As an example, we had a couple of sessions with our business units over the last few weeks and found another 15 processes that are either a nuisance or are going to be incredible value add to automate and have already identified another $500,000 in annual benefit by automating them. And so, there's a lot of upside opportunities. A couple other ones that are played, so anybody in the world of banking knows that we have to do these negative news searches on high-risk customers. We are in the process of automating that entire process where the RPA task goes out, does the Google searches, saves the results back, and then presents those results to an analyst versus them and having to go out and do the searches on their own in the new account review process to make sure that we've got all the documents we acquired for a new account opening, being able to do that.

Now some of the things that we think are important for ongoing enhancements that the... What really helps is that the RPA tool is not a black box, and the business unit has evolved, and you keep them in the loop. So a lot of times, we have within the processes that we will send the business unit, both, a positive affirmation for everything that was processed correctly, as well as an exception report for things that are not processed correctly so that they stay in the loop as to what is occurring versus coming back and asking a week after the fact. And so, we think that's really important to keep the human in the loop, overall, and then keeping it pretty transparent to the business unit because then it helps drive them to continue to find new opportunities for your organization to be able to automate for them.

With that, I think we can open up to any questions that we have out there.

Paul Necklen: 32:00: Again, thank you, Matt. That was wonderful. Really appreciate that insight from your experience with the Automate RPA tool in your organization. I'm pulling here from the Q&A section. We are a little bit over time, so we're probably only going to have time for one or two questions, so let's start right here.

First one that comes through says, "How difficult was it for you to get started? It seems like there's lots of opportunities for automation, but how did you ultimately get started?"

Matt Speare: 32:35: Yeah. For us, we did have that finding we needed to do about data correction. What we did is, I'll be honest, we used a trainer from HelpSystems right out of the gate. Literally, we purchased on a Friday, and on Tuesday, we started three days of training on how to use the tool. It was well worth it because it was almost like a jumpstart, because as part of that, you're going to automate during it, and you're going to automate a processor to get used to working with the tool. Then, we would carve out time every week so that we could get everybody that was going to be involved sitting in the same room, going through and working through task automation because it made it all better. And so, that was really beneficial because I'll be honest, that within the first 60 days, the tool had already paid for itself, and so demonstrated the RPA side.

I do see here, because I do see the second question, is, "Was it a corporate strategic objective, or was it driven from the bottom-up?" I say, for us, it happened to be in the middle in that I didn't have the CEO say, "We ought to be doing RPA." What I did have is that we had a need. We couldn't throw manpower at it, so they were really looking to us for a solution to be able to do this, and we selected an RPA tool with HelpSystems, and came back and said, "Look, this is what we think it will take to be able to do this, and we think there's going to be a lot of other benefits." Within the first 30 days, when we were able to demo, here's the first five things that we've automated with the tool, and oh, by the way, here's the cost savings for doing that, then a lot of lights went off, because I really do think that this is something that can't be driven from the top-down. You really need that middle tier that is going to come up with solutions to solve business problems.

Paul Necklen: 34:45: Okay, very good. We're about five minutes over. Time flies when you're having fun. Right, Matt? I want to thank the audience, and you, especially, for your participation today. I would just direct the attendees to our website if you want to take advantage of a free 30 day trial. So, the software that you saw running on the screen that's in place and production in Matt's organization is available to you, and we encourage you to download that software. We've got staff that can help you with a proof of concept and kind of get you started.

Again, thank you, Matt. Very appreciative of your time today, and again, thank you to the audience and your participation. Have a great day, everybody.

Matt Speare: 35:28: Thank you. Bye, thanks.