Chatbots in Collections: Smarter Workflows, Better Customer Experiences

What separates a successful collections chatbot from one that creates more problems than it solves?

In this episode of The Payoff Podcast, brought to you by Quantrax, host Andrew Wahl sits down with David Byrne to unpack how chatbots are transforming the collections industry—and why thoughtful implementation matters more than ever.

David shares practical insights on building chatbot strategies that actually improve operations, customer experience, and compliance. From continuous testing and staff collaboration to real-time system integration, he explains why chatbot management should be treated with the same rigor as compliance oversight.

The conversation also explores how consumer expectations have shifted post-COVID, with more customers preferring fast, self-service options over traditional engagement methods. David explains how AI-powered chatbots can help agencies meet those expectations while improving efficiency, data accuracy, and operational performance.

If your organization is exploring automation, AI, digital collections, or customer experience strategy, this episode delivers a grounded, real-world look at what works—and what doesn’t.

Meet our Guest: David Byrne https://www.linkedin.com/in/david-byrne-1862969b/

Learn more about Quantrax: https://www.quantrax.com/

Transcript

SPEAKER_00 0:01

Welcome to the Payoff Podcast, Project One. This is a show for general operating leaders who know the industry, even if the conversations around them are not. Many collection operators are stuck doing what they've always done. New it for smart systems and leadership choice. Each episode features real world insight. Operator. Experiment. No life, no recycle. Technology and leadership. Today is the table of podcasts. This is the payoff podcast, not the people of data. Welcome to the payoff podcast, sponsored by FonTract. I'm your host, Andrew. Today we are joined by David Burney. David is director of product and application support and our sponsor, Fontract. But before joining, he was a veteran collections operator with nearly 30 years of experience. One of the many skill sets he developed, running chatbots inside agencies. Over eight years and across multiple organizations, he turned chatbots into a core part of their operations. As you'll hear, the chatbots brought in millions of dollars in immediate money with much more in potential money. David brings both an operator's eye for detail and a strategist focus on cost, staffing, and scalability. So today we wanted to learn from his hands-on experience how automation can reshape the economics of a collection shop. Welcome, David. Great to have you on the podcast.

SPEAKER_01 1:47

Andrew, thanks for having me. It's going to be fun. One of my favorite topics.

SPEAKER_00 1:52

So you ran a top chatbot for years at multiple agencies. What did those early days teach you about what a collections bot can and cannot do well?

SPEAKER_02 2:03

We got to go back in time a little bit. And we wanted to try to, you know, expand. Labor costs were starting to creep up a little bit. And so we were looking at you know cutting edge technology. So we looked at some demos some chatbots. We were currently a RMX user, so we demoed a course contracts. And that's the one we went with. Various reasons, but the biggest one is the integration, the comfort. So, you know, we were an early adapter. We were like the second uh RMX user to use it. So we expected some bugs and some challenges. Um early days taught you that uh the first the first lesson I learned was it wasn't a set it and forget it, right? You you went through the system controls and you built this cool bot and it talked to you and you tested it a couple times. You're like, wow, this thing's pretty neat. Um but there's more to it, you know. You know, you have to continually test and and develop, um, and how AI gets smarter the more it interacts with people. So it was kind of growing. So you wanted to hear it and continually test it. So that was a big lesson in the beginning is hey, you've got to really listen to this thing. Um you know, log in as a consumer and go through the the different uh prompts and and say the different words and things. Um so that was a big that was a big learning curve. Um other stuff was just you know, how do you introduce it to your staff, right? Because the staff is like, hey, you know, where where's this thing coming from and and what is it gonna do to my paycheck, right? Um so you had to kind of introduce this this piece of new technology as a teammate, you know, somebody that will work alongside agents that won't affect their commissions or their collections, but will enhance it. And how do you do that? Well, you know, you you use it as part of your strategy, it becomes, you know, part of your strategy. So that was a learning experience because I'd never really introduced something that could potentially compete with an hourly agent, right? Uh it wasn't a dialer or a new workflow, it you know, that enhances an agent's capability. This actually competed with one, and and how do you get them to wrap their mind, you know, in their mind and those challenges around that? So that was you know, getting their buy-in was something new. So those are the two biggest um you know things that that I learned early on and how to you know how to overcome those challenges.

SPEAKER_00 4:56

Uh you mentioned that uh you uh it's not a set and forget it kind of thing. So what does that mindset um you know look like when you when it comes to weekly or monthly routines? How do you how do you actually operationalize that?

SPEAKER_02 5:09

Well, the the the best thing I can compare it to is uh compliance auditing. Uh typically an agency will have a compliance department or a compliance auditor that that may listen to live calls. And they may listen to, you know, they may have agents like on a rotation or they're listening, or they may be uh buying a service that screens calls and looks for words and influxes of voice and that type of thing. Well, the bot is kind of the same way, especially early on. You had to have somebody listen to it. You would not only want them to log in as a consumer, go through each step, whether it was the web portal or the you know the phone bot. Um, you also wanted to log in as a compliance person and listen to it and watch and see. Um, so the web bot, if you will, you were talking or uh typing back and forth. Um so you would want to read those transactions, read the notes and and things and make sure that the bot was understanding correctly. When I said, you know, when the agent said, Hi, what's my balance? You wanted the bot to respond, hi, this is my your balance. Not you said you wanted to pet a cat. No, I don't want to pet a cat. I want to, what's my balance? Oh, I'm sorry you wanted to pet a dog. No, I you know, so you were looking for weird things that you kind of see and make fun of on TikTok reels, but um, so you were making sure you didn't show up on TikTok. Um, other than that, it was just you know a lot of going through and and seeing how this thing worked. It was new for everybody, so that's what you do, and you do it on a rotation just like you do a collector. If you audit your collectors once a week, you audit the bot once a week, especially in the beginning, right? Maybe even more. Uh, maybe you treat it like a new collector, right? Where you're actually, you know, looking at every interaction for the first couple weeks to see how it reacted. And then as your comfort level rises, you can back that off a little bit.

SPEAKER_00 7:06

So, what so what were the most common technical issues that you've you've seen? How do you so do you just catch those only by the kind of that monitoring, or was there anything else that you did to stop it before it impacted a lot of consumers?

SPEAKER_02 7:18

Well, a couple things. Um, the the monitoring was the biggest, right? Um, consumers would call um and say, Hey, I tried to make a payment on the bot and it didn't understand me. Um, that was a great call because you went back and you could listen to that and see where it fell short. Um, but you know, it didn't happen a lot. The technology was was good out the gate. So there wasn't a lot of things that that you had to catch, but there were a few. Um, like you would get sometimes you would get stuck in a loop. Um, like our bot uh at the time uh would if you want to talk to an agent, you said agent. So if uh if early on, if a user might have said uh operator, it might get stuck in a loop and come back and say, I'm sorry, I didn't understand you because it was looking for the word agent. Stuff like that. Um and that's just an example. But you know, you watched for that and you listened for that. Um so yeah, I mean that's it's just monitoring and and and listening to the technology. But you know, out of the gate it was really good, and of course, this was I'm curious about that um that that the authentication step.

SPEAKER_00 8:33

Um uh and that that that critical piece of that. What methods did you land on and what what kind of impact did it have on completion rates? How did you make sure that people were um you know properly validated for what they're needed to do?

SPEAKER_02 8:46

Well, that's a great question because first it was scary Terry, right? Nobody, you know, you locked that thing down to where you almost had to, you know, withdraw blood and submit it online, and then you could get into the bot, right? I mean, it was tight. We made it as tight as possible uh because it was new. And you know, our compliance people are like, hey, I don't want this thing talking to Joe Smith about you know Jack, you know, his bill or whatever, or you know, I don't want it yelling at an attorney, or I mean nobody knew what it was gonna do, but it was new. Um so it was like, look, if you're not, you know, if you're not Andrew Wall, you're not getting in. And we want to make sure that you're the right Andrew, right? So we tightened it up. And you might have had to have a social security number, um, you might have had to add your account number, um, whatever it was. But what we found was that people don't necessarily want to give that information, right? Especially because guess what? Not only is it new to you, it's new to the consumer. You know, those 20 million people that you have in your database probably haven't used something like this. Um, so you would watch it and you would see your authentication rates. And then the good thing about our bot um that we were using uh is that we had control. We could control what we allowed uh people to access. So we might say, okay, you don't need your account number because we believe that identifying information, uh, maybe the last four of your social and the last four of your phone number, that's enough to say, okay, this is definitely Andrew, right? Because the account number sometimes gets lost on a piece of mail or they can't find it, or whatever the case may be. Um, and then we would look at those rates. Because good the good thing about having an integrated real-time bot is that I could see when people were getting to the authentication step and then when they were dropping off. And as you made these changes, you could see, hey, they're moving forward. You know, because this thing only only is top value when they get past the authentication rate. Up to the authentication rate, it it does the it's no value to the agency, right? They have to get past it and take a step or two or three or resolve, ideally, um, to get it. So yeah. So you would watch it and you would make changes, you would, you know, run it through your compliance department or your attorney and say, hey, what do you think about the last four of the social? It's really kind of tricky. What about the date of birth and the zip code? And or, you know, what about the account number and the name? You know, if if if I have an account number and my full name, you know, would that pass you know the mustard for you and that type of thing? Um, but the easier it is to authenticate, the more transactions you will have.

SPEAKER_00 11:34

Right. So um, so how did you uh you know strike that balance between uh compliance and documentation uh while keeping the conversation simple enough that people actually follow through on payment and and you deliver that value back to the agency?

SPEAKER_02 11:49

Trial and error. Um you know, uh hundred dollar bills to the compliance person. Please let me do this. No. Um all kidding aside, you really, it's trial and error, right? You you adjust. Um and every agency is going to be different where where their threshold lies and you know in this, what they're willing to do, because if you loosen it up, it is possible that maybe somebody does get the wrong account. Um it's unlikely, but it's possible. So you need to have controls in place. And of course, you have PCI and you have all these other governing bodies that that want to make sure things are safe and secure. So all that has to pass. So that's number one, right? You you have to make sure that your the consumer's data is protected, your company's protected, right? All kidding aside, those are the most important. But then from there, it's how can I let's cross check that box. We're protected now. How can I make this experience better for the consumer? And it starts with that authentication. Um, so you you have a balance. And you know, typically, you know, two pieces of identifying information is is what you're looking for. So, you know, if the phone number matches or this the date of birth matches, or the name and the social match, or you can put more weight on one, you know, social security numbers pretty guaranteed, right? So that's you know, that weighs higher than something else. So trial and error, but you get through it.

SPEAKER_00 13:18

Uh is I mean, AI has come a long way in eight years. Uh, so um, you know, uh agencies and collection operations have to be uh seeing all kinds of different uh AI kind of options that are that are out there. Um how is uh from from your experience, how is it how has it changed uh in in terms of these um the these types of chatbots and how it how it works?

SPEAKER_02 13:40

Oh, it's it's just growing. I mean, it's it's just getting better and better. Um, you know, I before this this meeting, um, as I was you know preparing for trying to refresh, uh, because I knew you were gonna ask me questions from a long time ago. Um, you know, I read a ring, a LinkedIn article, and it said, you know, it I don't even remember who it was, but it was it was uh, you know, the banking and and arm uh industry saved 1,500 FTEs through AI up to to date. Of course, that's a huge spectrum, right? Of of all you know, banking and credit unions, and you know, we're probably in there somewhere. But uh the idea is it's getting bigger. It's getting bigger, it's getting smarter. Um the voice, you know, when when you used to talk to it, um, you know, we have the bot that we deploy has three different ways of reaching the bot. Um, so you have like the script base what we call the CSP, the customer service portal. It's very scripted. You just click, click, enter, enter, that type of thing. We also have talking bots, we have multilingual bots, and but the the voices are getting better and better and better. And as a matter of fact, it's getting so good that some states now require you to say, I am a chat bot. Right. How scary is that, right? I mean, they're making you say it because they they don't want a consumer to think that it's a real person.

SPEAKER_00 15:00

So well, I was gonna ask about the consumer response too, because it consumers are becoming much more accustomed to this, um, to the point that they may even prefer to just deal with a chatbot than a real person. Uh is uh and I wonder what your experience has been on on that front. Like how are how are consumers reacting to this and what are they sort of more accepting? Um it must potentially cause some issues with with compliance and people you know don't really understand uh how it works.

SPEAKER_02 15:28

You know, it's funny you mentioned a story. I was um in a meeting with a very highly regarded, very, very I mean, expert in in this industry, long time vet, um, just a great guy, very knowledgeable, and and you know, one of the power players. And he told me, he said, self-service will never work in the collection industry. And I said, I have 20 million dollars that says you're wrong. And here's the deal. I wasn't convinced when when I originally brought this to the the owner of the the company I was working for back when when Quantracts just released it. I wasn't convinced that it would replace agents. I'm still not, it's it's not gonna replace agents. Um I wasn't convinced how well it was gonna do. What I thought was this will give our salespeople a cool tool to run out and say we are open 24 hours a day now, and it's not an IVR because at the time everybody had an IVR. Push one to do this and enter your day, you know, that horrible thing that that you know. Um, but me, I was like, hey, for you know, for transactional pricing and this thing's a uh great deal, you can tell you know our clients that if somebody wants to pay their bill at three o'clock in the morning or dispute or file bankruptcy or what whatever the case may be, notify us of these crazy things, they can do it. Maybe they work nights and they want to do it or whatever. So um fast forward, you know, eight, ten years, and now it's like, you know, wow, you know, it's um it's really it's part of a workflow now. And um, but you know, the consumer reaction, luckily, um, we didn't have to do this because the world did it for us, right? Along came this this this funky illness called COVID, right? And everybody started working from home, and everybody started, I'm not touching, I don't want to be in front of somebody, right? I want to self-serve. You know, Walmart decided it was a great idea to have one cashier and 50 self-serve kiosk now, right? Where you could you could stand in the uh the never-ending line because you want to make a statement, or you can go over there and check yourself out and leave. Um, so so we're getting used to self-serve gas stations now. You know, you walk in to to buy a pack of gum. Um, I was shocked, and it's self-service kiosk there. And the dude behind the gas station's selling cigarettes and beer and playing on his phone. That's it. Anything else, go go self-serve. So whether you like it or not, whether consumers like it or not, their opinion really doesn't matter because they have to use it now. That's the way the world's going. So we've kind of slid in there. The the value add is that nobody really liked talking to us, anyways. You know, it wasn't like, you know, the the the nice old lady at Walmart to check you out. They didn't want to talk to us because we were bill collectors. Yeah. So now they have a way to take care of their bill without ever speaking to us. So it's kind of a you know, it very seldom does this industry win. This is a win. So this is a win.

SPEAKER_00 18:48

With with all the change, there's there's a lot more uh options out there in terms of you know AI chatbots and AI different tools. Uh so let's talk about you mentioned it before, but let's talk more about platform integration. How does this real-time integration uh with a platform change your approach uh as an operator to building dialogues compared with an API-based chatbot that's outside the core system, it's kind of bolted on. How's it different?

SPEAKER_02 19:12

So it's it's night and day. So there's lots of chatbots out there, like you said. You know, R CRM, obviously, I represent Quantracts. R CRM, we can integrate with different chatbot providers. Luckily, we have our own. Um, but here's the difference. When we build an API for a for a chatbot, uh, a bolt-on chatbot provider, for example, that's usually, you know, we send the file to them, you know, with all this data, consumers access it, and then that night sometime, or the next morning, or whatever, they send this big file back or all these changes that that may have happened. Um in our industry, that's a long time. And here's a great example of that. Being completely integrated, if a consumer reaches the chat bot today, this morning, and they go on the chat bot and they say they file bankruptcy, we know about it instant within seconds.

unknown 20:12

Right.

SPEAKER_02 20:12

Right? We can apply a smart code that alerts an agent, we can remove the phone numbers, we can do all that cool stuff. If you go to a different chatbot and you say, I filed bankruptcy, if that chatbot can even handle that, a lot of times they're just basically glorified payment portals. Um, but if they can actually take that bankruptcy information and update it, uh you don't know till tomorrow. Well, if that account's on the dialer, guess what? You just you're gonna call that person, they're gonna say, Hey, I let your chatbot know three hours ago that I filed bankruptcy.

SPEAKER_00 20:44

Yeah.

SPEAKER_02 20:45

Big problem, right? Especially a compliance risk. That's just one of the many things. Um, you know, we have could, you know, it's funny, it it at a much larger shop where the bot was producing much, much more um, you know, millions of dollars. You know, we would have consumers that literally would pay the bot and then call in and let the agent know that they made their payment. It's the funniest thing I've ever seen. You know, they because they wanted the agent to know that their payment was made, but they didn't really want to make it with them. They wanted to make it with the bot. Um, so you know, okay. Um, it's just it was just, you know, it wasn't very many, but some people did that. And because they they still weren't trusting, right? Uh of this new tech, even though it's been around, they they're still trying to, they won't you know, grandma wants to make sure that you know the payment got posted. Um, and I get it, you know, I understand. Um so that stuff was kind of funny, but um, you know, the integration allowed the agent to see right then, yep, I see your payment, I see you went on the bot, you know, 13 minutes ago, you made your payment. Thank you very much. We'll talk to you next month. So uh another thing with uh with us is we have developed ways to transfer stuff to the chat bot. So agents, for example. Sometimes it takes people a long time to get their credit card out. And you know, there's PCI DSS regulations. Everybody's working from home now. So let's transfer this call to the chat bot. Let the chatbot take your payments so my agents never have to touch your credit card information. So you can't call me back and say, you know, hey David, you know, we got an issue, we got this weird charge, you know, and we've got a thousand people working from home, you know, whatever. So transfer to the chat bot, they can talk to the chat bot. So integration really makes makes a difference. And when I was looking and and exploring chatbots, it was three major things that that I was looking at. Compliance, of course, uh, because I wanted to make sure that that we were safe as an agency. Um customer uh satisfaction, the the I the experience when they log in, how the flow was, how it looked, um, and then integration to the point of how soon will I know what happened? And there's no sooner than immediate, right? It doesn't get any faster. Um, so that was a big deal. Um so yeah, I mean that that was that was that was very important in the decision making for us.

SPEAKER_00 23:18

Well, so for for an agency who's thinking about turning to chatbots uh and and trying to apply them into their own uh system uh just for the first time, what do you recommend is the minimum you know viable operational plan? What are they what do they need to do uh that they can need to have in place before really going live?

SPEAKER_02 23:36

They need to decide how they want to integrate this chatbot into their daily routines because this will be this is almost like hiring a new person, right? How does that new person fit in? What are they gonna do? What is their responsibilities? Uh what's their return on investment, right? Because you don't hire somebody and expect them to break even, you hire people and at least production people is what we're speaking of. Uh you hire production people to make a profit, right? Um, so how will that person fit in? And the same thing with the bot. How is it gonna fit? What are you looking to gain? Is this like me 15 years, 12 years ago, whatever it was? Now, am I looking for a sales tool, you know, to be able to say, hey, I could take payments and I could take disputes and all this good stuff in the middle of the night? Or am I looking for something that generates revenue? Well, if you're getting a chat bot today, you're looking for something that generates revenue. And then where does that fit in in your workflow, right? Today you have all these omnichannel, you know, it's a great big buzzword running across the industry's omnichannel, um, and then of course now AI. But um, so where does the omni channel meet the AI, right? Where do they intersect and what do you want to do with them? Have that discussion and get the buy-in from the people that you depend on, your managers, your frontline supervisors, and your agents. Explain to them what you're doing and why you're doing it and how it benefits them. Uh, at first, the agents were, you know, I'm not, you know, I don't trust that thing. You know, we it was named was Alex, right? Well, I don't trust Alex. And um, but then it got to where, you know, our agents were like, here, here's the link. Next month, just click this to make your payment instead of bothering me, because they could go on to something else, right? The idea is to free up your agents for complex phone calls, things that really need that personal touch. And there's plenty of those, guys. It's not going to replace an agent. Now, will it help with attrition? Uh, will it help with labor costs? I mean, all those things are certainly part of it down the road, but it is never going to replace collectors in general across the board. So, I don't know what you asked me, but that's the answer you got.

SPEAKER_00 25:54

Yeah, no, I think that's ideal. That uh gives pretty clear uh instructions, and I think you've given uh everyone a lot to think about uh and a lot to a lot to consider. Uh so thanks a lot for your time, uh David, and uh hope to be able to have you back on at some other point.

SPEAKER_02 26:07

Absolutely. Anytime.

SPEAKER_00 26:08

Thanks for having me. That's it for today's episode of the pain. Thanks again to our sponsor, Quank, Brigaders of the RMA Death Recovery Management. Learn more at Quankac.com.