Adam Parks (00:05)

Hello, everybody. Welcome to Applying AI, our podcast where we're talking about the things that actually matter related to artificial intelligence and how we can use these tools, apply them to our daily lives and to our businesses to find success. I'm here today with my co-host, Mr. Mike Walsh, joining us as usual. How are you doing today, Mike?

Mike Walsh (00:28)

Great doctor. Let's go. Let's have some fun.

Adam Parks (00:29)

Awesome. And today we've got a fantastic guest with a lot of experience in the collection space and a fellow- I'm gonna say digital nerd here- Kristyn Leffler joining us from Resurgent. Kristyn, how are you doing today?

Kristyn Leffler (00:44)

Great. Thanks for having me, guys.

Adam Parks (00:46)

Absolutely. We really do appreciate you coming on, spending a little bit of time with us. For anyone who has not been as lucky as us to get to know you through the years, can you tell everyone a little bit about yourself and how you got to the seat that you're in today?

Kristyn Leffler (01:00)

Absolutely. So I work for Resurgent Capital Services, a large debt buyer collection agency now, and I have for quite some time. I started as an analyst. My background is learning SQL, account segmentation. I spent some time doing some judgment work, deciding what non-performing judgment strategies should look like. And then I spent about eight years developing Resurgent’s digital presence from the ground up. 

So that looked like email, portal, SMS, figuring out what was the right way to engage customers in the collection space. And I think I benefited from having a background in how data flows, but my educational background was both in supply chain, economics and marketing. And those three things together really allowed me to bring some of that marketing knowledge that just didn't exist a lot in the collection space. 

I know Adam, you're very familiar with marketing in the collection space. And maybe the gaps that have existed there and I think are beginning to close. So I was just really in the right place at the right time to figure out that we are missing some key data in what is like the collections conversion funnel. How do we make sure we build those guardrails and we build that data infrastructure so now we can learn, iterate, experiment as we scale. 

And then I moved after spending a lot of time on the digital side, learned from spending all this time doing product and portal, and realized we have made the customer experience. Now I'm biased, but I think we've made an excellent customer experience, and we left the agent experience behind. All these things we've learned about minimizing friction, making the right choice easy on the customer side of the fence.

We didn't do that in all of our agent applications. And now I spend a lot of time doing that same thing on the agent side of the house. What can we do to surface the most relevant insight to the agent when they are having this live conversation with the customer and creating a seamless user experience across the voice channel as well as the digital channels. So I'm very lucky to have been able to work in all these spaces in one company.

Adam Parks (03:22)

Well, it's a lot of different pieces and parts to the same customer journey. And when I hear all of the pieces and parts, I think one of those common threads that I hear throughout is digital trust. Creating an environment in which digital trust can be born. Relationships are built through a shared experience. You don't have that opportunity when you're engaging online with a consumer. 

So you have to find other unique ways to develop that type of digital trust, which I think leads down the path of successful e-commerce across the board, regardless of whether you're selling pens or you're collecting debt, right? Digital trust, in order to transact online, is a mission-critical piece. As you went through the process of evaluating that, let's call it pre-Kristyn Resurgent

online presence, what did you see that made you say, I need to take action and then start putting some of those actions together into a plan?

Kristyn Leffler (04:27)

I think one of the first things that we spent time on was working on the website and the portal. And our website was originally developed by developers, right? And they said we'll just have this single-page application, and it will just change dynamically. But that meant we had no conversion funnel. So I was like, we need differentiated page views. Like, I have no idea where my friction is. I have no idea where my drop-off points are. 

All I know is they landed on this page and some of them converted, but I don't know what the denominator is. So that was like the very first step. And I think most collection agencies now have gotten past that. They don't have a single web app with dynamic content. They have differentiated pages. They probably have Google Analytics at a minimum. But yeah, that was like the very first step. 

And how are you going to tailor the customer experience to say, you forgot to enter your payment method. Here, come right on back when you don't know where they fell off, right? Understanding customer behavior is absolutely step one before I can start to optimize: where's my biggest on the Pareto, right? Where's my biggest opportunity that I should focus on first? Or what is my messaging? What is my segmentation and where do I focus my content? All of those things really have to work together to have a successful kind of e-commerce experience on the collection side of the house.

Mike Walsh (05:54)

I think though not all agencies are there yet, right? 

Adam Parks (05:59)

I guarantee you they're not. I got off the phone with one five minutes ago.

Kristyn Leffler (06:01)

I'm so spoiled.

Mike Walsh (06:05)

But it's a very good point, like I think what you just said there, Kristyn, is like that's what customers expect, right? Like you expect them to have that already, right, and if you're not meeting the expectation, then you're putting up a wall before you even get started, right. And then the lack of data, I mean, it's a killer.

Kristyn Leffler (06:31)

I say this all the time; you've probably heard me say this before. We all share a consumer. Our consumers are Amazon's consumers. Like they have been shaped and trained by their e-commerce experiences. We don't operate in this vacuum. We should stop treating our customers like we do.

Mike Walsh (06:49)

This is different.

Kristyn Leffler (06:49)

Like they don't say, well, today I'm gonna interact with a highly regulated industry, so I am going to lower my expectations. Like that's not how this works. We could all say, well, it's fine because the law says I have to do this. That's true. You do have to follow the law, but you have to understand what the customer's expectations are when you do. So you need to fulfill all of your regulatory requirements, but you have to do it in a way that the customer is used to interacting. You can't force them to lower their expectations. That's not how this works.

Mike Walsh (07:24)

Right. And it's a very good point too. It's not like Amazon and Nike don't have laws regulating them, right? Otherwise they'd say come buy this two-dollar shoe and charge you eighty bucks in fees. Like when you log out. So I constantly call this the Amazoning of debt collections. That's how you should think of it. 

Kristyn Leffler (07:42)

Yes. Or the FinTechs. The fintechs have a very seamless originations pro- well, but many

of them have a very seamless originations process and they're regulated as a lender. We're collectors; we're not underwriting.

Adam Parks (07:58)

But consumers are expecting to engage with a collection agency through the same channels in which they originated the debt. If a consumer went into the bank branch to fill out a credit card application, they have a different expectation than a consumer who borrowed that money from a mobile device.

Mike Walsh (08:19)

I think to a point.

Adam Parks (08:21)

Look, there's always to a point, right? But I think you can look at that general flow.

Mike Walsh (08:27)

I even still think that if I went to CarMax and went and picked out my car, it doesn't mean that I'm not a Carvana customer and I'm not gonna do digital, right? 

Kristyn Leffler (08:38)

That's true. 

Adam Parks (08:38)

No, it doesn't carry across all areas. Some things you want to go in and touch before you buy.

Mike Walsh (08:40)

I think that you're right, though. I think that's the minimum expectation, right? Like, I did this digitally; I expect to be done digitally. But just because it went brick and mortar doesn't mean digital isn't more convenient.

Kristyn Leffler (08:53)

That's true.

Adam Parks (08:53)

No, but even when you bought from Carvana, you bought it digitally. You may have used your mobile phone in a dealership, but you still bought that vehicle through some sort of a digital channel there, right? And I was watching something the other day that was talking about when Tesla, for example, started selling cars online. And it was the first time that your average person was seeing a hundred thousand dollar item being sold online. 

And I believe the first version of their website was like 62 clicks to complete the purchase. Right. 62 clicks, I mean, that's a commitment. I feel like that's a day's worth of clicking. And does it count if I click everything four times? Probably not.

Kristyn Leffler (09:33)

I think that's now you're like, you probably don't want to go here yet. But the idea of forms and clicking and friction, I think that's where maybe we're jumping too far ahead now is where agentic commerce totally rewrites the playbook, right? Like why are we even gonna fill out a form anymore? Like the concept of filling out forms was made for the ease of the consumer, not the customer.

Mike Walsh (10:03)

Processor. Yeah, correct. It's data processing.

Kristyn Leffler (10:03)

But like the ingester of that information, the processor of that information, right? It wasn't for the purpose of a good customer experience. It was like, I'm putting the onus on you, customer, to fill this out to make it easy for me. But that gets all flipped on its head if you can automate that. You know my payment credentials, just fill it out for me. You know my address, just fill it out for me. I think that's like a super interesting radical pivot away from where we're going.

Adam Parks (10:28)

I mean, it's a lot of data saved in Chrome.

Mike Walsh (10:31)

Well, that's what Google Pay and Apple Pay are, right? Like it's just laziness.

Kristyn Leffler (10:34)

Absolutely.

Mike Walsh (10:36)

And it's great, yeah.

Adam Parks (10:37)

Well, it's comfort. It's digital trust. You see the Apple brand and you say, I want to buy this pair of shoes. I don't know that company, but I know Apple, right? So you know that you're putting some sort of a barrier.

Kristyn Leffler (10:46)

Right. And they're gonna protect me.

Adam Parks (10:49)

Yeah, there's some sort of a barrier there in between you and the merchant themselves.

Kristyn Leffler (10:54)

It's fascinating. There's like a barrier. The consumer perceives it as a barrier, right? Like Apple's protecting me, but it's also a radical reduction in friction because I don't have to remember all that now. And I know personally I will make the purchase more often if I can Apple Pay it. 

Adam Parks (11:11)

Yeah. Maybe I could double click it and face scan it.

Kristyn Leffler (11:14)

And I think that is increasingly true. Like Stripe, here we'll probably quote Stripe a few times here. Stripe's report they put out on wallets says like, a manual card entry takes eighty-five seconds, and the digital wallets take forty-two seconds. So just think about that. You said like sixty-two clicks or whatever. Now we're talking less than sixty seconds.

Digital wallet transactions happen in less than sixty seconds. Like that is the speed at which customers are expecting their interactions to exist right now. 

Adam Parks (11:50)

No, I'm curious how the consumer engages or how that timeline changes when there's a negotiation at play. Right. And how digital negotiation, because if we think about the three things that matter in a negotiation, and this isn't me, this is Herb Cohen, it's time, information, and power. 

And when we move to an online environment, those negotiating pillars change fairly dramatically, right? Like we're not managing it the same way. And people do manipulate those from a negotiation standpoint.

I mean, I won't sit in a car dealership after I look at a vehicle, right? I'll take the cell phone number of the sales associate, I'll leave, and I'll close that deal via text message because now they no longer control my time. I won't make bad decisions just to get out of there before dark or whatever the case may be. How do you think the lack of that sense of urgency translates into consumer engagement through virtual negotiation?

Kristyn Leffler (12:46)

I have, like, a slightly radical take on this. Like, what I think is maybe slightly radical. So we think about the collections industry; we think a lot about negotiation. Like, is my agent a good negotiator? Am I getting somebody to like the right offer? I think increasingly about digital interactions.

It's less of a negotiation. It's about getting the right offer in front of them so there is no negotiation. It's about an offer of curation, selection, and reduction of friction. Like customers, from what we have seen, customers aren't really trying to slide the slider bar around a lot and figure things out. They want to click it and be done, just like the Stripe checkout, right? They want to click the offer, pay it, and be done, as long as it's an offer that appeals to them. 

But getting the offer that appeals to them correct on the first try becomes very important because then you start to lose that engagement. I think that it's still a different dynamic when a customer calls in and is having a discussion with a human. That is relationship-based. You do still have those three negotiating pillars, but I think the medium of interaction drives the expectation. And the expectation in a digital interaction is less about negotiation and more about transaction completion.

Mike Walsh (14:06)

But that's part time though, right? Like, because you don't want to be digitally there for twenty minutes going down each three percent cut, you know, reduction of settlement. You wanna go on a logical, fast way. But I think you're right. Like absolutely, if you can start by saving that digital customer time with an offer that's appropriate to that. 

Like when I started in collections, the first thing you did is ask for PIF. And I said, Why would I do that? If they were good at PIF, they wouldn't be in this office. Like I wouldn't be calling them. Like that's absurd. And they're like, you don't get what you asked for, right? No, you don't. But you also don't get PIF most of the time on the first call. 

Like when you get someone in collection from a seven thousand dollar debt, I think offer curation is huge, right? And getting it to them. And then Kristyn goes back to what we started with: the data. So if you have data that tells you what offers match, or if you have an offer engine that you trust, and it is like we deal with many clients. And we can see some offer engines like this are a little rough. And we see some that are very flexible, and customer behavioral data can now tell you what to offer in the first place.

To shrink that time. Every click, we're losing people, right? Like e-commerce 101. Every click, every, every wall, you have a drop. So I don't think that's radical. I think that is the future, what you just said. Like, I think having that knowledge, and I think AI is very good at this. Your data can tell you, and AI can make that very easy for your data to tell you what offers are getting taken and what aren't by balance, by segment, by product, by portfolio.

Kristyn Leffler (16:08)

Yeah. But whether or not you've seen that customer before.

Adam Parks (16:09)

How has the accuracy of that changed over time? Right. We've since, let's call it 2022, we've started seeing agentic AI. Have we seen an improvement in the ability to put the right offer in front of the right consumer? Is that orchestration getting to that point now?

Mike Walsh (16:30)

I think so.

Kristyn Leffler (16:31)

I think it comes back to the digital data and what I call this digital exhaust, right? All of this information that a customer is creating, whether they know it or not. Just like you and I interact, I'm gonna click over here, I'm gonna hit the back button, I'm gonna refresh the page, and then I'm gonna leave, and then I'm gonna go jump over to this thing, and then I'm gonna come back, and like all of that is data and is really gnarly.

Clickstream data that is not meant for human eyes to consume. It has to be parsed and cleaned up. And if you're lucky, you have data engineers who put it in a data lake for you. If you're not lucky, then you put it in an AI and hope for the best. But like all of that digital exhaust data is super powerful, but it has to be collected, stored, and then later analyzed.

To understand and to hone that engine like Mike's talking about. And I don't think all of the collection agencies or even a lot of the debt buyers have the capacity or have had the foresight to be storing that information to then leverage it. So whether or not we are getting better since 2022, I think is bifurcated between the folks who are data-driven shops and who are not data-driven shops. And Mike, you have more clients, so you would be able to speak to that probably a little better than I can.

Mike Walsh (17:57)

Yeah, this is Adam, and it's been a continued discussion for a year now. 

Adam Parks (18:02)

Well, it's who invested in the data lakes, right? That bifurcation comes down to which organizations made the investment into the data lakes to organize that data and how many truly have you not just data analytics tools, Power BI, or whatever their flavor is, but have the analysts to actually run it. To look at that data to dissect it.

Mike Walsh (18:28)

Or the vendor to run it, right? 

Adam Parks (18:28)

Yeah. Yeah, fair enough. But again, the same solution is necessary.

Mike Walsh (18:35)

Yes. So you're right. We have people with multiple systems because maybe they purchase someone on a different system, you make a data lake, you're gonna run behavior, and then all that tracking behavior, they can't even so many of our clients can't ingest all the data we provide, which is fine.

Sometimes we can store it for them, and they can just access it. We give each client a cloud, and they can go in and access that data, right? That's where analysts like to weep. They're so happy. Because even if their system isn't telling them that data, we could give it to them or just show it to them.

But I think that's gonna be the change in this next generation of, like, I'll call it a CRM, but operating systems for agencies. There's gonna be a lot more fields that they can track, right? Like there's gonna be a lot more data that's gonna come back to them that they can use. Like for me, in Kristyn's example, if I go to your portal and I'm on there for seven minutes looking around and I look at 10 different offers and I didn't make a payment, I sure want to call that person as soon as I can, right?

Like that information we send back to the client so they can phone call them, like with a human. Because even our virtual agent's not gonna help them. At that point, because it's running under the rules, under the guards, even our agentic, unless it's got special programming where it can go around settlements, or that person needs help, like they're trying to make an effort. So that's like where you can get that human intervention and figure out something like a special case, right? Like that's but I think that's where the industry is gonna go, right? Like, that's where it's going. But I don't know if it's going at the same rate. 

Kristyn Leffler (20:27)

Yeah, I think that's really interesting, and I know you guys talked to Michael Lamm not too long ago, like a really interesting direction of what will happen over the next few years as the haves and the have-nots are able to separate, because again, it's the same customer. We are all like, this is my collection agency, but chances are it's the same customer that you're working for, that I'm working for, and it's like a wallet share, right? 

Where Collections is still converted, it's a conversion journey, just like e-commerce, to kind of bring it back to that. Like customers decide when and who they will engage with. You cannot force a customer to engage with you. So whether or not that customer chooses to engage with you, are you a positive experience for them? And that means they had a friendly email. They called me, but they didn't harass me.

Those decisions, kind of back to your point, Adam, are shaped by trust, timing, the channel, the message, how that user experience happened, message relevance, like to your point, Mike, hey, noticed you looked around my portal for seven minutes. Ooh, creepy. Now that's creepy. But if, like, we saw you were on the website, and we just wanted to offer some help. Less creepy, but it really matters that you get that right because if you go one step too far, now your trust is eroded because, like, now you're spying on you, right? But I expect you to know that I was there. 

I expect that of you. And if you send me an email that's like, did you know we have your account? I'm like, Yes, I know you have my account. I was just there yesterday. It's really hard to thread that needle. And you have to have all the data; that's where firms start to separate. The ones who can do it well and who can't do it well, and the ones who do it well will get the wallet share.

Adam Parks (22:33)

So it sounds like the ones who have started building the data lakes are the ones that are poised for success, but success is actually activating that data into actionable insights that allow you to engage with that consumer, not just the right time message channel, but the right content in that message to hit.

Kristyn Leffler (22:52)

Yes. The right offer.

Adam Parks (22:54)

Write the right offer at the right time. And this is where the orchestration engines come in. And from what I've read, and I've written some research papers recently about the impact of artificial intelligence on the debt collection industry, and read a couple of Stanford papers and some from the National Bureau of Economic Research. And it sounds like the voice AI bot that everybody's been chanting about at the conferences lately is not the be-all, end-all solution. 

And that the most impactful solutions seem to be those that will enable the consumer down a self-service path or information like kind of what you're working on now, where you empower the live collector by providing them with new better data and information at the point of contact to help drive them towards resolution of an account, whether that be a collection or whatever that may be, but resolution being the the core objective.

Kristyn Leffler (23:50)

Yes, I think it's I think it's both of those things, right? Like self-service is great for a lot of reasons. Lower complaints, customers are happy. So yes, drive as much self-service as possible. And then, the goal is not to use AI to remove humans from collections. It's to remove all the sludge from the system so that humans can focus on what they do best. Like, what do humans do better than AI? I think maybe some cynical people say nothing.

But relationships and empathy, right? Like, the thing humans still have an advantage in is relationships and empathy. And that's what we should keep the humans doing. Like let them focus on those things.

Mike Walsh (24:41)

I think so too; there are certain cases that you just can't pre-program, right? That AI is gonna kick out, right? The lady who's not paying her utility bill because they ran over her flowers and they didn't replant like they said, right? Like that's a real case though. 

Kristyn Leffler (25:02)

I have heard that exact case before.

Mike Walsh (25:05)

Right, like so there are things that there are always exceptions. 

Kristyn Leffler (25:09)

Yeah. Or things that live in filing cabinets. Like AI can't know about it.

Mike Walsh (25:14)

So I think you get as much covered through AI because it's faster, cheaper, easier. Some people prefer it. It's private; there's no embarrassment or shame, right? So there's gonna be a portion, but there's always gonna be a portion that needs a human. I think it will never get over seventy-five percent. I hear people saying ninety, and I'm just like, good luck. Like I've been in collections too long. I've heard too many; there's just no way, right?

Adam Parks (25:43)

Sounds aggressive. It's too personal of a situation to be that impersonal.

Kristyn Leffler (25:51)

I think it depends a lot. It depends a lot on the paper type too. Like a lot on the paper type. Medical, like complicated hospital billing. Like I don't think you even get to 75%. But fintech paper, are you pro? I'm saying way higher, right?

Mike Walsh (26:08)

You could get way higher, right? Yeah. You're right.

Kristyn Leffler (26:11)

Is it ninety? I don't know if it's ninety, but it's higher.

Mike Walsh (26:14)

Not yet, but it will be, right? Like it's changing fast. 

Adam Parks (26:23)

Consumer comfort with those communication channels will also have an impact on how much of our volume can be moved down any given channel, right? 'Cause what we're seeing today versus three, four, five years of AI voice usage from now may land differently with the same consumer.

Kristyn Leffler (26:39)

Yep. It's comfort. I think we are all operating from the paradigm that we're talking to the consumer, right? But I know increasingly we see consumers have authorized a bot to negotiate on their behalf. So in that case, you're not dealing with a consumer. So what does that do when you're not actually talking to a consumer? Like they don't want to talk to a human.

Adam Parks (27:11)

The bot-on-bot warfare. 

Kristyn Leffler (27:11)

They would rather just do an API transfer. Right?

Adam Parks (27:15)

The bot-on-bot warfare is something I bring up frequently because, like, we've seen a rise in pro se litigants. I don't know why we wouldn't see a rise in consumer bot usage. We know that the debt settlement companies are developing them, right? Like we've all had some experiences, and we know that the consumers are going to have more and more of them, especially as time permits. And then you've got the phone manufacturers making their modifications and their layers. 

You've got AT&T with Active Armor, and Verizon with whatever they've got going on, both again trying to restrict the availability of communication channels. And you have to validate that trust in order to get through to that consumer.

Mike Walsh (27:57)

I think the bot-on-bot is like we see it happening right, like what I've seen Heath Morgan talk about. I think it was the Northeast ACA conference. Those calls can go twelve minutes with a consumer bot, and half the time the consumer has not authorized them to make a payment. So to me that's a perfect case for AI. Why would you have a human being waste their time talking to a bot, right?

Kristyn Leffler (28:28)

Of course. But I think it's only a matter of time until they are authorized to make payments. Like, I think it's only a matter of time.

Mike Walsh (28:37)

They are like, literally, there he was talking about a tool that you can say, Okay, if it goes to a hundred bucks you can make a payment, but if they don't go to that, don't. 

Kristyn Leffler (28:45)

Yeah. Set my reserve price, right?

Mike Walsh (28:47)

Right. Or just negotiate, see what they'll offer me. You can do it either way. Yeah.

Kristyn Leffler (28:51)

Yeah, that's more what we see. Tell me what the offers are. Thank you very much, hang up. You're like, collectors hate them. But they go forever. No one's happy.

Mike Walsh (29:02)

But I would think that's where Adam and I have talked about it a lot, is if you could identify the bots quickly, then your bot could go to a three percent decrease in negotiation. Let it negotiate all day. Who cares? 

Kristyn Leffler (29:16)

Yeah. Well, eventually your token costs. Eventually you care. You don't want to have a three-hour-long call going back and forth. But I think like here's the thing that the bot on bot, or even having a bot call me, like it's solving a problem that it's like working around the problem. 

The problem is information transparency, like a lack of information transparency, information asymmetry. If I could just give, if I knew what the right offer was for that consumer and just gave it to them in self-service, they'd be happy, and they wouldn't have to authorize a bot to call me and negotiate on their behalf, right? 

The whole bot-on-bot or using a bot to call in and negotiate, it is like a giant workaround for a problem that is just we haven't figured out how to get the right offer in front of you in self-service yet. That's my take.

Mike Walsh (30:09)

No, I think you're very accurate. Or it's a person who really what is the chance of them following through on a payment plan anyway? They might make one payment. And which is fine, right? As anything helps, but I also don't think

Adam Parks (30:26)

Does a broken promise become like an abandoned cart? Do you end up following similar e-commerce level strategies to bring that promise back into fruition that you would use on an abandoned cart?

Kristyn Leffler (30:35)

You should be. You should be. It's just like, I mean, it's like marketing101, right? It's way cheaper. It's like I don't know what the old stat is, like five times cheaper to keep a customer than get a new customer. Of course you should be following all the same playbook for your breaks.

Mike Walsh (30:55)

Yeah. It's just re-engaging instead of engaging, right? Like it's much simpler, faster. And the person showed a behavior that is very, very positive, right?

Kristyn Leffler (31:05)

Yeah. You have way more knowledge. You don't have the cold start problem on a broken payer because you already know they had willingness and ability. Now you just need to solve why they didn't continue. Lifestyle change, payment method failure, and then you can start bifurcating those, segmenting those just like you segment the other, the front end of the funnel, right? What's the reason? Target that reason.

Mike Walsh (31:29)

Everything: overzealous, overpromise, and just like realize they forgot they'd include the price of gas rising up in their budget. Right.

Kristyn Leffler (31:37)

Overextended, right? I mean, you guys know you've earned collections; most customers aren't like perpetual. No, people aren't trying to get into debt. There are some fraudsters, right? But it's a problem managing cash flow for most of these people. And it's no different when you are optimistically signing up for a payment plan. They want to pay. They just have historically had trouble managing their cash flow, and it's no different now.

Mike Walsh (32:04)

It'll be interesting to see, I think, especially for agencies, right? Or even Resurgent, like how flexible plans become payment plan-like guidelines because of that, because data's gonna show, right? Like creditor A, like you just said, we're all working the same customer, but is it the average debtor that has sixteen creditors after them?

So the ones that are Amazon, right? Over some others, like even Target's great, right? Going to their website, but it's still not Amazon, right? Like it's not gonna perform better because when you abandon that cart, they're gonna be right there and then they're gonna work even harder to keep that payment flow going. 

So I think that's underrated: the future of collections is that payment plans are gonna be very it's not gonna be like all primes can do this, all seconds can do this, all tech can do that. I think it's gonna be for this segment of customers; we've seen the data tells us that you have to be very flexible. This state, like those types of micro segments that collections is going to get broken down to, just like marketing is right, like it's gonna be microsegments of customers that have shown a behavior act.

Adam Parks (33:33)

Do you think there's any increase in the comfort of consumers to get on payment plans as subscriptions have become more prevalent? Every app they download has got a subscription associated with it. Everything that they watch on their television has got a subscription associated with it.

And we saw the Adobe model change about 2010, right? Subscription: they went from the core software packages that they were selling to a subscription-based model over time. Clearly their stock price has reflected that over the last 16-plus years. But I'm curious: as consumers have become more comfortable, have you seen that have an impact on following through on their payment arrangements?

Kristyn Leffler (34:26)

Yes, and I love this topic. I gave a whole presentation on this topic at the TU conference. I love this. It's like you teed it up for me. Yes, absolutely. I think we aren't thinking about how collections are still anchored to, kinda Mike what you were saying, like, payment in full or let me get a one-time PIF. And there is huge value in getting money while you have money, right? But customers are super comfortable now with recurring subscriptions. So it has to do with the trust factor. It has to do with, I think, some generational impact there. 

The growth of BNPL also makes customers very comfortable. I'm signing up for a recurring charge with an end date. I think that we can tap into that changing customer behavior, and just monitor it, right? Like what's the right payment size? It's no different than, you know, if you go buy a car, they don't tell you what the price is. They tell you, here's what your monthly payment's gonna be, right? Like that's how the game works. And we're yes.

Adam Parks (35:39)

Yeah, no matter how you ask, they'll tell you the payment. Yeah.

Mike Walsh (35:45)

And I also think like, I would say yes, you get more auto pay, or right, you can deduct this. You get a form of payment that you can use over time. I don't think we got that back in the day that often. It's like, no, just I'll send it in on this date. Like, I think that subscription is where you'll have tons and tons of just recurring payments. The break, you'll have a strategy for the breaks; you'll send all your reminders out in any form they want, but I think you get more. But I also think these people are hurting for money. 

And those subscriptions probably got them in a lot of this hack, right? Like they probably have three Hulu subscriptions, right? Like, and they're not managing real well. So I think it's yes and no. Like, debtors are an interesting population.

Kristyn Leffler (36:43)

I think collectors are remiss if they haven't studied a little bit about subscriptions, though, in particular, like frequency. Like what are customers used to? If people like and think about all the different kinds of subscriptions we say, like, Netflix, or like what about meal subscriptions, right? People are getting their HelloFresh box delivered. Like they're very accustomed to different frequencies, different payments, like different bundles, like on the streaming bundle, right? 

The customer, again, is all the same customers, right? Customers are very comfortable and very familiar with this. But it makes pausing and canceling your subscription- like that's another way to increase the commitment. Hey, I feel more comfortable signing up for this recurring subscription because I know I can pause anytime, even though they mostly don't. They can. And that makes you more comfortable to hit the buy button, right? Because like, it reduces the fear of commitment.

Adam Parks (37:46)

That's interesting, right? It is reducing that fear of commitment. You're putting them in a position where they can make a choice at a certain level, which I think is kinda interesting. 

Mike Walsh (37:55)

It's that power of negotiation, right? Like their power is okay; if I get in a bind as I did three months ago when I got in this mess, I can pause it. Pause my payments. So yeah.

Kristyn Leffler (38:07)

Power or perceived power.

Adam Parks (38:10)

Yeah, well, power is where it's perceived to be, right? That's a different book, but the subscription thing, though, is something that I've been looking at pretty closely. I've been studying the dissolution order of consumers, right? So the consumer runs out of money; what do they stop paying first, second, third, fourth, fifth? And so we've been kind of researching that and just trying to better understand it. I do have one last big question that I wanted to throw out there, and it's a little bit more about how.

How are you able to leverage artificial intelligence in this e-commerce experience outside of the orchestration data science that we kind of covered at the beginning? But what other areas are you leveraging data science tools or artificial intelligence tools within that consumer engagement experience?

Kristyn Leffler (39:03)

I think you use it to serve, so like in the absence of a perfect orchestration model, right? Stripe has some amazing tools; if you like, they change dynamically what payment methods are offered, and they do pre-notices, and they can change the checkout flow and change the level of friction based on the fraud risk. Like that's where that's the aspirational goal. Most of us are like not there, right? 

So, but you can use insights like that to say, we're on the Pareto curve, right? So, like, hey, here's this insight I found. All right, I'm gonna go ahead and hard code that because I have found this major insight that is gonna get me significant lift. Is hard-coding the order of payment checkout options like that what we want to be doing? No, but it's better to code the one that is more successful than the less successful one until you get to this amazing dynamic flow at the individual level, right? 

So using the tools, using AI to pull out the insights and then make firm decisions on the insights that you get is still better than flying blind, which I think is where a lot of the industry is still flying blind or basing it on, like, a collector's intuition, which may or may not be accurate. So that's my take.

Mike Walsh (40:21)

No, I think that makes a lot of sense, right? Like you kinda stole my answer, but I also think putting those simple things to test, right? Like what's the best time to call somebody? I worked for an agency; we thought it was lunchtime and right as they got home. And then we realized with cell phones the best time to call someone was now. It didn't matter, right? 

It was totally nothing, and we bought all these different scores, and they all told us different things, but none of it mattered. Like, so I think it's gonna be if you're not using it for that reinforcement learning and that orchestration, and that's kind of, you know, where people like us come in, then I would use it to just test your perceptions that you've had for a long time and make sure they're accurate.

Kristyn Leffler (41:14)

Yeah. Trust but verify.

Mike Walsh (41:16)

Yeah, that's a good book. Because those intuitions, they produce your money. A lot of like the resistance I get when I go into an agency is like, we've been doing this, I don't know if that's gonna work for this. You're not buying a phone call agent or a texting solution. You're buying the brain. And the brain is using all your interactions to decide what is the best next interaction or the best way to collect money. 

So you can do that in segments using just simple AI, like, you know, what have we learned about our small balance segment? What have we learned about our mid balance? Like maybe we go high and low, and we spend a lot of time looking there, but maybe that middle is. I was somewhere I had a client who said it's not moving the E's to the A's, it's moving the D's to the C's. That's you make that's how you make that. Yeah, you know?

Kristyn Leffler (42:05)

Marginal. It's on the margin.

Mike Walsh (42:08)

Yeah. And if you use a little bit of data or look at something differently and use AI to dump your data in a data lake like Adam’s Point and start looking at it. Okay, what are we not thinking of? I think you'll find it. 

Kristyn Leffler (42:21)

Yes. Yes. It's always hard to find a null, right? But that's where I find that to be one of the most helpful prompts. It's like, what haven't I thought of yet? What questions am I missing? And then it's always like, yeah. But if you just ask it like, tell me X you're like, great, that sounds great.

Mike Walsh (42:40)

It's gonna tell you X.

Kristyn Leffler (42:41)

Tells you X.

Adam Parks (42:41)

I am like, ask me 20 questions one at a time, but what you don't understand about my request, and then work my way through that because it's always gonna come up with things that I in my mind imagined that anybody would have understood based on my request versus kind of going through the process.

Kristyn Leffler (42:59)

Yep. What doesn't make sense? What have I forgotten? What other questions would you have? Like the Socratic method. It's not rocket science, but it is important. Otherwise you just get reaffirmation of your existing biases.

Mike Walsh (43:12)

Yeah, what you wanted, what you thought. Right. Yeah.

Adam Parks (43:15)

I think prompt engineering is a great way, or the core fundamentals of prompt engineering is a great way for us to look at the overall technology, its interaction with our other systems, and how we build our ecosystem as an organization between our tech stack partners or pieces or however you're structured.

Well, that's my nerd statement for the day. But you guys have been fantastic. Every time I get on with the two of you, I learn at least a little bit of something, especially you, Kristyn. Your experience on the e-commerce side really does make for some interesting conversation. So I really want to thank you for joining us today.

Kristyn Leffler (43:55)

Of course. It's so fun.

Mike Walsh (43:55)

Yeah, thank you.

Kristyn Leffler (43:57)

I have a great time every time we get to talk shop.

Adam Parks (44:00)

I look forward to the next one. I think Mike and I are gonna have a couple more here coming up in next year that we're gonna wanna go down a similar rabbit hole. So we'll definitely be back in touch as we start exploring some of those opportunities into 2027.

Kristyn Leffler (44:14)

Excellent.

Adam Parks (44:14)

For those of you that are watching, if you have additional questions you'd like to ask Kristyn, Mike, or me, you can leave those in the comments on LinkedIn and YouTube. We'll be responding to those. Or if you have additional topics you'd like to see us discuss, you can leave those in the comments below as well. And I'm willing to bet I can get Kristyn back at least one more time to help us continue to create great content for a great industry. But until next time, Kristyn, we really do appreciate all your insights.

Kristyn Leffler (44:37)

Wonderful. I had a great time. Thank you, guys, for having me.

Mike Walsh (44:40)

Thank you.

Adam Parks (44:41)

Absolutely. And thank you, everybody, for watching. We'll see you all again soon. Bye everyone.