WEBVTT
00:00:00.160 --> 00:00:10.160
What's the point of running these artificial intelligence tools, outbound calls, text messaging, orchestration engines, all of these things if you're just going to fuel it with bad data?
00:00:10.320 --> 00:00:14.240
It's like taking your brand new Ferrari and putting regular gas.
00:00:15.119 --> 00:00:18.079
It's just not going to function the same way.
00:00:18.320 --> 00:00:27.920
And honestly, that's where this topic today kind of came from was a lot of the conversations I've been having in the marketplace about the deployment of artificial intelligence.
00:00:28.239 --> 00:00:36.000
Everybody's talking about the AI pieces, but very few people are talking about the data that actually fuels those opportunities.
00:00:36.399 --> 00:00:37.920
What does that workflow look like?
00:00:38.079 --> 00:00:40.640
How can we be using it specifically for this use case?
00:00:40.719 --> 00:00:52.320
And we've seen a growth of the data appending in research use case as we talk about the six use cases of artificial intelligence specific to the debt collection industry.
00:00:52.560 --> 00:00:53.520
Yeah, absolutely.
00:00:53.600 --> 00:00:59.119
And the one that comes to mind immediately is you know FDCPA third-party disclosure.
00:00:59.359 --> 00:01:03.840
You know, you can't communicate about a debt as you know to somebody to the wrong consumer.
00:01:03.920 --> 00:01:16.879
And so if your contact information is wrong or your phone number's been reassigned to another consumer, you're communicating about a debt, you know, that's inviting risk, you know, regulatory risk, financial risks, litigation risk, reputation risk.
00:01:17.040 --> 00:01:17.200
Yeah.
00:01:17.280 --> 00:01:22.480
So beyond the performance side of it, that data can lead to other consequences.
00:01:22.640 --> 00:01:24.640
And that's where the conversation usually starts.
00:01:24.719 --> 00:01:34.480
And then, you know, based on that, we start analyzing, you know, whether their data processes and strategies, the quality of their information are aligned with those goals.
00:01:34.560 --> 00:01:43.760
And you know, typically there's an opportunity to improve whatever they're doing today to further optimize that desired performance outcome.
00:01:43.840 --> 00:01:45.359
And that's that becomes the measurement.
00:01:45.439 --> 00:01:52.799
And then we build a strategy, a data strategy around that, typically using their internal data as well as third-party data providers.
00:01:52.879 --> 00:01:58.079
And then we do we run an A-B test and we kind of look at what they're doing today, and that's the champion strategy.
00:01:58.159 --> 00:02:03.840
And then we come up with you know challenger strategies that are designed to outperform that champion strategy.
00:02:04.000 --> 00:02:16.639
Um, you know, and typically, regardless of kind of how mature a customer is in their data strategy, you know, typically when we're running an A-B test, we're looking for a meaningful lift.
00:02:16.879 --> 00:02:18.159
We're looking to generate a meaningful lift.
00:02:18.240 --> 00:02:24.400
By meaningful, I mean, you know, 20% or higher, you know, sort of increase in performance.
00:02:30.719 --> 00:02:35.680
Hello everybody, Adam Parks here with another episode of Receivables Podcast.
00:02:35.840 --> 00:02:39.199
Today I'm here with a returning guest, my good friend, Mr.
00:02:39.280 --> 00:02:42.800
Will Turner, joining me from TEC Services.
00:02:43.039 --> 00:02:48.319
Very excited to have him here today because just like me, he is a deep data nerd.
00:02:48.479 --> 00:02:57.439
He loves to look at the information behind the decisions being made and try to understand why or how decisions can improve operationally.
00:02:57.599 --> 00:03:00.319
That's what makes today's discussion very exciting.
00:03:00.479 --> 00:03:06.319
The last time we talked, we talked about data waterfalls, building your waterfall, testing your waterfall.
00:03:06.479 --> 00:03:15.520
And today I want to take this conversation a little bit further, and I want to talk about the real cost of bad data in your organization.
00:03:15.840 --> 00:03:19.919
But before we get to that, Will, thank you so much for joining me today.
00:03:20.080 --> 00:03:24.800
I really do appreciate your time and attention and coming on here, sharing your insights.
00:03:25.039 --> 00:03:25.759
Hey, thanks, Adam.
00:03:25.840 --> 00:03:26.879
Thanks for having me back.
00:03:27.120 --> 00:03:28.240
Just a quick introduction.
00:03:28.319 --> 00:03:29.439
My name is Will Turner.
00:03:29.520 --> 00:03:30.960
I'm a senior data consultant.
00:03:31.120 --> 00:03:42.080
I've spent roughly 30 years working for large data companies, consulting for data companies around the topic of consumer data, helping clients solve business problems with data and analytics and now AI.
00:03:42.319 --> 00:03:45.599
Well, and you've done a great job at that at a lot of different organizations.
00:03:45.680 --> 00:04:04.560
And I think your consulting experience is one of those things that really gets me excited about these types of conversations, because the depth of experience and understanding in different situations, scenarios, and circumstances really does change from a consultant's lens because you get access to so many different types of challenges.
00:04:04.800 --> 00:04:14.560
So, Will, as we kind of kick off this conversation today, when you think about the cost of bad debt for a debt collection operation, what's that first thing that comes to mind?
00:04:14.639 --> 00:04:17.040
How do you frame that in your own mind?
00:04:17.360 --> 00:04:18.639
Yeah, great question, Adam.
00:04:18.800 --> 00:04:25.360
So when I think about the cost of bad data and debt collections, the first thing that comes to mind is performance.
00:04:25.519 --> 00:04:31.360
You know, whether you're a lender, a debt buyer, or a servicer, you're trying to recover money, of course.
00:04:31.600 --> 00:04:37.519
And you know, bad data across that whole delinquent account lifecycle is a barrier to collecting money.
00:04:37.759 --> 00:04:42.879
And so, you know, from a performance standpoint, if you're a servicer, that's revenue and profitability.
00:04:43.120 --> 00:04:46.399
If you're a lender, you know, your losses are more significant.
00:04:46.560 --> 00:04:48.560
Yeah, that's the first thing that comes to mind is performance.
00:04:48.879 --> 00:04:55.120
Those performance issues is so you have this cost of not being able to optimize that performance.
00:04:55.199 --> 00:04:59.439
So you're going to have a lag to your competition from a timing perspective.
00:04:59.600 --> 00:05:02.480
You may not hit the same numbers overall.
00:05:03.040 --> 00:05:09.120
But when we think about bad data, what is some of the data that is going stale the fastest?
00:05:09.279 --> 00:05:17.680
And where should people be looking within their data workflows to say this might be the first place that I should look to identify the problem?
00:05:18.319 --> 00:05:18.720
Yeah.
00:05:19.040 --> 00:05:25.439
So as far as um, I think of things like addresses, about 37 million people move every year.
00:05:25.680 --> 00:05:30.720
A small percentage of those are posted with the United States Postal Service as an ON file.
00:05:30.959 --> 00:05:36.879
I think about there's a large volume of reassigned numbers that prevents risk and performance issues.
00:05:37.040 --> 00:05:42.480
I think about email addresses, not necessarily that email addresses go bad, but they go unused.
00:05:42.720 --> 00:05:49.360
And you know, typically consumers have multiple email addresses, so you know, having the right email address that the consumer is actually using.
00:05:49.519 --> 00:05:56.319
Those are things that come to mind in in the in the legal collections market, of course, you know, place of employment, people switch jobs.
00:05:56.480 --> 00:06:04.079
Yeah, so the the topic of bad data is pretty broad, but as it relates to the contact information, those are some of the data elements I think about.
00:06:04.399 --> 00:06:08.240
Well, let's think about it from the perspective of different disciplines of debt.
00:06:08.319 --> 00:06:21.680
So from a collection agency standpoint, it sounds like the biggest cost of bad data would be directly related to that contact information, the phone numbers, the emails, potentially addresses, whatever you're gonna need for that output.
00:06:21.759 --> 00:06:35.360
And it's not just the cost of not finding that person, it's not just the opportunity cost, but also there's got to be some other risks that are coming along with sending the wrong communication or communicating with the wrong people.
00:06:36.079 --> 00:06:37.120
Yeah, absolutely.
00:06:37.279 --> 00:06:42.800
And the one that comes to mind immediately is you know, FTCPA third-party disclosure.
00:06:43.040 --> 00:06:47.519
You know, you can't communicate about a debt as you know to somebody to the wrong consumer.
00:06:47.600 --> 00:07:00.480
And so if your contact information is wrong or your phone number's been reassigned to another consumer and you're communicating about a debt, you know, that's inviting risk, you know, regulatory risk, financial risk, litigation risk, reputation risk.
00:07:00.639 --> 00:07:00.879
Yes.
00:07:00.959 --> 00:07:05.839
So beyond the performance side of it, that data can lead to other consequences.
00:07:06.240 --> 00:07:15.040
And the cost of being wrong at scale is we've started to apply artificial intelligence, because now we're getting to a point where we can send more customized messages, right?
00:07:15.279 --> 00:07:22.000
We can do things at a larger volume and scale, but with that scale comes additional risk as well.
00:07:22.240 --> 00:07:34.000
Because if you're going to be wrong, if you're going to communicate to the wrong phone number, if you're going to send out emails to the wrong addresses, you're not just going to be wrong one or two accounts, you're going to be wrong at scale.
00:07:34.240 --> 00:07:47.040
How do you how do you kind of look at that risk tolerance and how does that go into the decision-making process when a collection agency is looking at how they might want to address the bad debt in their organization?
00:07:47.439 --> 00:07:49.120
Yeah, I totally agree, Adam.
00:07:49.360 --> 00:08:10.639
As far as when you when we think about using AI, especially virtual agents, or we think about digital collections, digital communications, comparing it to the voice channel and live agents, you're right, bad data amplifies this problem at scale because you're throwing typically a higher volume of activity, especially when it when it comes to outbound communications.
00:08:10.879 --> 00:08:38.320
And so from a consulting standpoint, when companies are migrating away from the voice channel or augmenting their voice strategies with AI virtual agents or digital communications, you know, part of it is helping the customer understand those risks, evaluating their portfolio and the accuracy of their information to identify those risks, and then coming up with a risk mitigation plan so that they step into those new channels carefully.
00:08:38.639 --> 00:08:51.120
Now, if we think about the same problem from the perspective of a law firm or a debt buyer who's purchasing accounts for litigation, the cost or the potential cost here seems to amplify even more so.
00:08:51.279 --> 00:09:09.200
Because if I'm a debt buyer or a law firm who's using that data to make an operational decision about which accounts belong in which operational channel, that could be a massive cost if I were to move the wrong accounts into the litigation channel.
00:09:09.360 --> 00:09:15.360
All those core costs, all those expenses, process service, a very expensive channel to operate in.
00:09:15.519 --> 00:09:28.639
So it seems like there's even more risk there, but potentially even more opportunity if you get your data waterfalls working correctly and you are actively moving the right data over your accounts.
00:09:29.039 --> 00:09:29.919
I agree, Adam.
00:09:30.080 --> 00:09:37.759
And that's something I think a lot of people overlook is that you know, you know, on brand new accounts, forget contacting the consumer.
00:09:37.919 --> 00:09:44.639
When we think about you know, scoring the account and segmenting the portfolio, the quality of the data.
00:09:44.960 --> 00:10:05.039
A good example would be, you know, having the correct name and address, as simple as that, impacts a third-party vendor's ability to score that account correctly, which leads to a lot of decisions from the very beginning, right, in terms of segmentation and which flow this is which which collection path this account is going to go down.
00:10:06.559 --> 00:10:07.360
Which is interesting.
00:10:07.519 --> 00:10:11.840
You know, we're we're finding ourselves in a situation where we're getting this increase of accounts.
00:10:11.919 --> 00:10:15.279
And I believe it was from the TransUn report in 2025.
00:10:15.360 --> 00:10:22.879
We said that 73% of firms were expecting account volumes to increase in 2026, up from 52% the year before.
00:10:23.200 --> 00:10:32.720
At the same time, we're seeing a stagnation or decline in liquidation of those same accounts, meaning we're having to do more with less.
00:10:32.879 --> 00:10:49.120
And being able to leverage data to make the impact of each action more powerful from a collections perspective seems to be the wave of the future, or at least the the end goal for a lot of organizations right now.
00:10:49.360 --> 00:11:02.720
And as a consultant who's helping to build these waterfalls, what kind of actions are you taking to help organizations down that path of exploration and understanding of how to unlock more value from that data?
00:11:03.120 --> 00:11:03.360
Yeah.
00:11:03.440 --> 00:11:05.919
So, Adam, this is the most fun part of my job.
00:11:06.080 --> 00:11:08.720
Like it is excited.
00:11:08.799 --> 00:11:26.480
That's why that's why I'm using data, um, helping companies use data in a in a more optimized fashion to make a measurable and meaningful performance impact is kind of at the core of what I do as a as a consultant.
00:11:26.720 --> 00:11:34.559
The steps that we go through to make that happen, I guess, you know, starts with a conversation about what's important to the customer in terms in terms of desired performance outcomes.
00:11:34.720 --> 00:11:39.759
And that might be very different by client or portfolio, or in the case of the lender by line of business.
00:11:39.919 --> 00:11:41.840
And that's where the conversation usually starts.
00:11:41.919 --> 00:11:51.679
And then, you know, based on that, we start analyzing whether their data processes and strategies, the quality of their information are aligned with those goals.
00:11:51.840 --> 00:12:00.960
And you know, typically there's an opportunity to improve whatever they're doing today to further optimize that desired performance outcome.
00:12:01.039 --> 00:12:10.000
And that becomes the measurement, and then we build a strategy, a data strategy around that, typically using their internal data as well as third-party data providers.
00:12:10.080 --> 00:12:21.039
And then we do we run an A-B test and we kind of look at what they're doing today, and that's the champion strategy, and then we come up with you know challenger strategies that are designed to outperform that champion strategy.
00:12:21.200 --> 00:12:33.840
Um, you know, and typically, regardless of kind of how mature a customer is in their data strategy, you know, typically when we're running an A-B test, we're looking for a meaningful lift.
00:12:34.080 --> 00:12:35.440
We're looking to generate a meaningful lift.
00:12:35.519 --> 00:12:40.879
By meaningful, I mean, you know, 20% or higher, you know, sort of increase in performance.
00:12:41.039 --> 00:12:51.120
And again, regardless of how good or bad somebody is today or how mature they are with their data strategies, there's always room, almost always room to opt further optimize it.
00:12:51.360 --> 00:12:52.960
There's always room to further optimize it.
00:12:53.039 --> 00:13:06.080
Whether or not you can further optimize to 20% each and every time, I think is a, you know, that that's a it's a big goal, but it's definitely a doable thing because the impact that data can have ripples.
00:13:06.240 --> 00:13:10.879
And when you figure out that process and you've got right, it's like finding the right recipe.
00:13:10.960 --> 00:13:12.720
You're gonna make the best pasta here.
00:13:12.799 --> 00:13:20.639
And when you when you know which ingredients have to go in, at which measurements, and in which order, it changes the way that that dish is gonna turn out.
00:13:20.720 --> 00:13:33.919
And I think if you can continue to tweak and understand that, what kind of testing programs do you build for that to clean out the bad data and append with more powerful, accurate information and data?
00:13:34.399 --> 00:13:36.639
Yeah, so it kind of depends on the data type.
00:13:36.720 --> 00:13:49.440
Um, you know, we're doing a lot right now with debt buyers and law firms around you know, address waterfalls, address hygiene, um, appending good addresses, just ensuring that they have a good address to reduce retard bail and increase process serving rates.
00:13:49.519 --> 00:14:00.480
And so, you know, for that, we look at you know, how well a company is cleansing and validating their addresses, how accurate is the information, what does their append strategy look like if they have one?
00:14:00.960 --> 00:14:07.120
Typically, we introduce some sort of address scoring component, which is very similar, you know, for for phones.
00:14:07.440 --> 00:14:19.279
I guess the methodology is pretty similar overall, how we evaluate how well somebody's current strategy is for whatever problem they're trying to solve.
00:14:19.519 --> 00:14:31.039
And then the methodology we use for testing kind of depends a little bit about some of it is how well they can maintain control of the test.
00:14:31.360 --> 00:14:34.720
I mean, we do all the the testing methodologies are pretty broad, Adam.
00:14:34.960 --> 00:14:43.440
So you know, we do everything from you know sending the same accounts to multiple third-party data providers, that all the data providers get the same accounts.
00:14:43.600 --> 00:14:56.240
We do a lot of champion challenger production testing where we'll take a portion of a client's inventory and we'll always have them running through sort of challenger strategies, multi multiple data providers, you know, in in a sequence.
00:14:56.480 --> 00:15:10.080
Sometimes we're testing with you know the the bad data, the the information that the customer knows is wrong, and then we'll demonstrate using third-party data providers the lift that we can provide with information that that they know to be bad.
00:15:10.480 --> 00:15:12.960
We'll take fresh portfolios, we'll take age portfolios.
00:15:13.120 --> 00:15:27.120
One of the one of the things that I find interesting is that when it comes to bad data, and we think about debt collections, the older that account gets, in some ways, the value of that bad data amplifies.
00:15:27.360 --> 00:15:53.360
And what I mean by that is you know, you take a take an account that went from a creditor and to a first party agency and it and it now becomes you know a tertiary or quad account, the the the bucket of bad data increases as those accounts age, which is you know, if that problem is resolved, it represents an opportunity for somebody to collect that money.
00:15:53.519 --> 00:16:08.559
So I find you know the the strategies that we deploy can sometimes depend on how bad the data problem is, and you know, the age of the account sometimes, you know, it often amplifies as it as the account ages.
00:16:09.039 --> 00:16:27.759
Well, as that account is going through the aging process and you've got more of that data that's decaying over time, I would think four, five, six years down the road, the data's not great, but the quality of the consumer is probably improved if it's going to improve over that time block.
00:16:27.919 --> 00:16:36.960
So the opportunity to re-engage, I think, is very real, but the cost of not having good contact data there or not appending.
00:16:37.120 --> 00:16:42.399
Let me ask you this when you think about new data that's coming over to a, let's say, a third-party agency.
00:16:42.480 --> 00:16:56.480
So they're receiving the data from the creditor, how well is the understanding of that data quality managed and communicated in that early onboarding process?
00:16:56.639 --> 00:17:04.000
Meaning, if I'm an agency who's getting accounts from a creditor, how well am I going to truly understand what that account has been through previously?
00:17:04.160 --> 00:17:11.359
And how much of that do I have to learn through experimentation and testing versus communication?
00:17:11.680 --> 00:17:11.839
Yeah.
00:17:12.000 --> 00:17:24.480
So I I think part of what you're hitting on there, Adam, is you know, what's the what's the communication needed between the servicer and the forwarder of that debt in terms of the history of that account.
00:17:24.880 --> 00:17:26.079
Especially in the early stages.
00:17:26.319 --> 00:17:32.319
Well, the hit not necessarily even just the history of the account, but the history of the data associated to that account.
00:17:32.559 --> 00:17:33.200
Yeah.
00:17:33.680 --> 00:17:34.160
Yeah.
00:17:34.480 --> 00:17:55.200
So so obviously that's that's key for uh you know when you're bidding on a portfolio to understand that when you're thinking about your contingency fees and how how rates are set compared to how how hard a portfolio is going to be to work, and how are you gonna likely perform and what investments and data are you gonna need to make in in order to perform.
00:17:55.680 --> 00:18:09.440
One of the things I often see when people get accounts, you know, typically a service or has, and lenders too, have some decent operational metrics to know you know things like right party contact rates and conversion rates and liquidation rates and recovery rates.
00:18:09.920 --> 00:18:23.920
I think sometimes people get the sense that if your right party contact rates and your conversion rates and your liquidation rates are what you expect them to be, that you don't have a bad data problem.
00:18:24.160 --> 00:18:35.039
And what we try to expose is what could your right party contact rates, your liquidation or your rates be if you further optimized the the data?
00:18:35.200 --> 00:18:42.720
And so I do think there's a false sense of confidence oftentimes that, hey, my data's good, my data strategies are good.
00:18:43.039 --> 00:19:03.359
And and typically for somebody that's satisfied with their current performance, I don't know that we can you know help them, but people that are are open-minded to saying, hey, we're doing things really well today, we're performing, we're winning, you know, we're top of a lot of scorecards, but but we want to know what we could be, or we want to stay ahead of the curve.
00:19:03.440 --> 00:19:06.240
We want to maintain our our forward position.
00:19:06.400 --> 00:19:09.759
Um you know there's a lot of opportunity there.
00:19:09.839 --> 00:19:12.079
I think there's a lot of sort of overconfidence.
00:19:12.400 --> 00:19:24.319
I I agree with that statement because you're only as good as you think you are right now, and your where you can go to is never determined if you're only operating on that same platform.
00:19:24.400 --> 00:19:40.799
So if you're doing everything exactly the same way, yeah, hey, you're not gonna necessarily get worse results, and you might investigate if you got results worse than expected, or you might investigate if they were better, but just because it's meeting your expectation doesn't mean your expectation is correct.
00:19:41.039 --> 00:19:43.279
Where how has your expectation been level set it?
00:19:43.359 --> 00:19:44.720
Is it through a champion challenger?
00:19:44.799 --> 00:19:45.920
Is it through a scorecard?
00:19:46.000 --> 00:19:48.880
You know, how how are you level setting your expectations?
00:19:49.119 --> 00:19:51.759
How do you suggest a client level set their expectations?
00:19:51.920 --> 00:19:53.519
I think is a good question for you, right?
00:19:53.599 --> 00:19:56.559
Like you see all these different situations and scenarios.
00:19:56.720 --> 00:20:02.720
How do you help someone to understand what level setting really means from a performance perspective?
00:20:03.039 --> 00:20:12.319
So as you as you were talking, Adam, I was thinking some one of some of the characteristics over the years that I've noticed about top performers, there's two things that stand out for me.
00:20:12.480 --> 00:20:14.640
One is there's some level of humility.
00:20:14.799 --> 00:20:19.119
They may just be super good and super competent, but there's some level of humility.
00:20:19.279 --> 00:20:23.759
And second, there's this mindset of ongoing improvement.
00:20:24.000 --> 00:20:35.680
And so one of one of the ways that we level set um it is to, you know, evaluate how somebody is performing today and you know, can can we get a lift?
00:20:35.759 --> 00:20:37.519
You know, is is there room for improvement?
00:20:37.680 --> 00:20:50.079
And so I don't I don't think like for for me when it comes to optimizing data strategies, it isn't like you arrive at something that's the the best and then you walk away from it.
00:20:50.240 --> 00:20:54.880
It's sort of this ongoing thing that you're evolving over time.
00:20:55.119 --> 00:21:05.920
And so I again I think the characteristics of top performers to me are people that they're not really satisfied, they're humble enough to know that what they're doing to today could be better tomorrow.
00:21:06.079 --> 00:21:09.519
And there is this sort of mindset of ongoing performance improvement.
00:21:09.599 --> 00:21:22.079
And so a lot of times we'll walk in, you know, people haven't touched their data strategies or haven't really evaluated their data strategies in years, or they've had the data strategy in place for years.
00:21:22.319 --> 00:21:26.640
Um and you know, typically that's almost always that's suboptimal.
00:21:26.720 --> 00:21:29.599
You know, it's something that we're optimizing over time.
00:21:29.759 --> 00:21:35.440
And again, with with top performers, we typically have a portion of their inventory always running in champion challenger test.
00:21:35.680 --> 00:21:40.079
So it's not like something that you arrive at and walk away from.
00:21:40.480 --> 00:21:54.480
Well, success is the first step in the journey of a thousand miles, and each success starts another new journey of a thousand miles that requires not only for you to get there, but for you to maintain along the way, which you know brings in more challenges.
00:21:54.720 --> 00:22:04.880
Now, we're living in a time right now where system conversions are happening at a pace that I've I've don't ever remember seeing before, at least not since like the early 2000s.
00:22:05.039 --> 00:22:16.960
And with all of this bad data getting locked in and all this bad bad data being collected, how do organizations start looking at that cleansing process at scale?
00:22:17.119 --> 00:22:21.920
Are organizations only looking at those new accounts that are coming in and trying to improve for the future?
00:22:22.079 --> 00:22:34.319
Are organizations trying to better understand what they could strip away from their current storage in order to reduce overhead, reduce cyber liability costs, other things?
00:22:34.559 --> 00:22:40.720
Yeah, how do you think organizations are looking at that ongoing data cleansing and management process?
00:22:41.119 --> 00:22:45.039
Yeah, so um, so I'm not an expert in all of those particular areas.
00:22:45.200 --> 00:22:49.519
You know, as a company, we manage a lot of data conversions and help companies through that transition.
00:22:49.839 --> 00:23:02.480
For me personally, the piece of it that is kind of most relevant for me and how I help clients is you know, during a data conversion is a perfect opportunity to look at at their data and optimize that data.
00:23:02.559 --> 00:23:06.640
It's for one, it's gonna make the data conversion to a new platform much smoother.
00:23:06.799 --> 00:23:17.440
And then all the technology and scoring products and the people and the processes that you add to that, the performance of that, it's gonna run more smoothly, it's gonna perform at a much higher, higher level.
00:23:17.599 --> 00:23:24.559
And so certainly when it comes to data conversions, uh, that's a perfect time to look at this and evaluate your data.
00:23:24.799 --> 00:23:35.680
And then certainly from a cybersecurity standpoint, you know, there's there's data, there's probably data there that doesn't need to be stored indefinitely, which you know presents, you know, creates risk, right?
00:23:36.319 --> 00:23:40.319
Creates risk and opportunity to reduce that expenditure.
00:23:40.559 --> 00:23:56.559
But let me go back to kind of the last piece of of my first question, which was more around as you talk to an organization about building a solution for data cleansing into the future, are they more focused on let me clean up what's coming in now?
00:23:56.720 --> 00:24:05.039
And from today forward, we're gonna do it this way, or people looking at trying to manage that entire ball of wax?
00:24:05.440 --> 00:24:08.079
How are people addressing the challenge?
00:24:09.039 --> 00:24:17.039
So if I understand the correct question correctly, Adam, I I see it more there's a specific problem that they're trying to solve.
00:24:17.279 --> 00:24:24.000
So, hey Will, we have too many, too many accounts and a high percentage of our accounts are are are still in Skip.
00:24:24.319 --> 00:24:28.079
That that's an opportunity um that we're not currently capitalizing on.
00:24:28.319 --> 00:24:31.440
Or, you know, hey, Will, our are for legal collections.
00:24:31.759 --> 00:24:38.960
I see that on client scorecards we have competing law firms that are achieving higher serve rates than than we are.
00:24:39.119 --> 00:24:44.960
And so oftentimes, you know, it it presents itself as a you know a specific performance problem.
00:24:45.119 --> 00:24:47.839
And it may not present itself that way.
00:24:47.920 --> 00:24:54.160
It also could be, you know, hey Will, we're our our liquidation rates are really strong, but our margins have compressed.
00:24:54.319 --> 00:24:59.279
And you know, we we have a we have a profit problem that that we need to solve.
00:24:59.440 --> 00:25:06.079
So how can we maintain our level of performance but fix that profitability problem?
00:25:07.119 --> 00:25:18.079
So you see the you see it manifesting in specific instances or challenges that they're trying to solve, and the end result may be a deeper cleanse of the data.
00:25:18.240 --> 00:25:26.400
Do you find that they need to taste a little piece of success before they're willing to go deeper into that data cleansing investment?
00:25:27.119 --> 00:25:36.799
Yeah, I uh I hope there are companies out there that are just saying, hey, I should be looking at you know data cleansing across my portfolio for lots of different use cases and reasons.
00:25:37.519 --> 00:25:39.039
I hope that that's the case.
00:25:39.279 --> 00:25:45.519
For me personally, it again it's it's presented itself more in um you know specific challenges.
00:25:45.599 --> 00:25:51.039
And you know, to answer your question, yeah, I think again, I think there's a little bit of overconfidence.
00:25:51.279 --> 00:26:10.079
And so when we can demonstrate the impact that can be made by cleansing your data, adjusting your data strategies, and somebody sees the financial and performance impact of that, certainly that's motivates them to say, hey, wow, you did that, you know, you help us do that.
00:26:10.160 --> 00:26:12.480
Now let's look at this might be the right investment for me.
00:26:12.720 --> 00:26:13.119
Yeah, yeah.
00:26:13.359 --> 00:26:14.000
Fair exactly.
00:26:14.160 --> 00:26:14.319
Yeah.
00:26:14.559 --> 00:26:22.079
So, well, are there any big misconceptions around dirty data in a debt collection organization?
00:26:22.240 --> 00:26:28.319
Is there anything else that we haven't talked about today that you want to make sure that our audience understands about bad data?
00:26:29.039 --> 00:26:38.880
So one one that comes to mind, so and often oftentimes with medical debt, I hear the misconception that you know we can't skip trace.
00:26:39.200 --> 00:26:44.720
We only use the contact information that comes over from um you know our client.
00:26:44.880 --> 00:26:47.599
And that that's a that's a common misconception.
00:26:47.680 --> 00:27:00.000
And and certainly when we work with servicers of medical debt that are high performers, uh they're definitely using third-party data uh to validate, to score, to append, to skip trace.
00:27:00.240 --> 00:27:12.319
And so, you know, medical's a big asset class in our industry, as you know, and I think there's a a lot of of opportunity there of just because of some of those misconceptions.
00:27:12.480 --> 00:27:19.680
And again, I I work with a lot of smart people and and customers, so not not not diminishing any of that.
00:27:19.920 --> 00:27:31.680
I don't know if everybody, um, I don't know if it's commonly known kind of the impact that data has on the investments that companies make in their people and in their technology.
00:27:31.839 --> 00:27:41.200
And so when I think about you know different technology initiatives, you know, there's an expected ROI, there's expected performance when they're making those investments.
00:27:41.359 --> 00:27:55.599
And sometimes I think it's overlooked the data side of that and what the ROI and the performance impact could be on those investments that they're making if data was a bigger priority or a bigger part of that conversation.
00:27:55.759 --> 00:28:00.880
And so that's something I see across the board, Adam, across all types of customers.
00:28:01.119 --> 00:28:03.119
Yeah, I think it's a great observation.
00:28:03.440 --> 00:28:10.400
I think the conversation has only started to come to the forefront because people have gotten excited about artificial intelligence.
00:28:10.640 --> 00:28:15.039
And now they're realizing they need the data to be available, right?
00:28:15.200 --> 00:28:22.000
They need to combine these things, they need to make it available to a single tool or location, and there's some data cleansing that goes into that.
00:28:22.079 --> 00:28:34.319
And that was really the introduction of that technology, let's call it over the last two years, I think has been that driving force that's pushing people towards optimizing the data.
00:28:34.640 --> 00:28:45.039
What's the point of running these artificial intelligence tools, outbound calls, text messaging, orchestration engines, all of these things if you're just gonna fuel it with bad data?
00:28:45.119 --> 00:28:49.680
It's like taking your brand new Ferrari and putting regular gas in it.
00:28:49.920 --> 00:28:52.960
It's just not gonna function the same way.
00:28:53.119 --> 00:29:02.799
And honestly, that's where this topic today kind of came from was a lot of the conversations I've been having in the marketplace about the deployment of artificial intelligence.
00:29:02.880 --> 00:29:10.880
And everybody's talking about the AI pieces, but very few people are talking about the data that actually fuels those opportunities.
00:29:11.119 --> 00:29:12.720
What does that workflow look like?
00:29:12.880 --> 00:29:15.440
How can we be using it specifically for this use case?
00:29:15.599 --> 00:29:27.279
And we've seen a growth of the data appending in research use case as we talk about the six use cases of artificial intelligence specific to the debt collection industry.
00:29:27.440 --> 00:29:28.960
And I think this is just the beginning.
00:29:29.119 --> 00:29:36.880
I think that's where we're going to see a lot more investment over the coming two or three years as organizations get the rails set up.
00:29:36.960 --> 00:29:41.920
And I look at it similar to the way that we saw the communication change over time.
00:29:42.240 --> 00:29:45.359
Setting up the rails to send a text message is one thing.
00:29:45.599 --> 00:29:52.000
Setting a strategy for content is another, setting an orchestration amongst all of the channels is yet still another thing.
00:29:52.240 --> 00:29:56.160
And I think the industry is going to phase its way into that approach.
00:29:56.319 --> 00:29:59.440
First, it's going to want to get AI installed.
00:29:59.680 --> 00:30:25.680
And then it's going to want to start taking it, and then it's baby steps going forward as we continue to grow as an industry and as we continue to test and understand not only the impact from a consumer standpoint, but who knows, in the next months or years, we may have the CFPB coming back, and who knows how they're going to view some of these tools, some of these processes, and some of our responsibilities from a data cleansing perspective into the future.
00:30:26.000 --> 00:30:26.880
Yeah, 100%.
00:30:27.359 --> 00:30:36.720
AI is opening up opportunities as it relates to data cleansing, identifying problems, resolving problems within a customer CRM.
00:30:36.880 --> 00:30:38.960
That's a piece of what you're talking about, right, Adam?
00:30:39.359 --> 00:30:41.759
It's just a touch of all of those things.
00:30:41.839 --> 00:31:06.319
And I think as we look at being able to cleanse that data, whether it be in our sales CRM, in our customer management databases, but our ability to do fuzzy matching logic and to do other things that allow us to better manipulate data at scale more reliably and more accurately, is where I see us spending a lot of our time and money in the coming two to three years.
00:31:06.640 --> 00:31:07.200
Yeah.
00:31:07.440 --> 00:31:18.799
So Adam, I want to go back to a question that you asked about the process in helping customers identify, you know, a problem or an opportunity to increase performance.
00:31:18.960 --> 00:31:24.640
And typically we'll start with a call, a discovery call to learn about a company's business.
00:31:24.799 --> 00:31:34.960
We'll set up a sort of data assessment where after we know a little bit about customers' business and their goals and the types of debt that they work, the services that they provide, we'll walk through their existing data strategies.
00:31:35.039 --> 00:31:38.799
And part of that is understanding their reporting and their KPIs.
00:31:38.960 --> 00:31:48.559
And so when we when we look at somebody's email strategy, for example, well, we know what's possible in terms of email delivery rates.
00:31:48.720 --> 00:31:53.440
We know what's possible in terms of um uh you know bounce rates.
00:31:53.519 --> 00:32:00.240
And you know, so so that becomes, you know, hey, you know, I know that you're performing at this rate today.
00:32:00.480 --> 00:32:06.079
We try to help clients using data achieve these kind of performance levels.
00:32:06.319 --> 00:32:07.359
We think we can do that.
00:32:07.440 --> 00:32:09.200
We could we think we can demonstrate that for you.
00:32:09.279 --> 00:32:11.039
Let's do a pilot and prove it out.
00:32:11.119 --> 00:32:13.119
And so that that's part of the process as well.
00:32:13.279 --> 00:32:19.759
It's just you kind of understand after working with a number of clients on certain strategies, you know, what's possible.
00:32:19.920 --> 00:32:24.720
And oftentimes people are not performing at that level, and we see an opportunity.
00:32:24.880 --> 00:32:28.960
So well, the opportunities are definitely abundant in the upcoming future.
00:32:29.039 --> 00:32:31.119
And Will, I could talk to you for hours, buddy.
00:32:31.279 --> 00:32:38.720
Every single conversation we have, whether it be on a podcast, walking through the halls at a conference, I learned something from every discussion.
00:32:38.880 --> 00:32:43.200
I value your friendship and I really appreciate you coming on here and sharing insights today.
00:32:43.519 --> 00:32:45.039
Yeah, appreciate the opportunity, Yan.
00:32:45.200 --> 00:32:45.440
Thank you.
00:32:45.599 --> 00:32:46.000
Absolutely.
00:32:46.079 --> 00:32:53.759
Will I really appreciate you for those of you that are watching, and I'm sure that you have some questions here for Will, you can leave those in the comments on LinkedIn and YouTube, and we'll be responding to those.
00:32:54.000 --> 00:32:58.079
Or if you have additional topics you'd like to see us discuss, you can leave those in the comments below as well.
00:32:58.160 --> 00:33:05.039
And I'm willing to bet I can get Will back here at least one more time to help me continue to create great content for a great industry.
00:33:05.200 --> 00:33:07.920
I know we will have no shortage of topics to discuss.
00:33:08.079 --> 00:33:09.200
But Will, thank you so much.
00:33:09.279 --> 00:33:10.160
I really appreciate you.
00:33:10.240 --> 00:33:11.200
Thank you for coming on.
00:33:11.440 --> 00:33:12.720
Thanks again, Adam.
00:33:12.960 --> 00:33:14.559
And thank you everybody for watching.
00:33:14.640 --> 00:33:16.160
We appreciate your time and attention.
00:33:16.240 --> 00:33:17.839
We'll see y'all again soon.
00:00:00.160 --> 00:00:10.160
What's the point of running these artificial intelligence tools, outbound calls, text messaging, orchestration engines, all of these things if you're just going to fuel it with bad data?
00:00:10.320 --> 00:00:14.240
It's like taking your brand new Ferrari and putting regular gas.
00:00:15.119 --> 00:00:18.079
It's just not going to function the same way.
00:00:18.320 --> 00:00:27.920
And honestly, that's where this topic today kind of came from was a lot of the conversations I've been having in the marketplace about the deployment of artificial intelligence.
00:00:28.239 --> 00:00:36.000
Everybody's talking about the AI pieces, but very few people are talking about the data that actually fuels those opportunities.
00:00:36.399 --> 00:00:37.920
What does that workflow look like?
00:00:38.079 --> 00:00:40.640
How can we be using it specifically for this use case?
00:00:40.719 --> 00:00:52.320
And we've seen a growth of the data appending in research use case as we talk about the six use cases of artificial intelligence specific to the debt collection industry.
00:00:52.560 --> 00:00:53.520
Yeah, absolutely.
00:00:53.600 --> 00:00:59.119
And the one that comes to mind immediately is you know FDCPA third-party disclosure.
00:00:59.359 --> 00:01:03.840
You know, you can't communicate about a debt as you know to somebody to the wrong consumer.
00:01:03.920 --> 00:01:16.879
And so if your contact information is wrong or your phone number's been reassigned to another consumer, you're communicating about a debt, you know, that's inviting risk, you know, regulatory risk, financial risks, litigation risk, reputation risk.
00:01:17.040 --> 00:01:17.200
Yeah.
00:01:17.280 --> 00:01:22.480
So beyond the performance side of it, that data can lead to other consequences.
00:01:22.640 --> 00:01:24.640
And that's where the conversation usually starts.
00:01:24.719 --> 00:01:34.480
And then, you know, based on that, we start analyzing, you know, whether their data processes and strategies, the quality of their information are aligned with those goals.
00:01:34.560 --> 00:01:43.760
And you know, typically there's an opportunity to improve whatever they're doing today to further optimize that desired performance outcome.
00:01:43.840 --> 00:01:45.359
And that's that becomes the measurement.
00:01:45.439 --> 00:01:52.799
And then we build a strategy, a data strategy around that, typically using their internal data as well as third-party data providers.
00:01:52.879 --> 00:01:58.079
And then we do we run an A-B test and we kind of look at what they're doing today, and that's the champion strategy.
00:01:58.159 --> 00:02:03.840
And then we come up with you know challenger strategies that are designed to outperform that champion strategy.
00:02:04.000 --> 00:02:16.639
Um, you know, and typically, regardless of kind of how mature a customer is in their data strategy, you know, typically when we're running an A-B test, we're looking for a meaningful lift.
00:02:16.879 --> 00:02:18.159
We're looking to generate a meaningful lift.
00:02:18.240 --> 00:02:24.400
By meaningful, I mean, you know, 20% or higher, you know, sort of increase in performance.
00:02:30.719 --> 00:02:35.680
Hello everybody, Adam Parks here with another episode of Receivables Podcast.
00:02:35.840 --> 00:02:39.199
Today I'm here with a returning guest, my good friend, Mr.
00:02:39.280 --> 00:02:42.800
Will Turner, joining me from TEC Services.
00:02:43.039 --> 00:02:48.319
Very excited to have him here today because just like me, he is a deep data nerd.
00:02:48.479 --> 00:02:57.439
He loves to look at the information behind the decisions being made and try to understand why or how decisions can improve operationally.
00:02:57.599 --> 00:03:00.319
That's what makes today's discussion very exciting.
00:03:00.479 --> 00:03:06.319
The last time we talked, we talked about data waterfalls, building your waterfall, testing your waterfall.
00:03:06.479 --> 00:03:15.520
And today I want to take this conversation a little bit further, and I want to talk about the real cost of bad data in your organization.
00:03:15.840 --> 00:03:19.919
But before we get to that, Will, thank you so much for joining me today.
00:03:20.080 --> 00:03:24.800
I really do appreciate your time and attention and coming on here, sharing your insights.
00:03:25.039 --> 00:03:25.759
Hey, thanks, Adam.
00:03:25.840 --> 00:03:26.879
Thanks for having me back.
00:03:27.120 --> 00:03:28.240
Just a quick introduction.
00:03:28.319 --> 00:03:29.439
My name is Will Turner.
00:03:29.520 --> 00:03:30.960
I'm a senior data consultant.
00:03:31.120 --> 00:03:42.080
I've spent roughly 30 years working for large data companies, consulting for data companies around the topic of consumer data, helping clients solve business problems with data and analytics and now AI.
00:03:42.319 --> 00:03:45.599
Well, and you've done a great job at that at a lot of different organizations.
00:03:45.680 --> 00:04:04.560
And I think your consulting experience is one of those things that really gets me excited about these types of conversations, because the depth of experience and understanding in different situations, scenarios, and circumstances really does change from a consultant's lens because you get access to so many different types of challenges.
00:04:04.800 --> 00:04:14.560
So, Will, as we kind of kick off this conversation today, when you think about the cost of bad debt for a debt collection operation, what's that first thing that comes to mind?
00:04:14.639 --> 00:04:17.040
How do you frame that in your own mind?
00:04:17.360 --> 00:04:18.639
Yeah, great question, Adam.
00:04:18.800 --> 00:04:25.360
So when I think about the cost of bad data and debt collections, the first thing that comes to mind is performance.
00:04:25.519 --> 00:04:31.360
You know, whether you're a lender, a debt buyer, or a servicer, you're trying to recover money, of course.
00:04:31.600 --> 00:04:37.519
And you know, bad data across that whole delinquent account lifecycle is a barrier to collecting money.
00:04:37.759 --> 00:04:42.879
And so, you know, from a performance standpoint, if you're a servicer, that's revenue and profitability.
00:04:43.120 --> 00:04:46.399
If you're a lender, you know, your losses are more significant.
00:04:46.560 --> 00:04:48.560
Yeah, that's the first thing that comes to mind is performance.
00:04:48.879 --> 00:04:55.120
Those performance issues is so you have this cost of not being able to optimize that performance.
00:04:55.199 --> 00:04:59.439
So you're going to have a lag to your competition from a timing perspective.
00:04:59.600 --> 00:05:02.480
You may not hit the same numbers overall.
00:05:03.040 --> 00:05:09.120
But when we think about bad data, what is some of the data that is going stale the fastest?
00:05:09.279 --> 00:05:17.680
And where should people be looking within their data workflows to say this might be the first place that I should look to identify the problem?
00:05:18.319 --> 00:05:18.720
Yeah.
00:05:19.040 --> 00:05:25.439
So as far as um, I think of things like addresses, about 37 million people move every year.
00:05:25.680 --> 00:05:30.720
A small percentage of those are posted with the United States Postal Service as an ON file.
00:05:30.959 --> 00:05:36.879
I think about there's a large volume of reassigned numbers that prevents risk and performance issues.
00:05:37.040 --> 00:05:42.480
I think about email addresses, not necessarily that email addresses go bad, but they go unused.
00:05:42.720 --> 00:05:49.360
And you know, typically consumers have multiple email addresses, so you know, having the right email address that the consumer is actually using.
00:05:49.519 --> 00:05:56.319
Those are things that come to mind in in the in the legal collections market, of course, you know, place of employment, people switch jobs.
00:05:56.480 --> 00:06:04.079
Yeah, so the the topic of bad data is pretty broad, but as it relates to the contact information, those are some of the data elements I think about.
00:06:04.399 --> 00:06:08.240
Well, let's think about it from the perspective of different disciplines of debt.
00:06:08.319 --> 00:06:21.680
So from a collection agency standpoint, it sounds like the biggest cost of bad data would be directly related to that contact information, the phone numbers, the emails, potentially addresses, whatever you're gonna need for that output.
00:06:21.759 --> 00:06:35.360
And it's not just the cost of not finding that person, it's not just the opportunity cost, but also there's got to be some other risks that are coming along with sending the wrong communication or communicating with the wrong people.
00:06:36.079 --> 00:06:37.120
Yeah, absolutely.
00:06:37.279 --> 00:06:42.800
And the one that comes to mind immediately is you know, FTCPA third-party disclosure.
00:06:43.040 --> 00:06:47.519
You know, you can't communicate about a debt as you know to somebody to the wrong consumer.
00:06:47.600 --> 00:07:00.480
And so if your contact information is wrong or your phone number's been reassigned to another consumer and you're communicating about a debt, you know, that's inviting risk, you know, regulatory risk, financial risk, litigation risk, reputation risk.
00:07:00.639 --> 00:07:00.879
Yes.
00:07:00.959 --> 00:07:05.839
So beyond the performance side of it, that data can lead to other consequences.
00:07:06.240 --> 00:07:15.040
And the cost of being wrong at scale is we've started to apply artificial intelligence, because now we're getting to a point where we can send more customized messages, right?
00:07:15.279 --> 00:07:22.000
We can do things at a larger volume and scale, but with that scale comes additional risk as well.
00:07:22.240 --> 00:07:34.000
Because if you're going to be wrong, if you're going to communicate to the wrong phone number, if you're going to send out emails to the wrong addresses, you're not just going to be wrong one or two accounts, you're going to be wrong at scale.
00:07:34.240 --> 00:07:47.040
How do you how do you kind of look at that risk tolerance and how does that go into the decision-making process when a collection agency is looking at how they might want to address the bad debt in their organization?
00:07:47.439 --> 00:07:49.120
Yeah, I totally agree, Adam.
00:07:49.360 --> 00:08:10.639
As far as when you when we think about using AI, especially virtual agents, or we think about digital collections, digital communications, comparing it to the voice channel and live agents, you're right, bad data amplifies this problem at scale because you're throwing typically a higher volume of activity, especially when it when it comes to outbound communications.
00:08:10.879 --> 00:08:38.320
And so from a consulting standpoint, when companies are migrating away from the voice channel or augmenting their voice strategies with AI virtual agents or digital communications, you know, part of it is helping the customer understand those risks, evaluating their portfolio and the accuracy of their information to identify those risks, and then coming up with a risk mitigation plan so that they step into those new channels carefully.
00:08:38.639 --> 00:08:51.120
Now, if we think about the same problem from the perspective of a law firm or a debt buyer who's purchasing accounts for litigation, the cost or the potential cost here seems to amplify even more so.
00:08:51.279 --> 00:09:09.200
Because if I'm a debt buyer or a law firm who's using that data to make an operational decision about which accounts belong in which operational channel, that could be a massive cost if I were to move the wrong accounts into the litigation channel.
00:09:09.360 --> 00:09:15.360
All those core costs, all those expenses, process service, a very expensive channel to operate in.
00:09:15.519 --> 00:09:28.639
So it seems like there's even more risk there, but potentially even more opportunity if you get your data waterfalls working correctly and you are actively moving the right data over your accounts.
00:09:29.039 --> 00:09:29.919
I agree, Adam.
00:09:30.080 --> 00:09:37.759
And that's something I think a lot of people overlook is that you know, you know, on brand new accounts, forget contacting the consumer.
00:09:37.919 --> 00:09:44.639
When we think about you know, scoring the account and segmenting the portfolio, the quality of the data.
00:09:44.960 --> 00:10:05.039
A good example would be, you know, having the correct name and address, as simple as that, impacts a third-party vendor's ability to score that account correctly, which leads to a lot of decisions from the very beginning, right, in terms of segmentation and which flow this is which which collection path this account is going to go down.
00:10:06.559 --> 00:10:07.360
Which is interesting.
00:10:07.519 --> 00:10:11.840
You know, we're we're finding ourselves in a situation where we're getting this increase of accounts.
00:10:11.919 --> 00:10:15.279
And I believe it was from the TransUn report in 2025.
00:10:15.360 --> 00:10:22.879
We said that 73% of firms were expecting account volumes to increase in 2026, up from 52% the year before.
00:10:23.200 --> 00:10:32.720
At the same time, we're seeing a stagnation or decline in liquidation of those same accounts, meaning we're having to do more with less.
00:10:32.879 --> 00:10:49.120
And being able to leverage data to make the impact of each action more powerful from a collections perspective seems to be the wave of the future, or at least the the end goal for a lot of organizations right now.
00:10:49.360 --> 00:11:02.720
And as a consultant who's helping to build these waterfalls, what kind of actions are you taking to help organizations down that path of exploration and understanding of how to unlock more value from that data?
00:11:03.120 --> 00:11:03.360
Yeah.
00:11:03.440 --> 00:11:05.919
So, Adam, this is the most fun part of my job.
00:11:06.080 --> 00:11:08.720
Like it is excited.
00:11:08.799 --> 00:11:26.480
That's why that's why I'm using data, um, helping companies use data in a in a more optimized fashion to make a measurable and meaningful performance impact is kind of at the core of what I do as a as a consultant.
00:11:26.720 --> 00:11:34.559
The steps that we go through to make that happen, I guess, you know, starts with a conversation about what's important to the customer in terms in terms of desired performance outcomes.
00:11:34.720 --> 00:11:39.759
And that might be very different by client or portfolio, or in the case of the lender by line of business.
00:11:39.919 --> 00:11:41.840
And that's where the conversation usually starts.
00:11:41.919 --> 00:11:51.679
And then, you know, based on that, we start analyzing whether their data processes and strategies, the quality of their information are aligned with those goals.
00:11:51.840 --> 00:12:00.960
And you know, typically there's an opportunity to improve whatever they're doing today to further optimize that desired performance outcome.
00:12:01.039 --> 00:12:10.000
And that becomes the measurement, and then we build a strategy, a data strategy around that, typically using their internal data as well as third-party data providers.
00:12:10.080 --> 00:12:21.039
And then we do we run an A-B test and we kind of look at what they're doing today, and that's the champion strategy, and then we come up with you know challenger strategies that are designed to outperform that champion strategy.
00:12:21.200 --> 00:12:33.840
Um, you know, and typically, regardless of kind of how mature a customer is in their data strategy, you know, typically when we're running an A-B test, we're looking for a meaningful lift.
00:12:34.080 --> 00:12:35.440
We're looking to generate a meaningful lift.
00:12:35.519 --> 00:12:40.879
By meaningful, I mean, you know, 20% or higher, you know, sort of increase in performance.
00:12:41.039 --> 00:12:51.120
And again, regardless of how good or bad somebody is today or how mature they are with their data strategies, there's always room, almost always room to opt further optimize it.
00:12:51.360 --> 00:12:52.960
There's always room to further optimize it.
00:12:53.039 --> 00:13:06.080
Whether or not you can further optimize to 20% each and every time, I think is a, you know, that that's a it's a big goal, but it's definitely a doable thing because the impact that data can have ripples.
00:13:06.240 --> 00:13:10.879
And when you figure out that process and you've got right, it's like finding the right recipe.
00:13:10.960 --> 00:13:12.720
You're gonna make the best pasta here.
00:13:12.799 --> 00:13:20.639
And when you when you know which ingredients have to go in, at which measurements, and in which order, it changes the way that that dish is gonna turn out.
00:13:20.720 --> 00:13:33.919
And I think if you can continue to tweak and understand that, what kind of testing programs do you build for that to clean out the bad data and append with more powerful, accurate information and data?
00:13:34.399 --> 00:13:36.639
Yeah, so it kind of depends on the data type.
00:13:36.720 --> 00:13:49.440
Um, you know, we're doing a lot right now with debt buyers and law firms around you know, address waterfalls, address hygiene, um, appending good addresses, just ensuring that they have a good address to reduce retard bail and increase process serving rates.
00:13:49.519 --> 00:14:00.480
And so, you know, for that, we look at you know, how well a company is cleansing and validating their addresses, how accurate is the information, what does their append strategy look like if they have one?
00:14:00.960 --> 00:14:07.120
Typically, we introduce some sort of address scoring component, which is very similar, you know, for for phones.
00:14:07.440 --> 00:14:19.279
I guess the methodology is pretty similar overall, how we evaluate how well somebody's current strategy is for whatever problem they're trying to solve.
00:14:19.519 --> 00:14:31.039
And then the methodology we use for testing kind of depends a little bit about some of it is how well they can maintain control of the test.
00:14:31.360 --> 00:14:34.720
I mean, we do all the the testing methodologies are pretty broad, Adam.
00:14:34.960 --> 00:14:43.440
So you know, we do everything from you know sending the same accounts to multiple third-party data providers, that all the data providers get the same accounts.
00:14:43.600 --> 00:14:56.240
We do a lot of champion challenger production testing where we'll take a portion of a client's inventory and we'll always have them running through sort of challenger strategies, multi multiple data providers, you know, in in a sequence.
00:14:56.480 --> 00:15:10.080
Sometimes we're testing with you know the the bad data, the the information that the customer knows is wrong, and then we'll demonstrate using third-party data providers the lift that we can provide with information that that they know to be bad.
00:15:10.480 --> 00:15:12.960
We'll take fresh portfolios, we'll take age portfolios.
00:15:13.120 --> 00:15:27.120
One of the one of the things that I find interesting is that when it comes to bad data, and we think about debt collections, the older that account gets, in some ways, the value of that bad data amplifies.
00:15:27.360 --> 00:15:53.360
And what I mean by that is you know, you take a take an account that went from a creditor and to a first party agency and it and it now becomes you know a tertiary or quad account, the the the bucket of bad data increases as those accounts age, which is you know, if that problem is resolved, it represents an opportunity for somebody to collect that money.
00:15:53.519 --> 00:16:08.559
So I find you know the the strategies that we deploy can sometimes depend on how bad the data problem is, and you know, the age of the account sometimes, you know, it often amplifies as it as the account ages.
00:16:09.039 --> 00:16:27.759
Well, as that account is going through the aging process and you've got more of that data that's decaying over time, I would think four, five, six years down the road, the data's not great, but the quality of the consumer is probably improved if it's going to improve over that time block.
00:16:27.919 --> 00:16:36.960
So the opportunity to re-engage, I think, is very real, but the cost of not having good contact data there or not appending.
00:16:37.120 --> 00:16:42.399
Let me ask you this when you think about new data that's coming over to a, let's say, a third-party agency.
00:16:42.480 --> 00:16:56.480
So they're receiving the data from the creditor, how well is the understanding of that data quality managed and communicated in that early onboarding process?
00:16:56.639 --> 00:17:04.000
Meaning, if I'm an agency who's getting accounts from a creditor, how well am I going to truly understand what that account has been through previously?
00:17:04.160 --> 00:17:11.359
And how much of that do I have to learn through experimentation and testing versus communication?
00:17:11.680 --> 00:17:11.839
Yeah.
00:17:12.000 --> 00:17:24.480
So I I think part of what you're hitting on there, Adam, is you know, what's the what's the communication needed between the servicer and the forwarder of that debt in terms of the history of that account.
00:17:24.880 --> 00:17:26.079
Especially in the early stages.
00:17:26.319 --> 00:17:32.319
Well, the hit not necessarily even just the history of the account, but the history of the data associated to that account.
00:17:32.559 --> 00:17:33.200
Yeah.
00:17:33.680 --> 00:17:34.160
Yeah.
00:17:34.480 --> 00:17:55.200
So so obviously that's that's key for uh you know when you're bidding on a portfolio to understand that when you're thinking about your contingency fees and how how rates are set compared to how how hard a portfolio is going to be to work, and how are you gonna likely perform and what investments and data are you gonna need to make in in order to perform.
00:17:55.680 --> 00:18:09.440
One of the things I often see when people get accounts, you know, typically a service or has, and lenders too, have some decent operational metrics to know you know things like right party contact rates and conversion rates and liquidation rates and recovery rates.
00:18:09.920 --> 00:18:23.920
I think sometimes people get the sense that if your right party contact rates and your conversion rates and your liquidation rates are what you expect them to be, that you don't have a bad data problem.
00:18:24.160 --> 00:18:35.039
And what we try to expose is what could your right party contact rates, your liquidation or your rates be if you further optimized the the data?
00:18:35.200 --> 00:18:42.720
And so I do think there's a false sense of confidence oftentimes that, hey, my data's good, my data strategies are good.
00:18:43.039 --> 00:19:03.359
And and typically for somebody that's satisfied with their current performance, I don't know that we can you know help them, but people that are are open-minded to saying, hey, we're doing things really well today, we're performing, we're winning, you know, we're top of a lot of scorecards, but but we want to know what we could be, or we want to stay ahead of the curve.
00:19:03.440 --> 00:19:06.240
We want to maintain our our forward position.
00:19:06.400 --> 00:19:09.759
Um you know there's a lot of opportunity there.
00:19:09.839 --> 00:19:12.079
I think there's a lot of sort of overconfidence.
00:19:12.400 --> 00:19:24.319
I I agree with that statement because you're only as good as you think you are right now, and your where you can go to is never determined if you're only operating on that same platform.
00:19:24.400 --> 00:19:40.799
So if you're doing everything exactly the same way, yeah, hey, you're not gonna necessarily get worse results, and you might investigate if you got results worse than expected, or you might investigate if they were better, but just because it's meeting your expectation doesn't mean your expectation is correct.
00:19:41.039 --> 00:19:43.279
Where how has your expectation been level set it?
00:19:43.359 --> 00:19:44.720
Is it through a champion challenger?
00:19:44.799 --> 00:19:45.920
Is it through a scorecard?
00:19:46.000 --> 00:19:48.880
You know, how how are you level setting your expectations?
00:19:49.119 --> 00:19:51.759
How do you suggest a client level set their expectations?
00:19:51.920 --> 00:19:53.519
I think is a good question for you, right?
00:19:53.599 --> 00:19:56.559
Like you see all these different situations and scenarios.
00:19:56.720 --> 00:20:02.720
How do you help someone to understand what level setting really means from a performance perspective?
00:20:03.039 --> 00:20:12.319
So as you as you were talking, Adam, I was thinking some one of some of the characteristics over the years that I've noticed about top performers, there's two things that stand out for me.
00:20:12.480 --> 00:20:14.640
One is there's some level of humility.
00:20:14.799 --> 00:20:19.119
They may just be super good and super competent, but there's some level of humility.
00:20:19.279 --> 00:20:23.759
And second, there's this mindset of ongoing improvement.
00:20:24.000 --> 00:20:35.680
And so one of one of the ways that we level set um it is to, you know, evaluate how somebody is performing today and you know, can can we get a lift?
00:20:35.759 --> 00:20:37.519
You know, is is there room for improvement?
00:20:37.680 --> 00:20:50.079
And so I don't I don't think like for for me when it comes to optimizing data strategies, it isn't like you arrive at something that's the the best and then you walk away from it.
00:20:50.240 --> 00:20:54.880
It's sort of this ongoing thing that you're evolving over time.
00:20:55.119 --> 00:21:05.920
And so I again I think the characteristics of top performers to me are people that they're not really satisfied, they're humble enough to know that what they're doing to today could be better tomorrow.
00:21:06.079 --> 00:21:09.519
And there is this sort of mindset of ongoing performance improvement.
00:21:09.599 --> 00:21:22.079
And so a lot of times we'll walk in, you know, people haven't touched their data strategies or haven't really evaluated their data strategies in years, or they've had the data strategy in place for years.
00:21:22.319 --> 00:21:26.640
Um and you know, typically that's almost always that's suboptimal.
00:21:26.720 --> 00:21:29.599
You know, it's something that we're optimizing over time.
00:21:29.759 --> 00:21:35.440
And again, with with top performers, we typically have a portion of their inventory always running in champion challenger test.
00:21:35.680 --> 00:21:40.079
So it's not like something that you arrive at and walk away from.
00:21:40.480 --> 00:21:54.480
Well, success is the first step in the journey of a thousand miles, and each success starts another new journey of a thousand miles that requires not only for you to get there, but for you to maintain along the way, which you know brings in more challenges.
00:21:54.720 --> 00:22:04.880
Now, we're living in a time right now where system conversions are happening at a pace that I've I've don't ever remember seeing before, at least not since like the early 2000s.
00:22:05.039 --> 00:22:16.960
And with all of this bad data getting locked in and all this bad bad data being collected, how do organizations start looking at that cleansing process at scale?
00:22:17.119 --> 00:22:21.920
Are organizations only looking at those new accounts that are coming in and trying to improve for the future?
00:22:22.079 --> 00:22:34.319
Are organizations trying to better understand what they could strip away from their current storage in order to reduce overhead, reduce cyber liability costs, other things?
00:22:34.559 --> 00:22:40.720
Yeah, how do you think organizations are looking at that ongoing data cleansing and management process?
00:22:41.119 --> 00:22:45.039
Yeah, so um, so I'm not an expert in all of those particular areas.
00:22:45.200 --> 00:22:49.519
You know, as a company, we manage a lot of data conversions and help companies through that transition.
00:22:49.839 --> 00:23:02.480
For me personally, the piece of it that is kind of most relevant for me and how I help clients is you know, during a data conversion is a perfect opportunity to look at at their data and optimize that data.
00:23:02.559 --> 00:23:06.640
It's for one, it's gonna make the data conversion to a new platform much smoother.
00:23:06.799 --> 00:23:17.440
And then all the technology and scoring products and the people and the processes that you add to that, the performance of that, it's gonna run more smoothly, it's gonna perform at a much higher, higher level.
00:23:17.599 --> 00:23:24.559
And so certainly when it comes to data conversions, uh, that's a perfect time to look at this and evaluate your data.
00:23:24.799 --> 00:23:35.680
And then certainly from a cybersecurity standpoint, you know, there's there's data, there's probably data there that doesn't need to be stored indefinitely, which you know presents, you know, creates risk, right?
00:23:36.319 --> 00:23:40.319
Creates risk and opportunity to reduce that expenditure.
00:23:40.559 --> 00:23:56.559
But let me go back to kind of the last piece of of my first question, which was more around as you talk to an organization about building a solution for data cleansing into the future, are they more focused on let me clean up what's coming in now?
00:23:56.720 --> 00:24:05.039
And from today forward, we're gonna do it this way, or people looking at trying to manage that entire ball of wax?
00:24:05.440 --> 00:24:08.079
How are people addressing the challenge?
00:24:09.039 --> 00:24:17.039
So if I understand the correct question correctly, Adam, I I see it more there's a specific problem that they're trying to solve.
00:24:17.279 --> 00:24:24.000
So, hey Will, we have too many, too many accounts and a high percentage of our accounts are are are still in Skip.
00:24:24.319 --> 00:24:28.079
That that's an opportunity um that we're not currently capitalizing on.
00:24:28.319 --> 00:24:31.440
Or, you know, hey, Will, our are for legal collections.
00:24:31.759 --> 00:24:38.960
I see that on client scorecards we have competing law firms that are achieving higher serve rates than than we are.
00:24:39.119 --> 00:24:44.960
And so oftentimes, you know, it it presents itself as a you know a specific performance problem.
00:24:45.119 --> 00:24:47.839
And it may not present itself that way.
00:24:47.920 --> 00:24:54.160
It also could be, you know, hey Will, we're our our liquidation rates are really strong, but our margins have compressed.
00:24:54.319 --> 00:24:59.279
And you know, we we have a we have a profit problem that that we need to solve.
00:24:59.440 --> 00:25:06.079
So how can we maintain our level of performance but fix that profitability problem?
00:25:07.119 --> 00:25:18.079
So you see the you see it manifesting in specific instances or challenges that they're trying to solve, and the end result may be a deeper cleanse of the data.
00:25:18.240 --> 00:25:26.400
Do you find that they need to taste a little piece of success before they're willing to go deeper into that data cleansing investment?
00:25:27.119 --> 00:25:36.799
Yeah, I uh I hope there are companies out there that are just saying, hey, I should be looking at you know data cleansing across my portfolio for lots of different use cases and reasons.
00:25:37.519 --> 00:25:39.039
I hope that that's the case.
00:25:39.279 --> 00:25:45.519
For me personally, it again it's it's presented itself more in um you know specific challenges.
00:25:45.599 --> 00:25:51.039
And you know, to answer your question, yeah, I think again, I think there's a little bit of overconfidence.
00:25:51.279 --> 00:26:10.079
And so when we can demonstrate the impact that can be made by cleansing your data, adjusting your data strategies, and somebody sees the financial and performance impact of that, certainly that's motivates them to say, hey, wow, you did that, you know, you help us do that.
00:26:10.160 --> 00:26:12.480
Now let's look at this might be the right investment for me.
00:26:12.720 --> 00:26:13.119
Yeah, yeah.
00:26:13.359 --> 00:26:14.000
Fair exactly.
00:26:14.160 --> 00:26:14.319
Yeah.
00:26:14.559 --> 00:26:22.079
So, well, are there any big misconceptions around dirty data in a debt collection organization?
00:26:22.240 --> 00:26:28.319
Is there anything else that we haven't talked about today that you want to make sure that our audience understands about bad data?
00:26:29.039 --> 00:26:38.880
So one one that comes to mind, so and often oftentimes with medical debt, I hear the misconception that you know we can't skip trace.
00:26:39.200 --> 00:26:44.720
We only use the contact information that comes over from um you know our client.
00:26:44.880 --> 00:26:47.599
And that that's a that's a common misconception.
00:26:47.680 --> 00:27:00.000
And and certainly when we work with servicers of medical debt that are high performers, uh they're definitely using third-party data uh to validate, to score, to append, to skip trace.
00:27:00.240 --> 00:27:12.319
And so, you know, medical's a big asset class in our industry, as you know, and I think there's a a lot of of opportunity there of just because of some of those misconceptions.
00:27:12.480 --> 00:27:19.680
And again, I I work with a lot of smart people and and customers, so not not not diminishing any of that.
00:27:19.920 --> 00:27:31.680
I don't know if everybody, um, I don't know if it's commonly known kind of the impact that data has on the investments that companies make in their people and in their technology.
00:27:31.839 --> 00:27:41.200
And so when I think about you know different technology initiatives, you know, there's an expected ROI, there's expected performance when they're making those investments.
00:27:41.359 --> 00:27:55.599
And sometimes I think it's overlooked the data side of that and what the ROI and the performance impact could be on those investments that they're making if data was a bigger priority or a bigger part of that conversation.
00:27:55.759 --> 00:28:00.880
And so that's something I see across the board, Adam, across all types of customers.
00:28:01.119 --> 00:28:03.119
Yeah, I think it's a great observation.
00:28:03.440 --> 00:28:10.400
I think the conversation has only started to come to the forefront because people have gotten excited about artificial intelligence.
00:28:10.640 --> 00:28:15.039
And now they're realizing they need the data to be available, right?
00:28:15.200 --> 00:28:22.000
They need to combine these things, they need to make it available to a single tool or location, and there's some data cleansing that goes into that.
00:28:22.079 --> 00:28:34.319
And that was really the introduction of that technology, let's call it over the last two years, I think has been that driving force that's pushing people towards optimizing the data.
00:28:34.640 --> 00:28:45.039
What's the point of running these artificial intelligence tools, outbound calls, text messaging, orchestration engines, all of these things if you're just gonna fuel it with bad data?
00:28:45.119 --> 00:28:49.680
It's like taking your brand new Ferrari and putting regular gas in it.
00:28:49.920 --> 00:28:52.960
It's just not gonna function the same way.
00:28:53.119 --> 00:29:02.799
And honestly, that's where this topic today kind of came from was a lot of the conversations I've been having in the marketplace about the deployment of artificial intelligence.
00:29:02.880 --> 00:29:10.880
And everybody's talking about the AI pieces, but very few people are talking about the data that actually fuels those opportunities.
00:29:11.119 --> 00:29:12.720
What does that workflow look like?
00:29:12.880 --> 00:29:15.440
How can we be using it specifically for this use case?
00:29:15.599 --> 00:29:27.279
And we've seen a growth of the data appending in research use case as we talk about the six use cases of artificial intelligence specific to the debt collection industry.
00:29:27.440 --> 00:29:28.960
And I think this is just the beginning.
00:29:29.119 --> 00:29:36.880
I think that's where we're going to see a lot more investment over the coming two or three years as organizations get the rails set up.
00:29:36.960 --> 00:29:41.920
And I look at it similar to the way that we saw the communication change over time.
00:29:42.240 --> 00:29:45.359
Setting up the rails to send a text message is one thing.
00:29:45.599 --> 00:29:52.000
Setting a strategy for content is another, setting an orchestration amongst all of the channels is yet still another thing.
00:29:52.240 --> 00:29:56.160
And I think the industry is going to phase its way into that approach.
00:29:56.319 --> 00:29:59.440
First, it's going to want to get AI installed.
00:29:59.680 --> 00:30:25.680
And then it's going to want to start taking it, and then it's baby steps going forward as we continue to grow as an industry and as we continue to test and understand not only the impact from a consumer standpoint, but who knows, in the next months or years, we may have the CFPB coming back, and who knows how they're going to view some of these tools, some of these processes, and some of our responsibilities from a data cleansing perspective into the future.
00:30:26.000 --> 00:30:26.880
Yeah, 100%.
00:30:27.359 --> 00:30:36.720
AI is opening up opportunities as it relates to data cleansing, identifying problems, resolving problems within a customer CRM.
00:30:36.880 --> 00:30:38.960
That's a piece of what you're talking about, right, Adam?
00:30:39.359 --> 00:30:41.759
It's just a touch of all of those things.
00:30:41.839 --> 00:31:06.319
And I think as we look at being able to cleanse that data, whether it be in our sales CRM, in our customer management databases, but our ability to do fuzzy matching logic and to do other things that allow us to better manipulate data at scale more reliably and more accurately, is where I see us spending a lot of our time and money in the coming two to three years.
00:31:06.640 --> 00:31:07.200
Yeah.
00:31:07.440 --> 00:31:18.799
So Adam, I want to go back to a question that you asked about the process in helping customers identify, you know, a problem or an opportunity to increase performance.
00:31:18.960 --> 00:31:24.640
And typically we'll start with a call, a discovery call to learn about a company's business.
00:31:24.799 --> 00:31:34.960
We'll set up a sort of data assessment where after we know a little bit about customers' business and their goals and the types of debt that they work, the services that they provide, we'll walk through their existing data strategies.
00:31:35.039 --> 00:31:38.799
And part of that is understanding their reporting and their KPIs.
00:31:38.960 --> 00:31:48.559
And so when we when we look at somebody's email strategy, for example, well, we know what's possible in terms of email delivery rates.
00:31:48.720 --> 00:31:53.440
We know what's possible in terms of um uh you know bounce rates.
00:31:53.519 --> 00:32:00.240
And you know, so so that becomes, you know, hey, you know, I know that you're performing at this rate today.
00:32:00.480 --> 00:32:06.079
We try to help clients using data achieve these kind of performance levels.
00:32:06.319 --> 00:32:07.359
We think we can do that.
00:32:07.440 --> 00:32:09.200
We could we think we can demonstrate that for you.
00:32:09.279 --> 00:32:11.039
Let's do a pilot and prove it out.
00:32:11.119 --> 00:32:13.119
And so that that's part of the process as well.
00:32:13.279 --> 00:32:19.759
It's just you kind of understand after working with a number of clients on certain strategies, you know, what's possible.
00:32:19.920 --> 00:32:24.720
And oftentimes people are not performing at that level, and we see an opportunity.
00:32:24.880 --> 00:32:28.960
So well, the opportunities are definitely abundant in the upcoming future.
00:32:29.039 --> 00:32:31.119
And Will, I could talk to you for hours, buddy.
00:32:31.279 --> 00:32:38.720
Every single conversation we have, whether it be on a podcast, walking through the halls at a conference, I learned something from every discussion.
00:32:38.880 --> 00:32:43.200
I value your friendship and I really appreciate you coming on here and sharing insights today.
00:32:43.519 --> 00:32:45.039
Yeah, appreciate the opportunity, Yan.
00:32:45.200 --> 00:32:45.440
Thank you.
00:32:45.599 --> 00:32:46.000
Absolutely.
00:32:46.079 --> 00:32:53.759
Will I really appreciate you for those of you that are watching, and I'm sure that you have some questions here for Will, you can leave those in the comments on LinkedIn and YouTube, and we'll be responding to those.
00:32:54.000 --> 00:32:58.079
Or if you have additional topics you'd like to see us discuss, you can leave those in the comments below as well.
00:32:58.160 --> 00:33:05.039
And I'm willing to bet I can get Will back here at least one more time to help me continue to create great content for a great industry.
00:33:05.200 --> 00:33:07.920
I know we will have no shortage of topics to discuss.
00:33:08.079 --> 00:33:09.200
But Will, thank you so much.
00:33:09.279 --> 00:33:10.160
I really appreciate you.
00:33:10.240 --> 00:33:11.200
Thank you for coming on.
00:33:11.440 --> 00:33:12.720
Thanks again, Adam.
00:33:12.960 --> 00:33:14.559
And thank you everybody for watching.
00:33:14.640 --> 00:33:16.160
We appreciate your time and attention.
00:33:16.240 --> 00:33:17.839
We'll see y'all again soon.
