Adam Parks (00:06)
Hello everybody. Adam Parks here with another episode of receivables podcast. Today I'm here with someone who is clearly about to become my new friend, Mr. Jack Mahoney, who's joining us from National Credit Adjusters. How you doing today, Jack?
Jack Mahoney (00:21)
I'm doing great Adam, thanks.
Adam Parks (00:23)
I really do appreciate you coming on and sharing your insights with us today. Data analytics has become a huge part of the debt collection industry and trying to predict and understand from large data sets. And from my understanding and from our introduction, that's kind of your area of expertise, being able to query in and understand or answer questions for executives based on the data sets within the organization.
Can't wait to dig into that. But before we jump into all of that, could you maybe tell everyone a little bit about yourself and how you got to the seat that you're in today?
Jack Mahoney (01:00)
Yeah, sure. You know, I was born and raised in Hutchinson, Kansas. That's where our corporate office at NCA resides. I spent a few years in Oklahoma. That's where I met my wife of now 32 years. Started a family at a young age and found myself really in need of some financial support. Right. And so I found in Hutchinson a place called Midland Credit Management where I started in debt collections. It was meant to be kind of a short term thing. Right. I didn't intend to stay there too awfully long.
But about three decades later, here I am, still in the industry. I've worn a lot of hats over the years. I've been from the collection floor to operations management to the role I play today in analytics, which is by far the best part of this journey that I've had. I love doing that job. I can take that, really that practical experience, industry knowledge and apply it to the...
business requirements, the challenges that we face today. It's been a lot of fun and look forward to continuing on for another decade.
Adam Parks (01:58)
That's fantastic. I can clearly tell that we are going to be friends. That is a great approach to it. And there's really no better way to be able to analyze a business or a business requirement than to have actually done the job. And for those people that have done all of these different roles throughout an organization, I mean, those are always your most powerful thinkers. And then when you start layering the data on top of it,
What an incredible opportunity. But for anyone who's not familiar with National Credit Adjusters, could you tell everyone a little bit about the organization and what you do there?
Jack Mahoney (02:31)
Sure, NCA is a company that's been around since 2001. That's when I started with the company, shortly after it opened its doors. NCA is a debt buyer in the space. So we go out and we find the receivables that we believe we can turn a profit on and bring those in-house to feed to our collection staff.
That's kind of an in and a show what NCA does. And I'm the Director of Analytics here at the company. So I'm responsible for really bringing out those data insights from the millions and millions of records of data history that we have to help drive performance in the company and gain efficiencies where we can.
Adam Parks (03:17)
Fair enough. So speaking of those efficiencies as someone who has done all the jobs and worked in analytics, the world has changed pretty significantly. When I was still brokering in the let's call it the early 2000s into the 2010s. The world was driven by spreadsheets, it was driven by pivot tables, and we were running stratifications of a file to try and understand what that data set, you know, really looked like, because
You got a million records, you can't really go and look at individual accounts and learn something you have to be able to categorize group in order to understand. But I can imagine in the last 15 years that the technology being used has changed pretty dramatically. Although I know you're a spreadsheet wizard as well, what kind of tool sets have you started to deploy or utilize in order to really...
Get your hands around this volume of data.
Jack Mahoney (04:12)
Yeah, I mean it did start with Excel right? It all goes it goes back to Excel I could still remember the the day that I learned about pivot tables. It was Eye-opening right made things just so much easier. It was fantastic And then then it progressed into using Excel in a little
Adam Parks (04:21)
Absolutely. Changed my life.
Jack Mahoney (04:30)
bigger way, right? Learning how to use VBA to control automation and do a lot with that, along with Microsoft's Access product for database use. But really that progressed into using a tool that we use on regular daily basis throughout the entire organization. And that's our CLICK analytics tool.
Folks are familiar with Qlik. It's similar to Tableau, if you know what Tableau is. So it's in that area. But Qlik and its Qlik Cloud platform really allows us to do a lot. We can bring in data sources from all kinds of areas, whether it's cloud-based storage areas, on-prem storage areas.
Adam Parks (04:53)
okay yeah alright that actually is a good parallel for me
Jack Mahoney (05:12)
and just really bring that data together so that we can create our wonderful dashboards. It's going to give us a lot of insight. know our management team uses our call center analysis application on a daily basis. It updates basically in real time so they can see how the performance of their account managers are doing throughout the day. Very key information allows them to
not only kind of seeing in real time how they're performing right then and there, but also look at some history to say, hey, here's an area where maybe you can make some improvements and get your numbers looking a little bit better.
Adam Parks (05:43)
So are you looking for like building out trending information to try and understand how things are performing over time and then kind of working your way back through the data to start looking for strategic opportunities? So I guess let me ask it in this way. Are you starting from a problem that you need to solve or is it more of a data exploration or do you get an opportunity to do both?
Jack Mahoney (06:07)
It's generally both, right? There's times where we're going to see what we have in the data, explore that, and find some insight that we then build out into a dashboard or information that can be used. But a lot of times we have a question that gets posed, a business question that needs to be answered. And when we hear that question, we ingest it and then start going down the path of how can we answer that question. And usually kind of start with what does the end result need to be?
And then let's build towards that end result. So for example, if the end result needs to be, can we improve our promise rate?
right, the conversion rate on the telephone. We might go and look at the data to get insight on, we talking to enough individuals to create the promised conversion rate that we intend to have? And if we are, that's fantastic. Where else do we need to look in order to make those changes to impact that conversion rate? If there is something in that RPC rate, the right-party contact, that's a little on the lower side.
we can go back and look at, what are we building in? What are we delivering to the account manager? What can we change there to make a difference in how they are receiving those calls and the amount of people that they're talking to? Because when it comes to phone conversations, phone collections, it's all about how many people can we get on the phone and give that opportunity for our team.
Adam Parks (07:35)
Very interesting. And as you've moved into digital communications as well, I can imagine that having an analytical infrastructure sitting behind that would help you to refine content scripting and really drive forward some of those performance related make tricks.
Jack Mahoney (07:51)
Yeah, yeah. On the digital side, it's something that I feel like we're just really getting into on a little...
little deeper, right? In our industry, you know, as anyone knows who's in this industry, we are regulated. And so we're, we've got to follow a lot of regulations and be very compliant, which is fantastic. We'll make sure that that happens, but it does handcuff us a little bit on how far we can go on the digital side. So we're doing everything that we can to progress and get better into that. And as we do, we can, we're making sure that we're getting in that information, retaining that data so it can
use that for insights down the road and make better decisions because ultimately, know, it's it's we haven't talked about modeling yet, but I'll just go there real quick. We let's do it. So we have we have a servicing score, which is a predictive model.
Adam Parks (08:39)
That's that's the next question so we can jump right into it.
Jack Mahoney (08:48)
using some machine learning with all of our historical data that helps us understand where we should focus our attention. It breaks it up into a decile. So one through 10, 10 being the best possible outcome, one being not great, right? Don't put a ton of effort into that. It's probability.
Adam Parks (09:02)
Is this a probability score? So you're thinking about it in
terms of probability of engagement or collection?
Jack Mahoney (09:07)
It
essentially is, right? We just put a numeric value to it to make it easier to understand. So we'll want to put a lot of effort into those higher scoring values, those say six through 10s, make sure that we're doing what we can to get those folks on the phone or get a marketing information out to them. And it might be on a little bit of a higher cost engagement. The ones that are on the lower side, the ones, the twos, maybe even threes and fours and fives,
we'll still want to apply some effort, just not as much as the higher value. And we'll want to try to do that at a lower cost, which is really where that digital aspect comes into play. So if we can send a text or an email at a few pennies versus a phone call or an engagement in that way that gets a little more costly, then we'll certainly do that on that lower value product. Because we know that the conversion rates, even the contact rates on those are going to be on the lower side.
Adam Parks (10:03)
looking at those probabilities or the propensity for one of these consumers to pay and really being able to hone it in based on the results of your own organization is impressive. There's a lot of organizations out there selling scoring and sure some of it has some predictive attributes to it, but being able to understand from your own data set. Now in order to build it, you mentioned the machine learning and the models as well. So with clicks, that's kind of your data lake and then you're able to run models over that.
data set, is that kind of how I should be thinking about it?
Jack Mahoney (10:34)
Yeah, on the click side of things, they do have a product that is using machine learning, AutoML, which makes it pretty straightforward to really bring the data together. mean, really, the most difficult part in running this stuff is deciding which features, which sets of data, how to create the data set that goes into the model itself.
Once we can get that piece figured out and feed it in, then we can push a few buttons really and see what it can produce for us. It does still take some kind of a look in order to decide which of these models is going to fit best. If any, it doesn't always work out that it is great. It's very important to make that distinction that if this isn't something that's going to fit well,
We're gonna force it in and say that it's a model, let's use it. If it's not a good fit, not a good model, it's gonna potentially do more harm than good. So we make sure that it's going to test it, make sure it's gonna do what exactly we need it to do.
Adam Parks (11:33)
So as you start to look at kind of experimenting with AI models and testing and trying these things.
I'm going to go back to my original question, say like, are you looking at those models as an exploratory, like, okay, let me put together a data set, try and run some models on top of it, or do you generally setting out on that exploration path with a particular, let's say solution in mind, I've got a particular problem that I'm trying to resolve, and now I'm going to try and find the model that fits that or is it okay, I've got a data lake, like, what can I find in the data lake?
Jack Mahoney (12:07)
You know, it's really going to be, here's a target. I'm trying to get to this target. And one of our models is in pre-purchase evaluation. So we need to take a look at a very limited data set of a file that we want to potentially purchase and load into our system for collections.
Obviously, we want it to be a performing portfolio. We want to be able to get our return on it. And so we've trained up a model that will essentially use a target, which is a predictive payor target, to say that there's a reasonable probability that based on the information we have, this consumer will pay us within the first year of collections.
and then we can use that information to then compile within that data set the number of payers and the potential of liquidation, potential cash gain for that portfolio to then value it, to then give it a price that we could pay to where we feel that we could be successful. Hopefully that helps to answer that.
Adam Parks (13:10)
No, it
does. It absolutely does. It sounds like you're you're setting off with a particular target and saying, OK, I've got this data set. I'm trying to solve for X, Y or Z. And you may not necessarily know what that result is going to be right as you start down that exploratory path is you've kind of started playing with some of these models. Have you had any aha moments where you just went, my God, that that was completely counterintuitive. But the data says it's X. And I really thought it was going to be why.
Jack Mahoney (13:15)
Yes.
You know, there's little things always that happen in that context. I can't think of a moment where it was just real eye-opening time, but as we continue to go down the path of using this more and more, I know we're gonna run into that. It's certainly gonna happen. We've got enough data. We're always gonna get surprised by it. I know one of the things that we...
intend to do in the future is, right now a lot of the models we are running it, we're getting information, we're applying it to decisions and driving the company. We want to get a little more real time.
right, and use this modeling to help make decisions in a real-time fashion while we have the consumer on the phone. If maybe you could tell us something through the effect of, look, if I'm able to make this sort of payment or do this sort of arrangement, is that something that could work? And we could plug that in, and the model would give us an idea of whether or not that would be feasible for this particular consumer profile.
I think once we go in that path, that's when we're going to have those. Okay, I didn't think that was going to be the outcome, but there it is.
Adam Parks (14:40)
I agree. I've had a few of those moments as I've started digging into the data from my own organizations and just trying to understand the types of content that we create, how frequently we create which types of content, what kind of results we're seeing based on a variety of different variables, variables, but you must live in a world of variables, Every problem in front of you has got all of these different variables that you're trying to solve for.
What kind of approach do you generally take to a new problem? Somebody comes into your office, knock, knock, knock on the door. Hey, Jack, I need to understand X, Y, or Z. How's your mind frame around trying to solve a problem?
Jack Mahoney (15:20)
Sure, that happens quite a bit. And this, what we're here to do is solve problems. So that's the fun of the job. I think when generally speaking, when somebody comes in with a business question or a problem that needs to be answered, it comes back to that, I'd like to figure out, because I've got to make sure we understand all of their...
parts and pieces to that question. Sometimes it might be one sentence, but it really should be a whole paragraph, right? Because we need to build that framework out. But I want to understand what is their end result. Let's say they're asking for a particular report on some set of data. I'd want to know exactly what structure would you like this to be in, right? What information do we need to resolve? Is it just this one question that you're trying to answer, or is it more?
and let's make sure that we put that in that report. So we kind of start with a vision of what the end result needs to be, work our way backward into what steps need to be taken in order to make that end result happen. Now we know exactly what data need to draw from and how to transform that data into the insights that they really need to have within that report if that's the case.
Adam Parks (16:25)
So it sounds like you're starting with a success criteria, meaning that, okay, I understand your problem. What does success look like? I find that to be a great way to start approaching literally any problem in an organization. What is, if I'm gonna do this experiment, if I'm gonna try and solve something, start by determining what does success actually look like, right? Because it's real hard to get somewhere if you don't know where you're trying to arrive.
Jack Mahoney (16:29)
Yes.
Exactly.
Adam Parks (16:51)
So having a destination in place and then being able to categorize the data sets that might be needed in order to resolve that problem seems like a pretty logical approach to resolution. As a team, do you have a team of people that you're working with or are kind of lone wolf working your way through some of the data analytical problems, challenges really?
Jack Mahoney (17:14)
Yeah, I mean, for a long time, I was the only one in the analytics area. But over the past few years, we've started to build out a team. So we do have three on staff in the analytics area here today. So it's definitely a team effort at this point to get all the questions answered and get everything out there that this organization needs.
Adam Parks (17:35)
Well, if there's one thing that a room full of executives has its questions, and it doesn't matter what organization you're in. And I have yet to meet the IT team on Earth that has enough bandwidth to take on the next project. So I think it's great that you guys have a
a team of people that are focused on such a mission critical item. So much of decision making at this point in the industry is driven based on the data sets and those organizations that are not empowering leadership with appropriate data sets and not only just the data, but the visualization and interpretation of what is being found in the data.
and it sounds like Qlik is kind of like a Tableau or a Power BI in terms of being able to visualize. Have you found there to be any challenges specific to the industry about visualizing some of these larger data sets?
Jack Mahoney (18:29)
Um, you know, it's, it really comes back to what do we want to see? Uh, the challenge is always just ensuring that we have the backend data organized in a way that we can make use of it. Right. It's not always, it's a little messy sometimes. Right. So we've got to. Yeah. Yeah. Um, so we can, we can generally speaking, bring that information together in a way to where we can make use of it and, and really get it out there in the hands of the, the end users that need it.
Adam Parks (18:43)
Structured versus unstructured data, sure.
Jack Mahoney (18:59)
It's one of those things that we know that bringing this information to the table is going to provide that insight that's really required to energize the company and give us what we need to be successful. And that's really what it's all about.
Adam Parks (19:15)
What's your favorite part of the process? What's your favorite part of this job? This particular role at NCA?
Jack Mahoney (19:21)
You know, it's answering questions. I'm the type of guy who, if you pose a question to me, maybe once, I don't know. I'm not sure that I've ever said no, can't do that. I'm gonna find a way. I'm gonna find a way to answer this question. And it generally happens. So that's what I love about doing it. I just love problem solving and getting out there and digging into information.
that provides that insight and it gets the person what they need.
Adam Parks (19:46)
And that I think is why we are going to be long term friends. I really like that approach. It's like it's not easy to be the guy with the answers and to have to go answer some questions that are maybe sometimes impossible to really answer. And for me, everything comes down to a.
degree of certainty, right? What's my confidence in this result? And I think going back to, you know, college and thinking about statistics, everything was a standard deviation from the norm. And I think my brain just started thinking about it in that format and being able to normalize the outliers and get a better understanding of what that real data set looked like. Because when you start plotting these things that are all over the place, you really can start to find some correlations
even when there are offsets and outliers, but with a million records, it's not like you can drop that into a spreadsheet and go back to your core. And a lot of people in your role immediately want to go back to their comfort zone. Like I want my, my Excel safety blanket and I want to be able to get in there and see it, but there becomes a limitation on just the volume of data that I can't stratify it in that format anymore. Have you had any of, I have those moments all the time. Have you had any of those moments where like, okay, now
starting to get out into all of this new technology, but feeling that need to kind of come back and stick to some fundamentals in which you know is your source of truth.
Jack Mahoney (21:12)
I mean, there's always times where, you know, having the data all aggregated together and everything, that's great and we need that. But once in a while, I just want to throw it into the Excel workbook and just work with it there. Excel can hold up to a million records now, so I can do quite a bit with that. I do try to stay away from it, but, you know, it's handy.
Adam Parks (21:36)
How fast does Excel run with a million records? I've never tried. Fair.
Jack Mahoney (21:40)
depends on how many calculations you got going on. It can run to the point of crashing
if you're not careful.
Adam Parks (21:46)
Fair statement. That's awesome. I honestly did not even know that I've been using Google Sheets a lot lately for my business just because we kind of switched over to a Google shop and for whatever reason, the Microsoft products just don't run well on my Macs. No matter how native they try to make it, I feel like that's still just gonna be a problem into the future.
Jack Mahoney (22:03)
Right.
Adam Parks (22:05)
But Jack, man, this has been a great conversation. I really do appreciate you coming on and sharing some insights from the world of data and data analytics for the deck collection industry. This has been great. Thank you.
Jack Mahoney (22:17)
I appreciate you having me. It's been a lot of fun.
Adam Parks (22:20)
Absolutely. For those of you that are watching, you have additional questions you'd like to ask Jack or myself, you can leave those in the comments below on LinkedIn and YouTube and 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 hopefully I can get Jack at least back one more time to help me continue to create great content for a great industry. But until next time, Jack, thank you so much for your insights today. I really do appreciate you.
Jack Mahoney (22:44)
Thanks a lot.
Adam Parks (22:45)
And thank you everybody for watching. appreciate your time and attention today as well. We'll see y'all again soon. Bye everyone.
