Hello and welcome to Risk Chats with Affirm. I'm your host, Paul Marshall. Today we're speaking with Brian Jones. I'm using data analytics in your ERN program.
Brian was a long-time government employee, Postal Service LIG, and now the owner of his own company, specializing in analytics. So without further ado, here's the show. Hello and welcome to the podcast. So today we have with us Brian Jones from Strategy First Analytics, and we're going to talk about data analytics.
So good morning, or good afternoon, Brian. How are you? Yeah, good afternoon, Paul. I'm great.
Thanks for having me. Absolutely. Well, why don't you introduce yourself and tell us a little bit about what you do and, yeah, just start maybe your career a little bit. Yeah, sure.
I'm recently retired from the government. After 31 years, I spent most of my time at the Postal Service Office of Inspector General working in the oversight community. And about 10 years ago, we opted to try to build an analytics program to build fraud models for fraud detection. Eight or nine years ago, analytics wasn't nearly the buzzword that it is today.
So I wish we could say that we did everything right the first time, but honestly, it took a lot of trial and error and building things as we went. So we had some success returning investment to the organization. And so once I retired, I went to work for a data science company for a couple of years, and I thought, you know what? I really want to get out and do this on my own.
So now I'm consulting with other government agencies and helping them improve their analytics capabilities. Great. And what are some agencies you've worked with so far? So I've worked with, as I mentioned, Postal Service OIG, Postal Service Agency, Department of Labor.
I'm working at HUD right now, USDA OIG, Treasury Inspector General, Education. So quite a few different agencies. Great. Well, why don't we start off with a real basic question that maybe folks think they know the answer, but maybe they don't.
And that's, I mean, what is data analytics? And, you know, how should you be utilizing this capability at your agency? In case anyone hasn't picked up on the fact that I'm not a technical person, that's far in the conversation. I'm from the business side of analytics.
And so it's evolved a lot over the years. We used to get hung up on terminology, and then we're inundated by tools that people build to do analytics for you. And everybody has a different idea of what analytics is and what the different levels are. But basically, I believe analytics is just utilizing your data, combining that with human intuition and experience together combined to make evidence-based decisions.
At its core, that's what it is. Okay. So it's not just some black box you kind of throw your data into and some magic happens? Well, I mean, if you want to believe that, there are some solutions out there that may profess that.
But in my experience, that usually doesn't get us to where we need to be. Right. Well, so yeah, I mean, it's a little bit to hear what, you know, what have you seen that agencies do in this realm and maybe some pitfalls they fall into or things they're not thinking about? Yeah.
So normally in government, at least when I was there, no one walked up to me and said, we're going to give you a $5 million budget. And you got three years to stand at this program and you have all the tools that you need. So my philosophy is, you know, in the agencies that I've worked in, there's usually already a lot of data. There's usually some sort of a technical infrastructure.
There are tools that have been purchased. And maybe what we haven't done is put the pieces together in the right order and try to solve the right problem. So my approach starts from obviously the name of my company and strategy first, you know, start with the strategy first and put that framework around that and then make an investment in how you're organized and how you put the pieces together. And don't spend a lot of money on tools until you understand what you're doing and you understand what tool you might need that's going to return value to the organization.
Right. So what tools my agency already has specifically? I mean, as basic as Excel or whatever, right? But what other things may they already own that they didn't, you know, put in your inventory of potential?
Yeah. I mean, Excel comes to the top of the list just about everywhere we go in ACL and the audit world. But also, you know, we're talking about MicroStrategy or Tableau or Esri, Geospatial. Of course, there's SAS and there's R and there's Python on the analytics side.
So there's tons of them out there, a lot more open source now, you know, than when I was in government. So usually we will find an organization has access to or owns something that they're using already or that they purchased and they're not using and they can utilize better. Yeah. These agencies are so large.
I mean, you see all kinds of software that you probably didn't realize that you had. Yeah. One time I went into an agency and said, hey, is there a chance that you already own this piece of software? And they said, no, we don't think so.
And the next week I was at that software company's headquarters and there was a cutout of the president of that agency standing there advertising. So I went back to them and I said, you know, you really do own this already. So a lot of times we just have to figure out who bought it, what do we have access to and what do we own? And just a little bit of legwork there usually gets you something that you can start with.
Right. Yeah. Pretty funny. Yeah.
Same thing everywhere I go, you know, you look at the inventory and you say, hey, there it is, but why can I get it? Or how do I get it? But that's another story. Either way, all you know that you have it.
That's the starting point. Right. So yeah, I mean, so what, you know, what do you recommend for folks to get a basic analytics program up and running? As you said, you're probably not going to get a big budget for these things in the beginning.
So how do you show values early on? Yeah. And up and running is misunderstood. I think a lot of people think they have to start from scratch.
But what we found over the past few years is a lot of organizations and agencies have done a very good job analyzing their data and they've put groups of statisticians together or analysts together. And so they're already doing some sort of forensic analysis or data analysis. It's just not understanding how to evolve to something more advanced and how to utilize the tools that are out there and put things together with the strategy. I mean, we're talking about just a basic mission and vision and a business model and a concept of operations and putting processes in place so that when you have a problem and you choose, are you choosing the right problem?
Are you choosing the right customer? You know, do you have the data to support this type of answer? What kind of algorithm can I throw at this? What level of analytics is this?
What is going to be a potential return of value to the organization? And so if we don't ask these questions ahead of time and we don't have this framework ahead of time, typically we do a whole lot of work. We may or may not get an answer to the question and the question may not have held much value to begin with. And we don't really know how much better than random that we're doing if we don't apply some sort of analytics to the problem.
So I think, so you mentioned that you really start with a framework. What do you, I mean, what's the very beginning to make sure you're, before you go off and, you know, on a tangent, how do you really focus and figure out what are you trying to do here? I mean, it all starts with what's the problem we're trying to solve? And a lot of, and a lot of times, especially in government, government has a lot of passionate, dedicated, talented employees.
And sometimes though, when we're in that seat for a long time, we ask the same questions all the time. And we think we need to, we think we already know the answer to a lot of things. And what analytics can do for you now is it allows you to ask more robust questions because now we're able to take data sets and combine them. We're able to look across the enterprise and understand how programs are connected and how can someone else's data inform, you know, the problem that I'm trying to solve.
So it all starts with that problem. Do I have the right problem? And if we solve that problem, is that problem even valuable to the, to the organization? You know, is, or is it just something that's really cool that one person's been trying to get an answer to?
So it's globally having that process in place where are we asking the right question? Do we have the data? Do we, when we answer that, what is the value? What kind of analytics or algorithm can we throw at that?
How are we going to visualize and deploy that? And at the end, when we deploy it, how do we learn from what we've deployed and how do we just circle that loop over and over again and iterate what we've done to improve the model itself? So just some basic processes to put in place rather than just sending a group of smart people out and giving them some data and a problem to solve and hope they're going to come back, you know, and give you the answer that you're looking for. Right.
And I mean, do you find that folks tend to have sort of preconceived notions of what the data is, you know, what's going on and then really haven't analyzed the data deeply enough to see what the true picture might really be, like what's really going on here? Yeah. And a couple of things that people often get in the way of just that exact problem that you stated is a lot of people say, look, we don't have good quality data. You know, our data is just poor quality.
And, you know, my response to that is, well, you're doing something with poor quality data now. And there's a lot of things that we can do to cleanse that, you know, and get it into a better condition to do some sort of analytics on it. Yeah. Oh, the other thing, I know what I was going to say.
The other thing is that people don't realize how much human bias they bring into the situation. So it's a double-edged sword. When you sit down with somebody who runs a program, they're a subject matter expert and they know everything about it, sometimes that doesn't allow them to look at it through an unbiased lens. You know, so how do we find a way to remove that bias so that that analyst who has that experience can look at the data in a new way and see new insights that, guess what, analytics and data is not going to tell the whole story for you.
You know, on the other hand, a lot of people say, well, let's just automate everything and put algorithms at it and eliminate the people. And that's really not going to get you, you know, very well, very far either. So it's that combination of equal parts of the human, the subject matter expert, the analyst, and the data analytics and the algorithm. And combining those two things together really is going to give you the best value for what you're looking for.
Okay. Well, so, you know, today's podcast is for a firm with an enterprise risk management organization. So we want to kind of tie this now with data analytics a little more, you know, directly. So I think you have some good ways of kind of tying the two together.
You know, for a risk manager, how does this connect? Yeah. So we started in the fraud area, but obviously there's more to fraud out there. And when we talk about risk, a lot of times the way that we have identified risk in our agencies before is just by asking people, let's do a survey.
Let's look at some data and ask some people the questions and find out what we think our problems are. And then let's score those ourselves and prioritize them. And then let's just go solve those problems. Analytics can help you, I believe, on both ends of enterprise risk management.
First of all, it can inform what your risks are. So allow the data to tell you in your organization where some of the risks are and not only where they are, but deeply, where do you need to go to try to resolve that? And so that underlying output of a model may inform where you want to prioritize your risk. On the other side, if you've identified a risk, analytics can also help you dig deeper into that.
So what's the root cause of what we're looking into? And in fact, is this as risky as we thought it was? And so really, I believe analytics works on both ends of that scale, both informing risk or reacting to risk. Right.
And from the informing side, kind of back to what we said before, do you even know really what the issues are if you haven't reviewed the data in depth enough or even gathered the right data? Right. And then, of course, I think what's really interesting to me is now, okay, we've identified a risk. We want to mitigate it as much as possible.
Let's attack that thing. Let's reduce that risk with some tools here. So along those lines, I mean, what have you found that agencies can actually do to fight against fraud and things like this? Well, I mean, I'll just use human resources as an example because everybody has an HR department.
Usually when somebody escalates an HR problem to a high level of risk, it'll be something like we can't hire to compliment or we have people that are leaving and we can't we can't hire people fast enough. And so the assumption is HR is a problem. We need to get them to hire people faster. How can we improve that process when, in fact, using your data and your analytics, you can look at it a little bit more nuanced way.
You can say who left? Why did they leave? How long have they been here? Who stayed?
I mean, learning from people who are staying in your organization is something that you can do, too. How long have they been here? Who are they working for? What is their background?
What's their education? How do we incentivize the people that we think might leave? And how do we learn from the people who are going to stay? And how can we recruit more of the ones that tend to stay?
So that's an example of different questions you can ask by combining different data sets and not just assume that HR needs to hire people faster, which is our typical knee-jerk reaction sometimes to, you know, coming up to staff compliment. So that's just one example in the HR area. That's a good one, right, because I mean, for years, you always hear, you know, all our problems will be solved. We can just bring people on quickly.
That's right, right. But, you know, it's been years and, I mean, not that that's really, I just did some speed-ups, I guess, a little bit, but you're right, overall, is that really solving the true issue of, you know, people not staying around or not getting the right people in the door? And on the recruiting side, you know, people tend to want to hire the perfect person. You know, they want to build a job description and a job posting, so if I could just hire this person, they'd be the perfect person.
Well, they probably don't exist to begin with. It's too rigid a requirement. Yeah, and second of all, do you really want to hire the best person or the most, do you want to hire somebody who's going to leave tomorrow, or do you want to learn from who have become your better employees and hire who looks like they do? Who's coming in the door with that same attitude and skill set that we can develop into that employee?
So I'm not saying that that's the answer. I'm just saying that looking at data a different way, combining it in a different way, allows you to ask these questions that we've never really taken the time or had the opportunity to ask at that level. Right, so this is a good example. This is like an HR department wanting to, you know, solve the problem of retention or hiring or finding the right people, so just using the information you've already got somewhere in a system or out there, you just pull it in and really analyze it, you know?
Well, talking about what tools that you might have that you aren't aware of, agency after agency is not even aware of the data that they have access to, or sometimes they've created data and, you know, they've not captured it in a way that's meaningful, or they can only query it, or they can only run reports on it. So just for an organization to understand, we've got so much data that we didn't realize that was there, how can we learn about that and tap into it and understand what's there and combine it with other pieces of information to help us put us in a position to be able to answer that question, which comes back to the whole put your strategy first thing, because if you don't understand the environment and what you can put together and the data that you can bring into it, then you're never going to ask a question that's going to bring you the true value that analytics can. Right, so I would assume another nice advantage is being able to pull in data from multiple sources, right, and kind of compare it to each other, such as, I used to be an IG person too, just like you, so we would have some folks come in with a fancy, you know, algorithm and try to figure out some fraud flags and correlate things, but, you know, say back to, you know, we had maybe there's a risk flag, people that tend to stay late and never take vacation, that's kind of something funny going on there, and then you might correlate that to financial transactions and some spikes and weird things and things happening at weird times of day, I don't know, just throwing it out there, but, I mean, is that something you guys run across too, kind of fraudulent? Yeah, it's more learning from what you already know.
So if you have some bad actors that you've already identified in the past, then what can you do, how can you go back, pull all the data that you had on that type of employee and then model it in a way so that if somebody starts to look familiar, if somebody starts to go down that path and they start to show the behaviors or the traits of someone who ended up to be a bad actor, how can we intervene ahead of time? Because in the oversight community, we're waiting until a crime has already occurred or fraud has already happened, but the agencies really have the power because if we can intervene before it gets to that point, not only is it better for the employee, it's better for the organization, we don't lose the productivity, and we just become more efficient at managing our resources. So pulling analytics up as far as you can in front of the organization and at the head of the program or the employee, you know, that's where we're going to get the biggest value is avoiding the problem to begin with. Right, and I think also, you know, that's kind of alluding to, okay, we have all these data sources and this fancy tool that can aggregate and spit out some, you know, some result, but I think you mentioned before that it's not necessarily technology is not the problem, it's more of a people issue.
You know, just like you had talked, I was listening to one of your other podcasts about how enterprise risk management is going to be successful, and it's the same formula with analytics. If you don't have executive buy-in, if you don't have leadership that invests in this and really prioritizes it as an organization, and you don't have a culture of learning and a culture of open communication, those are the things that are going to sink you quicker than anything. It's not a technology problem. The technology is out there, and a lot of times the technology is low cost.
Everybody's got a plethora of data, so that's really not the biggest problem either. It's not having effective leadership or not having a culture that understands the value of analytics. Analytics is a really sexy name and thing that people want to do, but the misperception that you can just go buy a product, bring it in and install it, and it's going to solve all my problems is really part of the problem. So yes, it's always the human side of it that's the most difficult to solve.
Right, so it's really, again, it's a couple things. What really is the problem we're trying to solve? Do we have the right information? Have you done your descriptive analytics, so to speak?
And then, yeah, I mean, do we have the leadership to help us make these changes and put the right processes in place and things like that, right? Correct, and descriptive analytics is a good place to start because, you know, you don't necessarily, I mean, like I said, everybody's talking about AI, you know, everybody's talking about machine learning and sensor data. Well, I think that's great, but the reality is a lot of times in government, we just need to do basic diagnostic and maybe even some. descriptive and diagnostic analytics.
There's really a low bar. A lot of times we'll come into an organization and just combining data sets together from different systems is a huge win. So just making yourself aware of what's going on in your environment more completely is a great place to start. And look, the advanced stuff is going to come down the road, but you kind of have to learn how to walk before you can run.
And why invest in all of this predictive analytics when you may not understand what it is, you may not even have the data or the problem set that lends itself to that kind of a solution. So yeah, let's just start with something descriptive, get a few wins, prove this out to the organization, help people understand the value of analytics. Then, as I was saying before, then the investment starts to come. Then the organization starts to see the value.
Then the different programs are coming to you and want to invest in you because they understand and they've seen the value of it. So you almost have to have a small business mindset if you're trying to start an analytics program in your organization and quit waiting for all the money and all the budget problems to work themselves out because we've been in Washington, D.C. long enough to know that more than likely that's not going to happen. So in your experience over the years, have you seen things go from the starting just doing ourselves with the tools we have all the way to some huge program?
So have you seen this occur or is this actually still kind of rare? I think it's kind of rare. And I say that to maybe comfort people because I think a lot of agencies think they're really far behind the other ones, but usually we see the same issues in the same agencies. The good news about this is that, like I said, there's a lot of talented people in the government.
There's a lot of good data in the government. There's a lot of good problems to solve in the government. So just putting all of that together and solving a quick problem and putting the pieces together in the right order, a lot of agencies are right there. They're not afraid of analytics anymore.
They have an idea of what it is and they've been trying to put the pieces together and they're not quite as successful as they could be. So it's really not a heavy lift as much as it was in the past for you to make a change here, add a skill set here, get some input here, utilize this tool a different way, put these things together that you already own and have access to in a practical way and you can get a quick win. So that's from the optimistic side of it is that people are really close to being able to do this well. Right, okay.
Well, let's talk a little bit about some of the laws and regs out there that also have spurred people to look more into data analytics. So we've got the Fraud Reduction Data Analytics Act. It's got data analytics in it. And then we have, what's that new one, evidence-based, foundations of evidence-based decision-making.
Yes. So have you seen that out there with some folks you work with? Is that part of the reason that they're looking to do this? Yeah, actually, Frida has been a great foundation from which agencies can leverage a regulation to say this is something that we have to do.
And I'm working at housing and urban development there and that's what they're standing on is Frida. And so, yes, having that law or regulation in place that you can point to in an organization and say, here's something that we can stand on that's going to allow us to give this a try. What we're really looking for in government, and when I say we, I guess I mean me, is there are a lot of rules and regulations out there that support this. It's just that a lot of times we don't understand them or we ignore them or we think we'll just wait until that goes away and not have to do anything about it.
So there are enough policies out there, there's enough new laws with the Data Act and with Frida that an organization can grab a hold of that, build yourself a charter and then you have something, a foundation from which to stand on in your organization because as I mentioned before, it's the agencies that have the biggest opportunity to improve what they're doing and a bigger return on investment. Don't wait on the oversight community. Don't wait for GAO to come in and tell you. You and I came from the IG community.
Let's put the IG community out of business and fix everything upstream as far as we can. So I'm really curious. I don't know how much you've really gotten into it but on this Evidence-Based Act, how do you envision agencies might want to do or what do you think they might want to do to start showing, getting true evidence for the successor of their mission or their program? I want to go back to what I talked about before about this being a leadership issue.
So many times, and I was an executive in a government organization and sat into a lot of senior staff meetings and executive meetings and time after time after time, week after week, month after month, the head of the agency was informed by PowerPoint. They were informed by the head of a program or the head of an organization who just knows all the answers and that answer, I'm not going to say that it's wrong but we got into a habit of allowing people to just tell us things or show us things on a big screen in a PowerPoint presentation and we believe that. So whoever's running that executive staff meeting, if you really want to make analytics a priority, you need to start asking, show me or why. What's behind that number?
And when you turn it on the direct reports and the executives to, they know that they're going to have to defend what they say, they're really not going to have much of a choice then to go back and look for data to bring into support what they're selling. So rather than me telling you there's a tool out there that I think you should do or a methodology or an approach, if it starts from the top of the organization and that kind of attitude and that kind of culture is put into place, very quickly you'll discover that you'll learn a whole lot more, they'll bring things to the boardroom that are backed up and can be supported by data and then all of a sudden you have some evidence-made decisions going on and that's, but really that's a, I mean I've got plenty of bloody noses in my career, you know, trying to make that happen, you're going to step on toes and you don't expect the ship to turn right away. It's a long, it's going to take a while but again I'm optimistic and I've seen a lot of organizations start to move that way and I'm happy to be a part of it. Yeah, I mean just a slightly techie thing here but I've seen some great tools out there, dashboards, you know like the tableaus, the BI, all the different intelligence tools out there.
I mean it's great especially if they have real-time data to get connections and things to systems which is sometimes hard but I mean instead of having a PowerPoint, I'm going to bring that dashboard up and say here, here's the graph and they click on that and it drills down and drills down and shows you yep, these are the reasons why this happened. You ask me any question you want to and we're going to click down into it and we're going to show you the underlying data and then we're going to learn something from that too. We shouldn't be afraid to acknowledge that we don't know everything. Well and something I love is when I'm in those meetings with executives and I'm here to present this but then they see the graph and they're like, why is that?
And you know, that's a question I hadn't even thought about. let's get back to you. Yeah, but still it's interesting. It is.
You see something from a fresh perspective you're going to see something that the people preparing and probably weren't thinking about. And that's good actually. It's not something you're scared of. It's something that's good.
You want to see something, a fresh perspective you know or a perspective from an executive that you may not have even thought about. Exactly. Because that's not your world. You're doing this every day and they're thinking about the bigger picture.
Yeah, that's right. You and I get it. I don't know why it's so hard. I don't know.
It's pretty simple to me. But all right. Well Brian, any last words? I really appreciate you coming on here to speak to us all.
Yeah, I mean. No, I mean, just to sum things up, this is not a hard problem to solve. There's a practical approach to it. You don't have to spend a lot of money.
You have talented people in your organization. You're sitting on a lot of data and if you're not sitting on it, there's a lot of publicly available data that we can use to solve your problems. So it's just putting the pieces together in the right order to take what you've already started, all the hard work that you've done to get you to this point and put you over the top and allow you to do not just descriptive but some more diagnostic work and then prescriptive and predictive down the road. So yeah, I guess if I had parting words, I would say, you know, let's be optimistic and I'd encourage everyone to stop and think about how you're doing it now and can we help you put the pieces together better.
Great, well, Brian, thanks for joining us. All right, thanks, Paul. Well, that's our show. Thanks for tuning in.
Check us out on Affirm.org. There's all our podcasts there. A couple fun ones coming up with HUD and USAID talking about risk appetite. So I hope you will join us.
So until next time, this is Paul Marshall signing off for the risk chance for the firm.