Ep. 19 Part 1: 1 in 48 — Why Hasn't the Needle Moved?

From the floor of RILA RAP 2026 in Phoenix.

15 years of better cameras. 15 years of better analytics. 15 years of faster alerts. And the retail theft arrest rate is still roughly 1 in 48.

DJ Carr took that question to the people building the technology. In Part 1, he sits down with Jason Davies of Sekura Global, whose answer points past hardware tagging at the same structural problem detection alone cannot solve.

Then DJ and Jim Cords debate where they diverge. Jim brings 28 years across FBI, DHS, and OIG. He'd start with funded enforcement. DJ brings private-sector roots in ground-level security and several years as a systems architect. He'd start with the data layer. They do not resolve it.

They agree on one thing. A system that knows and cannot act is not a security system. It is an evidence system with no consumer.

Part 2 drops next week with Chris Ochs of SAFR.

Loss. Liability. Law.

Transcript

SPEAKER_02 0:02

This is the City Safe Podcast, a conversation at the intersection of community safety, technology, and leadership. Co-hosted by Don Carr and Jim Kortz. Together, we examine the critical issues facing cities today and the smarter tools that can help reduce crime and protect our communities. From instant communication systems to emerging technology, we break down what works, what doesn't, and what's next for urban safety. Because in today's world, keeping people safe requires more than good intentions. It requires innovation, data, and decisive action. Listen to the City Safe Podcast, available now on Spotify, Apple Podcasts, and YouTube Music. Subscribe today and join the mission to make cities safer for everyone.

SPEAKER_01 0:49

I'm your co-host Don, and I come from the private security side, running and operating private security guard firms.

SPEAKER_03 0:57

And I'm your co-host Jim, and I come from over 28 years on the federal law enforcement side.

SPEAKER_01 1:04

So, Jim, 15 years of better cameras, 15 years of better analytics, and 15 years of faster alerts.

SPEAKER_03 1:15

Yeah, and the arrest rate for retail theft is right where it was.

SPEAKER_01 1:20

Right. Roughly one in 48, depending on the source. But that's the number that this industry usually hovers around. One in 48 arrests.

SPEAKER_03 1:32

Right. So I I think you, DJ, you you took a question to the people building the technology, right? A simple one.

SPEAKER_01 1:39

Why? Their answers were exactly what we expected, Jim. Just articulated through their lens. And that's exactly why we wanted to hear from them. So a couple of weeks ago, I was on the floor at the Rela Asset Protection Conference 2026, right here in Phoenix. Four days, hundreds of vendors, some of the sharpest LP minds in retail under one roof. But Jim couldn't make it out this year. So I went solo with the podcast mic.

SPEAKER_03 2:10

Right. Which means I'm hearing some of these conversations the same way your listeners are. Fresh with no how to start.

SPEAKER_01 2:18

Right. And I want to do something at the top of this episode that I don't always do, Jim. I want to disclose where I'm coming from because the framing of this episode points at the gap my own company works on. So you should hear that out loud before we go any further.

SPEAKER_03 2:36

Okay, I think that's important.

SPEAKER_01 2:38

So I've spent the past six years on the architecture side of the problem. I'm not a neutral observer. So when I tell you the gap is structural, I'm telling you what 15 years in this line of work has shown me, not what a research paper told me to believe. So you should weigh what I say against that.

SPEAKER_03 3:03

Um yeah, and I I think my lens is a little bit different, right? So I spent a career uh doing federal investigations. So when DJ talks architecture, I tend to think more about the case file, right? And whether or not a particular prosecutor will pick it up. So different angle, but the same problem.

SPEAKER_01 3:22

We're saying because we're gonna disagree on at least one thing in this episode. And I think you will hear where. Maybe even more than one. Probably more than one. Okay, so setting the table. I walked in to the Rela floor with one question. The same question. I put it to the solution providers who were able to take a few minutes and chat. And I asked, okay, over the last 15 years, the industry has spent billions on retail crime technology. Better facial recognition, better gesture detection, better analytics, better integrations, better dashboards. Yet the percentage of retail theft incidents that result in an arrest is still about one in 48. So why has that needle not moved?

SPEAKER_03 4:17

Yeah, I think that's a fair and honest question.

SPEAKER_01 4:20

So the answers I got, Jim, were not surprising. Um they weren't surprising to me, and they're not surprising to Jim. Um really not to anyone who's been in this work at this layer for as long as we have. But hearing them say it out loud, the people who actually built these tools, that was a different kind of confirmation. So that's what this episode is. So before we run the first clip, I want to be honest about the one in 48 number because numbers get loose in this space, and I want to be careful with it. So the one in 48 figure traces back to NASP, that's the National Association of Shoplifting and Theft Prevention Specialist. It gets quoted by NRF um and other industry research firms by basically everybody in the space. But it's the most cited stat in the LP world. It's not the only stat by any means, but the Capital One shopping research, drawing on a different methodology, also has a rate, but they put their rate closer to one in 100. Okay, wow, okay, yeah, that's a bigger gap. Much bigger gap. And I should also say there's a real critique of how the retail industry reports its own numbers because the Center for Just Journalism has written pretty carefully about this. And they point out that police reported crime data is a better source than industry surveys, even though it's also imperfect. So in 2023, the NRF uh retracted its widely circulated claim that organized retail crime accounts for nearly half of total shrink. So that retraction is the part as to why I'm being extra careful.

SPEAKER_03 6:10

Yeah, I think that's the right way to handle it. The number is directionally correct, even if the precision is debatable. The order of magnitude, I think, is what matters here.

SPEAKER_01 6:21

Right, Jim. So whether it's uh one in 30 or 1 in 60 or 1 in 100, the picture's the same. The vast majority of retail theft is not generating a case, it's not generating a charge, it's not generating a record that anyone can act on defensively later.

SPEAKER_03 6:42

And that's the problem, right? Because what we know from the federal side, you know, from organized retail crime task forces, is that the same offenders are hitting the same chains over and over and over.

SPEAKER_01 6:55

Right. So, Jim, walk the listeners through that data because that data's pretty striking.

SPEAKER_03 7:01

Yeah, so the Manhattan Institute, uh, they pulled this together a couple years ago, and the numbers are, I think, are are pretty interesting. So the NYPD reported that 327 individuals accounted for 6,600 shoplifting arrests in 2022. So 30% of all shoplifting arrests in New York City came from fewer than 400 people.

SPEAKER_01 7:27

Wow. 327 people in a city of what over 8 million?

SPEAKER_03 7:32

Yeah, so Manhattan's uh DA's office, they reported that about 18% of shoplifting arrestees, uh, they count for about 42% of arrests uh across a six-year window. Seattle reported 168 people responsible for 3,500 misdemeanor referrals over a five-year period.

SPEAKER_01 7:55

And that's the pattern.

SPEAKER_03 7:57

Yeah, right, that's the pattern. Uh, but that's what gets ignored when every alert is treated as a one-off transaction.

SPEAKER_01 8:05

Well said, Jim. So when I asked the vendors why the needle has not moved, I was not asking why detection has not improved. Detection has improved. I was asking why detection improving has not translated into outcomes. And that's an entirely different question.

SPEAKER_03 8:25

Yeah, different question entirely, exactly.

SPEAKER_01 8:28

So let's run the first one. Should we go here with Jason Davies of Sapira Global? Jason, thanks for joining us on the program. The preface is 15 years of better technology, faster detection, smarter analytics, earlier intelligence, depending on the source of one in 48 incidents. And in an unrest. And approximately half of those end up getting prosecuted. So the technology keeps improving, the insights are getting better and better and have never been better. Yet the needle on the outcomes don't tend to move.

SPEAKER_00 9:17

Well, I think there's multiple issues there. So from an ORC standpoint, you know, you have some companies that do have ORC teams that operate uh with law enforcement, with the prosecutor's office. They're following it up from you know the instant, you know, to conviction. Alright, and that's very, very important in running ORC investigations. We just don't hand off these cases to law enforcement or the prosecutor. We are an investigative partner. We help from you know setting everything up for the court hearings, providing evidence, providing video uh evidence as well, any technologies that were used with data analytics that can actually help prove out the losses to that individual or to that state, its defensive location, whatever it may be. So I think, you know, that goes into play there, and then also the teaching and training of law enforcement and the project on basically what we've done over the years have proven to be, you know, tactics that can get convictions. Um now in some cases we have laws that prevent us from being being able to do that. Uh pop 47 uh basically tied the hands of law enforcement and the prosecutors that came back with uh 36, uh, which you know heightened, you know, uh heightened the bill for retail benefit for law enforcement to be able to go after specific individuals to product that's a product. Um part of that was also uh making available two law enforcement training modules, um, you know, uh additional uh resources such as uh FH uh sorry, CHP's um task forces being able to have them in more than just two or three locations that was expanded to uh I believe five throughout California, and dedicated ORC prosecutors, prosecutors that knew how to to really really push a case through to fruition. Um I think the other piece of that, if we step back away from the law enforcement, prosecutor, and investigative side, you got retail companies that do not have the resources to go after these individuals, right? They'll file their police reports, and that's it. They never hear back from from law enforcement uh because they're not following up with law enforcement. You have to have that relationship. So if you don't have an ORC, you know, you still need to have somebody that's building relationships with the local law enforcement, local officers office to bring back to those employees because the employees want to know, hey, I filed this report, whatever happened. If it becomes a black hole to them, then the quality of information when the report is gonna be diminished, uh the frequency of them making reports is gonna be diminished, and the frequency of them calling law enforcement when they're supposed to will be diminished. We saw that for the pandemic. So uh there's a multitude of issues there, and I think it all starts with education. Educating the employees on what happens with their um their their police reports, educating the law enforcement on how an impacts we tell and how to investigate it, and then uh teaching and training prosecutors with local, state, or federal on the impact of RC to the community as a whole, not just to the retailers, um, and then try to do tactics to actually go after um multi-state actors or large fence.

SPEAKER_01 12:43

Understood. Strong signal, right? I mean, many layers to what appears to be a simple question. Um, not a simple question, and certainly appreciate your lens and your perspective on it. Um it's meaningful and it's very layered. So, Jim, what's your first reaction?

SPEAKER_03 13:02

Oh, well, honest answer is he named it. The technology produces the alert. The alert lands somewhere that doesn't have the authority, the time, or the legal cover to do anything with it.

SPEAKER_01 13:16

And that's the part I keep coming back to. Because detection without intervention authority, that's not a security improvement. It's really just a paperwork improvement.

SPEAKER_03 13:27

And maybe not even that, right? Because at least paperwork creates a record. A lot of these alerts um don't even live long enough to become a record.

SPEAKER_01 13:36

That's a point, Jim.

SPEAKER_03 13:38

Yeah, and here's the kind of the federal lens on it. The single hardest thing to do in a federal case is building the timeline, right? Yeah, the same offender, same MO, multiple sources, multiple jurisdictions, but to prosecute that as anything more serious than a series of misdemeanors, you have to assemble that pattern deliberately and across sectors.

SPEAKER_01 14:00

Okay, I want to hand this one to you first, Jim, because I think this is where you and I are going to diverge.

SPEAKER_03 14:07

Okay, yeah. So what I heard him say was that the prosecution threshold is the wall. And I have to agree, right? Because in a lot of jurisdictions, everywhere I've worked, the the thresholds have been raised. Um, some places have raised it a lot. Uh the practical effect is that when the case is built, even when the evidence is clean, the system declines to charge because of those rising thresholds.

SPEAKER_01 14:33

Right. And technology cannot fix that.

SPEAKER_03 14:36

Right, technology can't fix that. And here's where I want to push back on you, DJ. Because it's it's not just on tonight's argument, it's sort of on the whole frame.

SPEAKER_01 14:46

All right, go ahead.

SPEAKER_03 14:47

So you've been saying architecture first for as long as I've known you, and I get it. I understand that logic absolutely, but I want to put a couple things on the table that I think the listeners should hear. All right, so first, if we look at California, um they stood up a dedicated organized retail crime task forces led by uh the California Hyper Highway Patrol, CHP, uh, and they had about two years worth of results. 29,000 arrests, 226 million dollars in recovered stolen goods. And that happened because somebody funded law enforcement, right? Somebody funded that enforcement. So that happened because there were a lot of bodies and a budget pointed at that problem. Beautiful pattern recognition system in the world, and if the prosecutor won't file and the police won't respond, then really you built a very expensive filing cabinet. But California shows what funded enforcement looks like.

SPEAKER_01 15:44

That's a super strong argument and perspective, Jim.

SPEAKER_03 15:48

Yeah, and a second thing, so I'm gonna say this carefully because I respect you. You disclose at the top of this of this episode that the architectural framing of this problem is also the commercial framing of the company you run. The listener should hold that in mind right now, because when you tell us the gap is architecture, you're also telling us where you sell. That's a fair thing to put on the record. I'm not saying you're wrong, I'm just saying the listener should weigh it.

SPEAKER_01 16:16

And I appreciate you saying it. And let me try to answer both because they're both the right questions. All right, go ahead. In California, you're right that it is the strongest counter argument. But I want to be careful about what California actually shows. Because that 29,000 arrests over two years, that's real. The recovery of stolen goods is also real. But what it does not show, at least not yet, is whether the underlying pattern of repeat offenders changed. Whether that 327-person dynamic in NYPD's data shifted, or whether the same people who were driving the volume before the task forces stopped driving it after. The arrest went up, that was established. So whether the structural problem of pattern blindness was solved, that's a different question.

SPEAKER_03 17:14

That is a different question.

SPEAKER_01 17:16

And here's where I think that data layer matters because if California is arresting more people, but the same small core is still driving most of the activity, all you've done is increase the throughput of the misdemeanor pipeline. Right? You've not changed the structural problem. You've funded enforcement against the symptom. So to know whether the pattern actually shifted, you have to be able to see the pattern. And that is the data layer.

SPEAKER_03 17:45

Okay, I hear what you're saying. Not sure I'm fully convinced because California did something. I think they did something big. And the data layer you're describing is theoretical until somebody builds it. Fair enough. And it's worth saying out loud. There is a real version of this argument where the listener says, look, California funded enforcement and got measurable results. And DJ is asking us to fund a data layer that has not been demonstrated at scale. It's a fair comparison.

SPEAKER_01 18:16

It is a fair comparison. And I won't run from it. Okay, good. So on the COI point, you're right. The architectural framing is my commercial framing. I don't think it makes it wrong, but I will not pretend that the two are independent. They're not. I would believe what I believe even if I had no commercial position in it, because I believed it before I had a commercial position in it. But the listener does not have to take my word for it.

SPEAKER_03 18:45

Okay, so I'm gonna land somewhere different from you, DJ. I think the architecture is necessary. Uh I'm just not sure it's sufficient. And I'm not sure it's the thing that would move the needle fastest if you handed me a billion dollars and told me to fix the one in 48 problem in five years. Well, what would you do? Honestly, I would put half of it into prosecutor and law enforcement capacity. So ORC task forces, dedicated retail crime units, enough bodies to take the cases. California showed that works. Um, I think I'd put the other half into that data layer that you're describing. But I would do enforcement first because I've seen what happens when you build evidence systems that nobody is funded to use. They become very expensive shelves.

SPEAKER_01 19:32

That's a point. But I would do the data first because California is one state with a particular political environment that funded a particular response. And in my opinion, you can't replicate that nationally on the political timeline that anyone's willing to wait. So you can build the data layer everywhere. And once that exists, it makes the case for enforcement undeniable. Even in jurisdictions where today the case is invisible.

SPEAKER_03 20:05

Okay, good point. But I think we probably disagree here. We disagree indeed, and I love it. Yeah, but it's you know, it's worth letting that sit.

SPEAKER_01 20:15

Absolutely. So here's the thing I know we agree on, even with the disagreement on order. The current state, the state where the technology has improved and nothing else has, is the worst possible configuration. You have built a system that knows and cannot act. That's not a security problem. That's an evidence system with no consumer.

SPEAKER_03 20:41

Yes, that part we agree on.

SPEAKER_01 20:43

100%.

SPEAKER_03 20:45

And the listener gets to decide which one of us is closer to right.

SPEAKER_01 20:49

And Jim, before we wrap it up, what's the line you want the listener to take with them?

SPEAKER_03 20:54

I think it's that patterns matter more than incidents. So if your loss prevention program is producing incidents and not patterns, you're not feeding the only part of the system that can actually generate an outcome. That's what I think is important. That's a good line. What about you?

SPEAKER_01 21:15

Well, mine is the one I keep coming back to. Uh the the technology is not the problem. The technology is only a part of the chain that has improved. Which means every other part of the chain is the problem. And we are not going to fix that with a better camera. Very well said. So thanks to every solution provider on the floor at Rela this year who gave us some of their time, Chris Ox at Safer, Jason Davies at Secure Global, and a number of others who also took the time to share their perspective.

SPEAKER_03 21:53

Fifteen years of better cameras, fifteen years of better analytics, fifteen years. Of faster alerts.

SPEAKER_01 22:01

And the arrest rate? Well, still hovering around 1 in 48.

SPEAKER_03 22:06

And you've now heard the people building the technology say why.

SPEAKER_01 22:10

And you've heard us disagree on what to do about it, or rather, in which order to do about it.

SPEAKER_03 22:16

The argument isn't over. If you want to push back on me, on DJ, or on both of us, the email is in the show notes.

SPEAKER_01 22:24

So, until next time. Stay alert, stay well. And stay city safe.

SPEAKER_02 22:30

That wraps up this episode of the City Safe Podcast. Thanks for joining us in the conversation about smarter strategies for safer cities. If you found today's discussion valuable, share it with your network and help us spread awareness about the tools and technology making a real impact. Be sure to subscribe on Spotify, Apple Podcasts, and YouTube Music so you never miss an episode. Until next time, stay informed, stay connected, and stay city safe.