It is a little after seven in the morning and the store is quiet. The overnight person has gone home, the coffee is brewing, and the manager sits down in the back office to do the part of the job nobody trained them for. On the screen is yesterday. Forty one flagged transactions. Sixteen cameras that recorded all night whether anything happened or not. Somewhere in that pile is the reason the numbers did not tie out on Friday, and the truck arrives in about ten minutes.
That is the daily loss review as most stores actually run it. Not a formal investigation, not a case file. Ten or fifteen honest minutes, squeezed between a delivery and a shift change, done by someone whose real job is running the store. Any approach to loss that assumes more time than that is not describing a convenience store.
The report is not the problem. The pile is.
Exception reporting is old and it works. You take the stream of transactions, you flag the ones that break the pattern, and you look at those instead of all of them. Voids over a threshold. Refunds with no customer. Drawer opens with no sale. Discounts stacked on the same loyalty number. Nobody in loss prevention needs to be sold on the idea.
The trouble starts one step later. A flag is a question, not an answer, and the questions arrive faster than anyone can answer them. DTiQ puts it plainly in its guide to exception based reporting: with so much data to sift through, many managers do not have time to review daily reports from point of sale systems. The report gets generated. It does not always get read.
When it does get read, the next step is worse. A flagged transaction sends you to the footage, and the footage is where hours go to die. Writing for Loss Prevention Magazine in December 2025, Lauren Fritsky quoted a former convenience store loss prevention professional on what the work used to feel like: "As a c store LP professional, I spent a lot of hours looking at unproductive video and unproductive leads." That is the honest version. Most of the searching finds nothing, and you cannot tell which searches those will be until you have already done them.
The industry answer has usually been more people, and there are not more people. When the National Retail Federation published its retail theft and violence study on October 28, 2025, it reported an 18% increase in the average number of shoplifting incidents per year in 2024 versus 2023, and found that 64% of retailers reported less than half of their store related theft incidents to law enforcement. Among the top reasons retailers gave for not being able to investigate or prosecute was limited retail asset protection resources. The cases are not being dropped because nobody cares. They are being dropped because nobody has the hours.
What a good review actually looks like
Turn the problem around. Instead of asking how the manager can review more, ask what would have to be true for a useful review to fit inside ten minutes.
It has to be ranked. Not forty one flags in timestamp order, but a short list in order of money at risk, so that stopping halfway through still means the important things got seen. A review that has to be finished to be worth anything will not get done on a truck day.
It has to be attributable. Each row should already say who was on the register, when, how much, and how it compares to that person's normal. If the manager has to open a schedule in another tab to find out who was working, the review just became a research project.
It has to carry its own evidence. The clip belongs on the row, cued to the moment, eight or ten seconds long. Not a camera name and a timestamp to go look up. The single largest cost in the old workflow is the trip from the flag to the footage, and that trip should not exist.
And it has to end in a decision. Four are usually enough: coach it, watch it, escalate it, or dismiss it. Every one recorded, including the dismissals, because a pattern of dismissed flags on the same register is itself a signal. What the manager does in those ten minutes is decide. Not search.
Attribution is what makes it short
The reason most reviews run long is that the flag arrives without context, so the manager has to rebuild it by hand. A void at 2:14 in the morning is not evidence of anything. Registers void things. Customers change their minds.
The same void looks different when the other systems weigh in. If the camera shows nobody at the counter. If the drawer opened afterward and the cash count came up light by roughly the voided amount. If it is the fourth one that week on the same login, all on the overnight shift, when the store average is under one. Those four facts live in four systems that historically never spoke: cash, point of sale, camera, and labor. Joined, they turn forty one flags into three events worth a manager's attention, and thirty eight that quietly answer themselves.
This is the part we spend most of our time on at ARGUS, and it is less glamorous than it sounds. The agents run on the cameras and registers a store already owns. Most of the work is the join, continuously, so that a flag shows up already carrying the answers to the first four questions a manager would have asked.
It has to survive a manager change
One more constraint, and it is the one most loss programs ignore until it breaks them. The person doing this review will probably not be the same person in eighteen months.
NACS reported in January 2026 that among top decile convenience companies, the ones performing best on operating profit, employee turnover ran 60.8% and manager turnover 17.1% in the first half of 2025. Those are the good numbers. Industry wide, employee turnover typically tops 100%.
So a review routine that depends on one experienced manager knowing which registers to watch and which vendor always shorts the milk order is not a routine. It is that person, and they are leaving. The test of a daily review is whether a competent new manager can be handed it on their second day and run it correctly on their third. Short, ranked, evidence attached, four possible outcomes. That survives a handover. A folder of tribal knowledge does not.
Worth saying clearly: most of what surfaces in a good daily review is not theft. It is a new hire who was never shown the refund policy, a promotion that was priced wrong, a vendor whose delivery keeps landing short. Treating every flag as an accusation is how these programs lose the staff they depend on. The point is to find the small problems while they are still small and still fixable, which is the same reason to do it daily rather than at the annual count.
If you run a store or a few hundred of them and this sounds like your morning, we would be glad to hear how you handle it now. ARGUS is in private beta. You can talk to us or write to business@useargus.co.