A count night looks about the same in every store. The doors lock a little early, a crew comes in with scanners, and for a few hours people walk the aisles reading barcodes off shelves that are normally being shopped. In the morning a number lands in the back office. It says the store is short by some amount of money. It does not say when the shortfall happened, or how, or who was on shift.
That is the strange thing about how retail measures loss. The count is careful work and it produces a real figure. But the figure is a residual. It is what is left after you subtract what is on the floor from what the system believed should be on the floor, and it quietly blends every possible cause: product that was stolen, product that was damaged and thrown out without a note, a case of soda that was delivered short, a scan that rang the wrong item, a promotional price that was never entered. One number, many stories, all of them months old.
The count answers a question you asked too late
The industry figures come from the same place. When the National Retail Federation published its last full National Retail Security Survey in September 2023, it put the average shrink rate at 1.6% of sales for fiscal 2022, up from 1.4% the year before, and translated that into $112.1 billion in losses. Those are counted numbers. They exist because thousands of stores did exactly what the store above did and reported the gap.
A year later the federation stopped publishing the report after 32 editions. Retail Dive covered the decision in October 2024, and the reasoning is worth sitting with. The NRF said a broad study of shrink no longer captured the key challenges retailers face, and an asset protection expert quoted in the piece argued that shrink is ambiguous by definition, because retailers do not hold item level financial data behind the gap. The industry standard measurement was retired in part because of what the measurement method could not say.
None of this is an argument against counting. Somebody has to reconcile the books, and audit and tax filings require it. The point is narrower. A residual computed once or twice a year is a poor instrument for running a store day to day, and most operators already suspect as much.
Accuracy decays the moment the crew goes home
Inventory people have been writing about this for a long time. Jon Schreibfeder, who has spent a career on inventory accuracy, describes the annual physical as a comprehensive but disruptive snapshot that is essential for audit and tax, while noting the part everyone in the building already feels: accuracy decays quickly after the count. His recommended answer is cycle counting, small continuous counts spread through the year, so the books do not drift while you wait for the next full physical.
Cycle counting genuinely helps. It is also still counting. It is still hands and scanners and hours, and in a convenience store with thousands of fast moving items it competes with every other thing a two person shift has to do on a Saturday. In practice it slips, and then it slips again, and the drift comes back.
Meanwhile the loss itself is not periodic. It happens on a Tuesday at four in the afternoon, in a specific aisle, at a specific register, during a specific shift. By the time a count discovers it, the schedule has turned over a dozen times, the footage has been overwritten, and nothing in the number points you anywhere. You are handed a symptom with no case history attached.
What changes when the number is current
Think about a smoke detector. A detector is not more perceptive than a person. It is just always on. Walking your house once a year looking for evidence of a fire would be diligent, careful, and useless. The entire value of the detector is that it reports while the event is still happening.
That is the design idea behind what we call the closed loop. Argus joins four streams a store already produces and almost never compares: cash, point of sale, camera, and labor. Joining them continuously turns a discrepancy into something you observe as it occurs rather than something you infer later from a shortfall. The camera saw product leave the cooler. The register has no matching transaction in that window. The schedule says who was on. The three facts arrive together, so the event carries its own explanation and its own clip.
From that join we compute a Revenue Integrity Score, one figure per store per day, answering a simpler question than the stocktake does: of everything that left this store, how much was actually paid for? It is not a shrink percentage and it does not replace the annual count for accounting. It is the operating number. When it drops four points, you can open the events that moved it this morning, while the shift is still on and the person who can explain it is still standing behind the counter.
The real difference is not precision. It is latency. A yearly figure can only tell you that something went wrong. A daily one tells you what is going wrong, and that is the only version an operator can act on.
Why the lag hurts small operators most
Convenience retail is not mostly chains. NACS counted 151,975 convenience stores in the United States as of December 31, 2025, and 63% of them belong to companies that run ten stores or fewer. Those operators do not have an asset protection department, a data team, or the spare hours to run a disciplined cycle counting program. What they have is a camera system, a register, a schedule, and a nagging sense that something is off.
A large chain can absorb a slow measurement because it employs people whose job is to chase the number until it explains itself. A single store owner cannot. For that owner, a measurement that arrives once a year is close to no measurement at all, because by the time it lands there is nothing left to do with it except feel bad. Continuous measurement is the only kind that fits how they actually work.
What we are not claiming
A daily score does not make counting obsolete, and we would be suspicious of anyone who told you otherwise. The physical count still sets the accounting baseline, still catches receiving errors that no camera would flag as suspicious, and still belongs in the year. What a daily score changes is what the count finds when it comes around. Fewer surprises, and the ones that remain are mostly paperwork rather than mystery, because the events that would have become mystery were reviewed the week they happened.
The honest framing is that the stocktake was never designed to be an operating control. It was designed to close the books. Retail has been using it as both for decades because there was nothing else available at the store level. There is now, and it runs on equipment stores already own.
Argus is in private beta with convenience, gas station, and grocery operators. If you would like to see what a daily integrity number would say about your stores between counts, talk to us or write to business@useargus.co.