It is a slow Tuesday afternoon and the store is empty. Nobody is at the counter. Then the register prints a refund. Thirty eight dollars back on a carton of cigarettes, cash out of the drawer. The receipt looks exactly like every honest refund the store has ever done: a reason code, a dollar amount, a timestamp. The one thing the receipt cannot show is that no customer was ever standing there and no carton ever came back over the counter. That refund is a second transaction that never happened, and on paper it is invisible.
Most loss at a store is something that should have rung and did not, the item that walks out the door unpaid. Refund fraud is the mirror image. It is a transaction that runs backward. Instead of money coming in, money goes out, and a slip of paper says that was correct. It is worth understanding on its own terms, because a store can watch its shelves all day and still bleed at the one button that sends cash the other way.
A slice of every return is fake
Returns are enormous. The National Retail Federation estimates that 15.8 percent of annual sales will be returned in 2025, a total of 849.9 billion dollars, and that same report found that 9 percent of all returns are fraudulent. Nine cents of every hundred dollars that comes back is not a real return. Across the industry that works out to roughly 76 billion dollars walking back out as refunds that should never have been paid.
Appriss Retail, which studies this every year, put the figure even higher the year before. Its research found 103 billion dollars in losses tied directly to return and claims fraud in 2024, with 15.14 percent of all consumer returns classified as fraudulent. The gap between the two numbers is mostly a matter of how each study counts, but the direction is not in dispute. This is not a fringe problem, and it is not only organized crime. A Loop Returns survey found that nearly four in ten online shoppers say either they, or someone they know, has committed returns abuse or fraud in the past year.
A refund is a sale that runs in reverse
The reason refund fraud is so hard to catch is that it hides inside a normal, expected part of the day. Stores issue real refunds constantly. A customer changes their mind, a product is defective, a price rang wrong. So a refund is not a strange event the way a broken lock is. It is routine. That is exactly what makes the fake ones easy to bury: they look like the honest ones, because on the record there is nothing to tell them apart.
The tactics are familiar to anyone who has worked a counter. The NRF named the common ones: overstated quantities, empty boxes, and counterfeit items returned for a genuine refund. Appriss lists more of them, including the box of rocks where an empty carton or a worthless substitute comes back, price switching, and merchandise stolen off the same shelf handed back for cash. Every one of them ends the same way. A refund is issued, money leaves the drawer, and the item that supposedly came back either never existed or is worth nothing. And the hardest version has no customer at all: a staff member keys a refund into an empty store and takes the cash, with a clean receipt left behind to explain the shortage.
What the refund report cannot see
The instinct is to find this in the point of sale report. The report is honest about what it knows. It lists every refund that rang: the amount, the reason code, the register, the time. What it cannot tell you is whether the story behind the refund is true. It cannot see whether a customer was actually standing at the counter. It cannot see whether merchandise physically came back across it. It cannot tell a thirty dollar refund handed to a real person with a real return from a thirty dollar refund keyed into an empty store. From the register point of view, both are just refunds, and both look correct.
The camera has the opposite blind spot. There is almost certainly a camera pointed at the counter already, and it records whether a person was there and whether anything came back over the counter, in full detail. On its own it cannot read the transaction, so it has no idea a refund was ever issued. So the store owns both halves of the answer and still cannot see the loss. The register knows the refund and not the floor. The camera knows the floor and not the refund. Refund fraud lives in the gap between them, which is why a store can own a camera and a register and still pay the bill.
Closing the loop at the refund button
The thing worth building is the join. ARGUS runs what we call the closed loop. It crosses the camera with the point of sale, continuously, on the equipment a store already owns. At the moment a refund is issued it checks the transaction against the counter: was there a customer, did merchandise actually come back, does the reason on the receipt match what the camera saw. A refund with nobody at the counter is a mismatch. A refund with no returned item is a mismatch. Nobody has to accuse anyone first. The loop raises the flag on its own, attaches the clip, and ranks the events by how much they cost, so a full day becomes a short list a manager can review in a few minutes.
Because a refund moves cash out of the drawer, this is where several of the store's numbers meet at once. The refund shows up against the drawer as a Cash Anomaly Index. It shows up against the person who rang it as a Workforce Honesty Score. And it lands in the daily Revenue Integrity Score, the single figure that tells an operator whether the money that left the building today was supposed to. Instead of finding a run of odd refunds weeks later, buried in an export, the operator sees it on the day it happens.
None of this asks a store to buy new cameras or replace a register. It asks the two systems already on the wall to do together what neither can do alone: tell a real refund from a second transaction that never happened. We are in private beta with convenience, gas station, and grocery operators. If refunds you cannot explain are quietly draining the drawer, talk to us or write to business@useargus.co.
Sources
- National Retail Federation: Consumers Expected to Return Nearly 850 Billion in Merchandise in 2025, reporting a 15.8 percent return rate, 849.9 billion dollars in total returns, and that 9 percent of all returns are fraudulent.
- Appriss Retail: What Is Return Fraud, on the common types of return fraud and its finding of 103 billion dollars in return and claims fraud in 2024, with 15.14 percent of returns classified as fraudulent.
- Loop Returns: The State of Returns Fraud and Abuse, reporting that nearly four in ten online shoppers say they or someone they know has committed returns abuse or fraud in the past year.