The returns-fraud number can’t hold still.

Retail’s most-cited returns statistic is that fraud cost about $103 billion in 2024. That figure was published two different ways in the same year, from bases $205 billion apart. Then it “fell” by roughly a quarter in 2025 while every signal beside it rose. The number is not measuring behaviour. It is measuring how the number was made.

The series does not hold together

Look at the two 2024 rows. Two official totals for the same year, $205 billion apart, a gap larger than the entire fraud problem being measured. Both land on exactly $103 billion of fraud through offsetting maths: a larger base times a smaller rate, and a smaller base times a larger rate.

The total-returns column lurches too, across years in which e-commerce grew every single one. Real behaviour is not that jumpy. The lurching is the measurement moving, not the world.

Why the number is built to move

The fraud rate is self-reported perception, with no shared definition of fraud: one retailer counts wardrobing and bracketing, another counts only empty boxes and receipt fraud. Much of the fall to 9% in 2025 is definitional, and NRF says so itself: its 2025 report notes the years are not directly comparable. When the publisher tells you not to compare years, the series is not a trend.

The dollar figure is that rate applied to a Census sales total, so two noisy estimates are multiplied and reported to three significant figures. And the panel is 60 of the top 100 US retailers: big-box chains with in-store returns, a different channel and a different customer from an apparel brand doing mail-back returns off Shopify.

In the same 2025 report, the directional signals point the other way: overstated-quantity claims up at 71% of retailers, empty-box returns up at 65%, and 93% calling fraud and abuse a significant problem. A number that moves opposite to everything it describes is telling you about its method.

The part you can measure

Returns fraud cannot be measured from the top because a return is a customer-level behaviour reported as an aggregate. Your own blended return rate fails the same way inside your business: your best customers and your worst customers can both return a lot, and the rate cannot tell them apart. Hard fraud, the empty box and the counterfeit swap, belongs to your payments stack. Soft abuse, the wardrober and the serial bracketer, is invisible to fraud tools, because every event is a legitimate customer making a legitimate return. The pattern only exists at the customer level, across orders.

Run it yourself. The Returns Self-Audit prices one return at your costs and scores a customer on net contribution after returns, with the four profiles and the move for each.

Sources