Return fraud in footwear: sizing abuse vs genuine fit problems
Footwear has the worst return economics in apparel, and most of it is not fraud. Shoes are hard to buy online: sizing runs vary by brand, widths are inconsistent, and a half size means different things across manufacturers. High return rates are structural. The scoring problem is separating the customer gaming the system from the customer whose feet simply do not match the size chart.
Treating all multi-size orders as abuse punishes honest buyers in the category where honest buying looks the most suspicious. The signals exist to tell them apart, but they are footwear-specific.
Why footwear bracketing is usually honest
Ordering two sizes of the same shoe and returning one is the rational response to inconsistent sizing, and in footwear it is the norm rather than the exception. A customer who buys an 8 and an 8.5 in the same sneaker, keeps one, and returns the other within a week is doing exactly what the size chart failed to help them do. Penalizing this behavior trains customers to buy from competitors with better fit guidance.
The pattern to watch is the ratio over time. Occasional bracketing with fast returns and one kept pair is honest shopping. Bracketing on every order, across many styles, with low keep rates and slow returns starts to look like a rental operation. The difference is in the longitudinal pattern, not any single order.
Wear patterns separate abuse from fit problems
The strongest signal in footwear returns is condition on arrival. Genuine fit-problem returns come back unworn: soles clean, insoles uncreased, no outdoor wear. Abuse returns show wear: creased toe boxes, dirty soles, stretched uppers. A returns operation that photographs and grades condition on intake has the single best abuse signal in the category.
Weight this signal heavily in scoring. A customer with a 60 percent return rate whose returns all arrive unworn is a fit problem, not a fraud problem. The fix for them is better size guidance, not a warning email. A customer with a 30 percent return rate whose returns arrive worn is the one the policy should address.
Size consistency across brands is a signal
Honest fit-problem customers are consistent: they order the same size across brands and return the pairs that do not match their feet. Their order history shows a stable size with variation only where brands run large or small. Abusers show no such consistency, ordering wide size ranges in the same style or sizes that do not match their purchase history.
Cross-reference the customer's kept sizes. If they keep a 9 in three brands and return a 9 in a fourth, the fourth brand runs small. That is product intelligence, not a fraud signal, and it should feed back into size recommendations instead of the risk score.
Policy responses that do not punish honest misfits
The goal is reducing returns without accusing good customers. Fit-finder quizzes, brand-specific size guidance ("this brand runs half a size small"), and free exchanges for size swaps all cut the honest return rate without touching the abuse rate. Exchanges deserve special handling in scoring: a size exchange is the opposite of a loss, it is a retained sale, and scoring should treat it that way.
Reserve the fraud response for the wear-pattern cases: worn returns, repeated wardrobing signals, multi-size orders with nothing kept. Those customers get warnings and restrictions. Everyone else gets better fit tools. Footwear brands that score this way keep the honest majority buying while the abusers find softer targets.