Wardrobing vs friendly fraud: separating the signals

October 10, 2026

Every return arrives with a reason code, and almost nobody trusts them. Customers pick whatever option gets the return approved fastest, which means the codes measure friction, not truth. But distrusted data is not useless data. Read at the aggregate level, reason code distributions reveal wardrobing patterns that no individual code admits to. The trick is knowing what to look for.

October 10, 2026

Every brand knows its return rate. Almost no brand knows its wardrobing rate, the share of returns that were worn before coming back. The two numbers get conflated constantly: a 25 percent return rate is treated as a 25 percent abuse problem, or wardrobing is dismissed because "most returns are honest." Both mistakes come from the same gap. The return rate is measured; the wardrobing rate is estimated. And the gap between them, read correctly, tells you where the abuse actually concentrates.

Wardrobing means wearing an item and returning it; friendly fraud means keeping the item and claiming a refund anyway. Both show up as returns with complaints, but they are different behaviors with different tells and different fixes. Treating them as one category means applying the wrong remedy to half your problem.

Why they get confused

Both produce the same surface pattern: a customer reports a problem and gets money back. In wardrobing, the problem is real in the sense that the item was used and no longer wanted; the fraud is in the return itself. In friendly fraud, the problem is invented or exaggerated and the customer keeps the goods. From the refund log alone, the two are nearly indistinguishable, which is why teams that only look at refund reasons miss the split entirely.

The wardrobing tells

Wardrobing leaves physical evidence. Returned items show wear: deodorant marks, stretched seams, faint odors, missing tags. The timing clusters around events, weekends, and holidays, because the item was bought for an occasion. The customer often has a clean payment history and no chargebacks, because wardrobers think of themselves as clever shoppers, not criminals. Size and style consistency matters too: wardrobers return the same categories repeatedly, usually occasion wear.

The friendly fraud tells

Friendly fraud leaves data evidence instead. The customer claims non-delivery while tracking shows delivery, or claims damage on items that arrive intact. There is often a dispute or chargeback rather than a clean return, because the goal is keeping both the item and the money. The pattern skews toward high-value items where the payoff justifies the lie, and the accounts may be newer or have thinner history, since established customers have more to lose.

Why the distinction matters

The remedies are opposites. Wardrobing is deterred by friction on the return side: inspection, restocking fees, wear detection, return limits. Friendly fraud is deterred by friction on the claim side: delivery confirmation, photo evidence requirements, dispute representment. Apply return friction to a friendly fraudster and they simply switch to chargebacks; apply claim friction to a wardrober and you punish a customer who might have been retained with a warning.

The measurement differs too. Wardrobing rate is a returns-analytics problem, measured through inspection outcomes and wear signals. Friendly fraud rate is a payments problem, measured through dispute win rates and claim patterns. One team rarely owns both, which is why the split has to be explicit in the reporting, or each team optimizes against half the picture.

Building the split into your process

Start by tagging returns at inspection with wear indicators, separate from the customer's stated reason. Then join that with payments data on disputes and chargebacks per account. The accounts that cluster on wear indicators are your wardrobing population; the ones that cluster on disputed claims are your friendly fraud population. The overlap, customers doing both, is your highest-risk segment and deserves the strictest handling.

The tagging has to be consistent to be useful. Train inspectors on a fixed checklist of wear signals with photo examples, and audit a sample of inspections monthly for drift. Inconsistent tagging poisons the split: if one warehouse flags aggressively and another barely flags, the data reflects the warehouses, not the customers. A short calibration exercise each quarter keeps the signal clean, and clean signal is what makes the account-level clustering trustworthy.

The verdict

Same refund log, two different crimes. Separate them with inspection evidence on one side and dispute data on the other, then apply return friction to wardrobers and claim friction to friendly fraudsters. The distinction is the difference between a deterrent that works and one that just moves the abuse somewhere else.