Bracketing vs wardrobing: how to tell the two apart in your returns data

September 28, 2026

Most returns dashboards lump bracketing and wardrobing into the same number: the return rate. That is a costly blur. Bracketing, buying two sizes to keep one, is a fit behavior from customers who often convert well. Wardrobing, buying to wear and return, is fraud that destroys margin and inventory. Treat them the same and you either punish good customers or let abusers operate freely. The two behaviors leave different fingerprints in your data, and reading them correctly is the first step to handling each one properly.

What bracketing looks like in the data

Bracketers are predictable. They order multiple sizes of the same item in one order, keep one, and return the rest quickly, usually within a week of delivery. The returned units come back in original condition with tags attached, because they were tried on, not worn. The customer has a healthy purchase history: they keep things, their kept orders have normal AOV, and their returns cluster around fit-risky categories like denim, footwear, and occasionwear.

The key signal is speed and condition. Bracket returns arrive fast and resellable. The customer is not trying to extract value from the product; they are trying to extract certainty about fit. Their behavior responds to fit interventions: better size charts, fit quizzes, and consistent sizing reduce their bracketing rate, which is the proof that the behavior was fit-driven in the first place.

What wardrobing looks like in the data

Wardrobers are also predictable, but in the opposite direction. They order one unit, keep it for most of the return window, and return it late, often in the final days before the deadline. The returned unit shows wear: stretched fabric, deodorant marks, faint perfume, scuffed soles. Tags may be reattached or missing. The return reason is usually generic, "didn't fit" or "changed mind," chosen to end the conversation rather than describe the problem.

The purchase history tells the rest of the story. Wardrobers cluster around event-driven categories, occasionwear, premium outerwear, statement pieces, and their orders spike before holidays and wedding season. Their kept AOV is low relative to their order volume, because most of what they buy comes back. And unlike bracketers, their behavior does not respond to fit interventions: better size charts do not reduce wardrobing, because fit was never the point.

The three signals that separate them

First, return timing. Bracket returns cluster in the first week; wardrobing returns cluster near the deadline. Plot return latency for your top returners and the two populations separate cleanly. A customer whose returns consistently land on day 28 of a 30-day window is not bracketing.

Second, unit condition on inspection. Bracketed returns resell as new. Wardrobed returns need cleaning, repair, or liquidation. If your warehouse is not grading return condition and feeding it back into the customer record, you are blind to the single most reliable wardrobing signal you have.

Third, category concentration. Bracketing concentrates in fit-risky basics where size uncertainty is genuine. Wardrobing concentrates in high-visibility, event-driven pieces where wearing once has social value. A customer who brackets denim and wardrobes evening gowns exists in every dataset, and the category split is what lets you treat the two behaviors differently for the same person.

Why the distinction changes the response

Bracketing is a conversion and fit problem. The right responses are fit tools, keep-your-size incentives, and bracket-aware return fees that nudge without punishing. Brands that treat bracketers as fraudsters lose customers who would have been profitable for years.

Wardrobing is a fraud problem. The right responses are decision layers: risk scoring at checkout, inspection feedback loops, and policy enforcement that makes the behavior unprofitable. Brands that treat wardrobers as fit-confused customers subsidize abuse with free rentals.

The audit we run starts here, with the data split, because every policy downstream depends on it. Window length, fee structure, auto-approval rules, all of them need to know which returns are fit behavior and which are fraud. Get the split wrong and every lever pulls the wrong way.