What return reasons predict future return fraud?

September 26, 2026

Every return your store processes carries a reason code, and most brands treat those codes as accounting trivia. In audits, we find they are one of the strongest early signals of future return fraud. Customers who will go on to abuse the return process select systematically different reasons than customers who will not, and the difference shows up months before the return rate crosses any threshold.

The reason is straightforward: reason codes describe the story a customer is telling about the return, and fraud requires a cover story. A customer who intends to keep a dress for a wedding and send it back afterwards cannot select "wore it once" from the dropdown, so they pick something plausible: "did not like," "wrong size," "changed mind." Those selections are not random. They cluster.

The reasons that predict trouble

Across audited catalogs, three reason families over-index among customers who later show clear fraud signatures. First, vague preference reasons: "did not like," "changed mind," "not as expected." These are the easiest cover stories, and serial returners and wardrobers select them far more often than ordinary customers. A customer whose returns are 80 percent vague preference reasons is behaving very differently from one whose returns skew toward specific fit complaints.

Second, condition-sensitive reasons on the wrong categories: "damaged" or "defective" selected on simple basics or repeat purchases. Occasional damage claims are normal. A pattern of damage claims from one customer, especially on items that are hard to damage in transit, predicts claims fraud specifically.

Third, size reasons across full size runs. "Too small" or "too big" is a legitimate reason. Selecting every size reason across a full run of sizes on the same SKU is bracketing, and the reason codes document it in plain text.

The reasons that predict innocence

The flip side matters just as much. Specific, verifiable reasons predict honest customers: "too short," "zipper broken," "color different from photos," "runs small, sizing inconsistent with last season." These reasons give the brand actionable feedback, they vary across orders, and they rarely repeat identically month after month. A customer who writes a real reason is usually a customer you want to keep.

This is why blunt policy responses to return reasons backfire. Penalizing all "changed mind" returns punishes honest customers with the same tool that barely inconveniences abusers, who simply switch reasons. The goal is not to punish the reason. The goal is to read the pattern across reasons.

How to weight reasons in a risk score

Start by computing, per customer, the share of returns in each reason family over a trailing twelve months. Then compare each customer to the brand-wide baseline: what share of all returns use vague preference reasons? Customers running far above the baseline on vague reasons, combined with a high return rate, are the ones to review.

Weight matters more than counting. One "changed mind" return means nothing. Twelve in a row, all on occasionwear, all returned within a week, means a lot. The score should combine reason skew, return rate, category skew, and return timing. Any single signal is weak. The four together are strong.

Two practical moves this week

First, add reason skew to whatever customer review process you already have. If a customer crosses your review threshold, look at their reason distribution before deciding what to do. A threshold-crossing customer with specific, varied reasons is probably a loyal buyer with fit problems. One with twelve identical vague reasons is a different conversation.

Second, clean up your reason list. Long lists with overlapping options produce noise: customers pick the first plausible option, and analysis drowns. Fewer, sharper reasons produce cleaner signal. Every reason on the list should be something a customer might honestly mean, and something you can act on.