Short answer
Return fraud is using a store's return process to get money or goods the customer is not entitled to: wearing an item and sending it back, returning a different or empty package, or claiming a refund for an order that is not theirs. Its milder cousin, return abuse, bends the policy without breaking it, such as returning most of what one buys.Online stores see a recognizable set of patterns. Wardrobing is buying an item for one occasion, a party dress or a camera for a trip, and returning it afterwards as unused. Item swapping sends back a lower-value or broken unit in place of the one that shipped. Empty-box returns rely on the warehouse not checking contents quickly. And account or order takeover uses someone else's order number to request a refund to a new card or address.
| Pattern | What arrives back | Where it is caught |
|---|---|---|
| Wardrobing | a worn item with tags reattached | inspection at the warehouse |
| Item swapping | a different or damaged unit | inspection against the order |
| Empty or light box | nothing, or the wrong weight | receipt and carrier weight |
| Order takeover | often nothing at all | identity check before acting |
Support sits at the start of this chain. The person asking for a return or refund in chat is the first point where a store can confirm who they are and what they bought. The popular response to fraud runs the other way: tightening the policy for everyone, with shorter windows and stricter conditions, which punishes the honest majority to slow down a few. A better line is to make the checks precise. Confirm that the person owns the order before discussing it, keep refunds behind inspection or approval, and watch for the same customer returning the same kind of item again and again.
Aide only discusses or changes an order once the shopper proves it is theirs, by signing in to their store account or by giving the order number with the matching email. Returns it starts land in Shopify as requests the store approves, refunds wait for a teammate by default, and the customer's earlier conversations sit beside the current one, so a pattern is visible to whoever reviews it. Every order action is recorded with who did it. More on guardrails for AI customer service.