Skip to main content

Refund Fraud

TL;DR
  • Refund fraud is exploiting return or refund policies to get money or goods without legitimate cause
  • Schemes to know: wardrobing (using an item, then returning it as unused), empty box returns, false not-received claims, double-dipping (refund + chargeback)
  • "Refund-as-a-service" groups market on Reddit, TikTok and Telegram (CNBC, March 2024)
  • Detect via: refund patterns outside comparable customer history, weight mismatches and serial number changes
  • Related to friendly fraud and chargeback management

The same customer keeps getting refunds on claims you can't check? Compare their returns with the signals below before you approve the next one. Each claim on its own looks ordinary.

Refund fraud is getting money or goods back on a claim that isn't true. Sometimes it's a used item returned as new, sometimes it's a package that did arrive, and after the fact, both are hard to disprove.

One refund looks like a bad day. The same name every month is worth a closer look.

Common Schemes​

Each of these needs a lie or a false entitlement. A lawful return and a real delivery or quality complaint aren't fraud.

Return Fraud Variants​

TypeDescription
WardrobingUse item, then return it as unused with tags still attached
Receipt fraudUse found/stolen receipts to return shoplifted goods
Price switchingSwap tags, return at higher price
Empty boxReturn empty box or box with wrong item
Cross-retailerBuy at Store A, return at Store B while claiming you bought it there

Refund Fraud Variants​

TypeDescription
Double dippingGet refund from merchant AND chargeback
Did Not Receive (DNR)Falsely claim a package wasn't delivered (see 13.1 reason code)
Item Not As Described (INAD)Exaggerate issues for refund (see 13.3 reason code)
Partial returnReturn part of a multi-item order while claiming a full refund
Refund-as-a-serviceCriminal services that process fraudulent refunds

"Refund-as-a-Service" Economy​

Refund services for hire

CNBC reported in March 2024 that these groups market on Reddit, TikTok and Telegram. A shopper buys online and sends a group the order details. The group poses as the customer to request the refund, and the customer pays it a cut, usually 15% to 30% of the refund by CNBC's account. In one case, court documents describe carrier employees recruited to alter a package's tracking history.

Detection Signals​

Check these against comparable products, case outcomes and what a wrong call costs you.

Customer-Level​

SignalPriority to Check
Refund rate outside the customer's comparable product and purchase historyInvestigate; set alerts from your own returns data, not a universal percentage
Multiple "not received" claimsFirst
Refund requested immediately after deliveryNext
Returns always at max allowed timeframeNext
Returns of high-theft items (electronics, designer)Next

Transaction-Level​

SignalPriority to Check
Weight difference (return vs. original)First
Serial number mismatchFirst
Item condition inconsistent with claimNext
Photo evidence doesn't match claimFirst

None of these proves anything alone. Set your own trigger from your products and outcomes. Two on one customer is one place to start reviewing before you refund.

Prevention Strategies​

Policy Design​

  1. Proof of purchase: Ask for a receipt, an order number or other proof the law where you sell accepts. UK guidance, for one, accepts a bank statement or packaging (GOV.UK)
  2. Restocking fees: A fee cuts the wardrobing incentive
  3. Time limits: Shorter return windows, within any legal minimum, give the pros less room
  4. Condition requirements: Write down what "used" means before they argue it
  5. Refund method: The money goes back the way it came in. And if a chargeback is already open and you've decided to give the money back, don't refund separately. Accept the chargeback through your processor

Ship the restocking fee first, where local consumer law allows one. It goes at the incentive instead of the effort, and effort is exactly what the refund-as-a-service operators sell.

Taking an app deposit or handling an agreed partial refund? Check the original payment and any notice.

Operational Controls​

  1. Verify delivery: Photo, signature, or GPS at the door
  2. Weight checking: Weigh what comes back against what you shipped
  3. Serial number tracking: Record it going out, match it coming back
  4. Abuse tracking: Flag the names that keep showing up
  5. Refund velocity limits: Cap discretionary refunds per customer per period. What the law or your terms owe the customer still stands

Weight checking and serial number tracking each give you a number to put in front of the customer, for the price of a scale and a spreadsheet. Pilot them where your products and records support it, and count labor and matching errors, before you buy anything.

Technology​

  1. Return probability scoring: A machine-learning score for how likely a return is
  2. Customer abuse scoring: Score the lifetime behavior, not one order
  3. Network analysis: Link related accounts (via device fingerprinting)

Start with the policy. Tools won't save a return desk that pays anyone who asks.

Next Steps​

Detecting refund abuse?

  1. Check customer-level signals: High refund rates
  2. Verify transaction-level signals: Weight, serial mismatches
  3. Set up velocity limits: Caps on discretionary refunds

Preventing refund fraud?

  1. Review policy design: Tighten return policies
  2. Implement operational controls: Weight checking, serial tracking
  3. Track abuse history: Flag serial abusers

Fighting refund-related chargebacks?

  1. Review compelling evidence: What wins
  2. Check refund strategy: When to refund vs fight
  3. Set up alerts: Pre-dispute resolution

Something wrong on this page? Email matt at paymentsandrisk.com and say what needs fixing.