Holiday Return Fraud vs Friendly Fraud at BFCM

The answer, first The December scale, with sources Return fraud: signals and response Friendly fraud: signals and response The December triage table FAQ

The answer, first

Return fraud is a refund obtained by abusing the return process: the customer (or an organised ring) gets money back without returning the item in saleable condition — or without returning it at all. Friendly fraud, also called first-party misuse, is a chargeback on a purchase the customer made and received: instead of returning the item or asking for a refund, the customer tells their bank the charge was unauthorised, not received, or not as described. The distinction is the door the money leaves through. A return-fraud case is decided at the returns desk, with inspection, serial numbers and return history as evidence. A friendly-fraud case is decided at the bank, with delivery proof, order records and descriptor clarity as evidence. Sending a return-fraud response to a bank dispute — or a dispute response to a returns-desk case — loses both, because each decision-maker is looking at different evidence.

The December scale, with sources

Every figure below carries its publisher and year, because the two big return studies measure different things and their numbers must not be blended:

NRF and Happy Returns, 2025 Retail Returns Landscape: retailers expected 17% of holiday sales to be returned, consistent with previous years; 15.8% of total annual sales returned ($849.9 billion); 19.3% of online sales returned; and 9% of all returns fraudulent. Among retailers tracking return-fraud incidents, the reported increases were overstated return quantities (71%), empty boxes or "box of rocks" (65%), and decoy returns such as counterfeit items (64%).
Appriss Retail and Deloitte, 2024 Consumer Returns in the Retail Industry: total returns of $685 billion (13.21% of sales), with $103 billion lost to fraudulent and abusive returns and claims — 15.14% of all returns. The gap with NRF's 9% is definitional, not a contradiction: Appriss counts fraud and abuse together (including wardrobing and bracketing), while NRF's 9% counts fraud alone.
ACI Worldwide, 2025 holiday forecast: friendly fraud projected to rise 25% over the Black Friday–Cyber Monday peak, with the average transaction value associated with friendly fraud expected to reach $291 — up 21% ($52) on the previous year.
Merchant Risk Council, Visa Acceptance Solutions and Verifi, 2026 Global eCommerce Payments and Fraud Report: resolving one first-party misuse dispute costs about $82 on average in staff time and fees, on top of the lost order.

Read together, the picture is a December pincer: roughly one in six holiday sales comes back, about one in eleven returns is fraudulent, and at the same time the bank-dispute route spikes by a quarter with a $291 average ticket. Staffing only the returns desk, or only the dispute queue, leaves the other door open.

Return fraud: signals and response

Return-abuse signals cluster around the item and the return itself: wardrobing (a worn or used item returned as new — the most commonly reported type in the Appriss/Deloitte 2024 survey, reported by 60% of executives), returns from fraudulent or stolen tender such as gift-card fraud (55%), returns of stolen merchandise (48%), empty boxes, decoy or counterfeit swaps, and overstated quantities. The response is operational, not legal: inspect before refunding during peak (the NRF survey found most large retailers refund or exchange before inspection — that convenience is the opening), check serial numbers and item condition against the original shipment, flag accounts with return rates far above the norm, and tighten the specific abuse you actually see rather than the whole policy — a blanket "no returns in December" punishes the 91% of returns that are legitimate and drives those customers to the bank-dispute door instead, which is how return policy creates friendly fraud.

Friendly fraud: signals and response

Dispute-abuse signals cluster around the claim, not the item: "unauthorised" claims from the account holder's own device or address, "not received" claims with a signed delivery record, "not as described" claims where the listing, photos and order record match the item, and repeat disputes from the same customer after a refund was already given. The response is evidential: a recognisable billing descriptor (unrecognised charges are the commonest trigger for a "fraud" dispute on a legitimate purchase), delivery and tracking records, order and account history showing prior use, and pre-dispute alerts that turn the complaint into a refund before it posts as a chargeback — the one move that keeps the case out of the ratio entirely. Where a dispute does post, fight it on the evidence, priced case by case: at roughly $82 to resolve plus the order value, a $20 item is usually cheaper refunded than fought, while a $291 average-ticket case is usually worth the evidence file. The Fight-or-Refund Decision Calculator prices that call per case.

The December triage table

Signal	Which fraud	Response that works
Worn/used item returned as new	Return fraud (wardrobing)	Inspect before refund; condition photos; account return-rate flag
Empty box or counterfeit swap	Return fraud	Weight/serial check on receipt; hold refund pending inspection; report the pattern
Return with stolen tender or stolen goods	Return fraud	Tender verification; refuse and document; law-enforcement route for organised rings
"Unauthorised" claim, own device/address	Friendly fraud	Device and account evidence to the bank; descriptor fix to prevent repeats
"Not received" with delivery proof	Friendly fraud	Tracking and delivery record as representment evidence
Dispute after a refund was already given	Friendly fraud (double recovery)	Refund record as evidence; alert the issuer to the double recovery
Customer confused by an unclear descriptor	Neither — preventable	Fix the descriptor; answer the customer before they call the bank

One December rule covers both doors: never let a frustrated legitimate customer become a dispute statistic. A slow or refused return is the most reliable friendly-fraud generator there is — the customer who cannot get a refund through your door will get one through their bank's, and the bank's version costs you the sale, the fee, and a ratio point. Staff the returns desk to be fast for honest returns and strict for flagged ones, and staff the dispute queue with evidence, not apologies. That split — fast for the honest, evidence for the abusive — is the whole December triage.

FAQ

Is wardrobing return fraud or friendly fraud?

Return fraud. Wardrobing — wearing or using an item and then returning it — abuses the return process to get a refund. Friendly fraud is a bank dispute on a purchase the customer keeps. Different doors, different evidence, different responses.

How much of holiday selling actually comes back?

Retailers expected 17% of holiday sales to be returned, according to NRF and Happy Returns' 2025 Retail Returns Landscape — consistent with previous years. For online sales the expected return rate was 19.3%, and for all retail sales 15.8%.

What share of returns is fraudulent?

9% of all returns, according to NRF and Happy Returns (2025). Appriss Retail and Deloitte (2024) put fraudulent and abusive returns together at 15.14% of returns, worth $103 billion — the higher figure includes policy abuse such as wardrobing and bracketing, not fraud alone.

Why does friendly fraud spike at Black Friday?

Volume, gift purchases and unclear descriptors collide: more first-time buyers, more gifts the buyer never sees, and more statements full of unfamiliar merchant names. ACI Worldwide forecast a 25% rise over the Black Friday–Cyber Monday peak in 2025, with an average friendly-fraud transaction value of $291.

Should I fight every friendly-fraud dispute in December?

No. Resolving one first-party misuse dispute costs about $82 on average (Merchant Risk Council / Visa / Verifi, 2026), on top of the order value. Price each case: fight the high-value, well-evidenced ones, refund the low-value ones before they post where an alert allows it.