Loss Prevention Glossary

A chargeback is a forced payment reversal initiated by the cardholder’s issuer, in which the merchant loses the sale, the goods, and a dispute fee — and the dispute starts a representment window measured in days, not weeks.

In practice, "chargeback" covers two very different disputes. The friendly-fraud variety (a buyer who keeps the product and disputes the charge) and the genuine fraud variety (an unauthorized card use) both arrive through the same reason-code rails, and the representment window is the same. For beauty merchants, the threat is that small-ticket replenishment orders let one customer cycle through four or five chargebacks a quarter without ever flagging on the issuer side — the textbook profile a representment program is designed to defend.

Friendly fraud is a chargeback filed by a cardholder who actually received the goods — disputed to keep both the product and the refund, then monetized through personal use or a marketplace resale.

The term is a misnomer: the buyer is not "friendly," but the issuer-side workflow is. The cardholder contacts the issuer, claims non-receipt or unauthorized use, and the network treats the dispute as if it were genuine fraud. The merchant has a narrow 7-to-10 day Visa/Mastercard window to file a representment — and Amex an even more demanding SAFE-driven process — which is why a tamper-evident verification record captured at packing is the single most decisive artifact in the package.

First-party fraud is a category of chargeback fraud in which the legitimate cardholder is the bad actor — the dispute is real only in the sense the network treats it, never in the sense of the underlying transaction.

First-party fraud is the umbrella term under which friendly fraud sits, and the reason it shows up here: most beauty chargebacks resolve through first-party, not third-party, channels. A distinction drawn at the network level between friendly fraud (cardholder-scored, mostly small-ticket) and chargeback fraud (organized, often marketplace-resold) is what shapes each respond-with-evidence step in the representment playbook.

A BIN attack is a fraud pattern that scripts small-ticket authorizations across a bank identification number range to discover valid card numbers — a card-test that almost always precedes a friendly-fraud wave downstream.

The card-test amounts are intentionally low (under $5), the issuer-permitted AVS behavior makes many pass, and the resulting "approved" cards are the inventory a friendly-fraud wave is built from two to six weeks later. For beauty merchants, the early signal in the issuer-side chargeback ledger is the giveaway: a BIN-attack cluster at the same issuing bank almost always precedes a chargeback uptick on high-AOV prestige launches.

Wardrobing is return fraud in which the buyer uses an item and then returns it as unworn, untested, or "didn’t fit" — and the category shows up most in apparel, beauty, and prestige fragrance launches.

Appriss Insights pegs return fraud at $101B in 2024 and 13.7% of total merchandise returns as fraudulent — the wardrobing slice is the largest sub-bucket in beauty because every prestige fragrance and color cosmetics launch is essentially a try-at-home event. Wardrobing crosses into retail shrinkage, which is the line item most beauty LP teams carry at 1.6% of sales.

A BOPIS swap is a return-fraud pattern unique to buy-online-pick-up-in-store, in which a buyer exchanges a delivered-but-different SKU (typically a counterfeit) for the genuine unit at the return desk.

BOPIS swaps show up alongside $137B-projected return fraud losses because the swap window — the moment the customer hands the associate a unit and walks out with a replacement — is the gap most retailer workforces do not staff for. The counter-side verification record that closes the loop is the same artifact the representment deck hangs on two chargeback cycles later.

Organized retail crime (ORC) is fraud conducted by coordinated groups who divert, counterfeit, or steal inventory in volume — usually for marketplace resale — and increasingly target prestige fragrance and color cosmetics.

The 2026 NRF survey shows 73% of LP leaders reporting ORC activity rising in the past year; the per-unit marketplace resale on prestige fragrance is high enough that ORC crews treat the category as a default target. ORC is the threat model that drives the case for a serialized verification record at pack-out, since the dispute chain that follows a stolen-batch consumer is what a representment deck has to defend.

Retail shrinkage is the gap between recorded and counted inventory, and in 2024 amounts to roughly $112.1B in US retail — a 19% jump in two years that hits beauty hardest per dollar of inventory on hand.

The NRF pegs shrink at 1.6% of total retail sales in 2023 — and for a $500M prestige-beauty business, that line item is roughly $8M a year before chargebacks, vendor chargebacks, and counterfeit returns are even tallied. Beauty over-indexes on shrink per dollar of inventory on hand, which is the asymmetric line that motivates the LP program in the first place.

Chargeback win rate is the share of disputed orders the merchant recovers in representment — the single metric that determines whether a beauty chargeback-defense program is paying for itself or not.

Without a structured defense workflow, beauty merchants recover only ~32% of chargebacks (Retail Dive, 2024); the remaining 68% is overwhelmingly lost to the friendly-fraud, small-ticket tail. A representment deck anchored in a serialized verification record typically lifts win rate past 60% — the recovery dollar math is what earns the program C-suite funding.

Return fraud rate is the share of returns that qualify as fraudulent (wardrobing, BOPIS swaps, empty-box returns, BIN-test-influenced volume) — and the figure now runs above 13% of total merchandise returns per year.

Appriss puts US retail return fraud losses at $101B in 2024 and projects the figure to exceed $137B by 2028. Appriss’s framing — "more than 1 in 8 returns is fraud-tagged" — is the right metric to anchor a C-suite read on the program, because the dollar figure translates back to gross-margin recovery and not just refunds avoided.

Product authentication is the workflow that confirms a returned SKU is the genuine unit the brand produced — and is the missing artifact in most failed representment decks at the close of the dispute window.

Authentication in beauty rests on a paired home-and-counter image capture, an on-device metadata fingerprint, and a brand-rule lookup against SKU-specific rules. The result is a tamper-evident verification record attached to the SKU that the issuer can independently re-check, which is the asymmetry the rest of the chargeback-defense stack is built to exploit.

A metadata fingerprint is a structural hash computed from a photo’s packaging geometry, label-print vector signature, and ink density profile — computed on-device so the raw image never leaves the device.

The fingerprint is what gets uploaded for the similarity score, never the original photo — a privacy boundary that keeps the audit record compliant. The same fingerprint powers the SKU-level ruleset lookup: brand-registered distribution geography, packaging-batch expectations, and the known counterfeit signatures for the line.

Representment is the formal rebuttal a merchant files against a chargeback — a structured evidence package that rebuts a specific reason code with proof the order was legitimate, delivered, and not counterfeit.

Visa and Mastercard give merchants a narrow 7-to-10 day window; Amex offers more generous calendar terms but consultative requirements (SAFE reporting, fuller customer-communications trail) and a higher evidence bar. The metric that justifies the program is the win rate, and the recovery dollars follow.

A serialized verification record is a per-SKU audit artifact — serial number, scan timestamp, geographic origin, fingerprint hash, verdict — that gives the issuer one document to re-check at the close of the dispute window.

The record is the missing artifact in most representment decks; it lets the issuer independently re-verify the disputed unit after the dispute window closes, which is what a generic rebuttal cannot. Amex overweights this record relative to store-floor photos, while Visa weights it inside the larger Compelling Evidence 3.0 framework.

Compelling Evidence 3.0 (CE 3.0 / CBE 3.0) is Visa’s evidence framework for representment — four categories (customer profile, transaction history, shipping/fulfillment, device/IP) that the issuer scores against the disputed reason code.

CBE 3.0 rewards merchants who submit evidence across all four categories rather than a single one. In the beauty lane, the dominant three reason codes are 10.4 (CNP fraud), 13.1 (non-receipt), and 4837 (no cardholder authorization) — and each one demands a different evidence shape, with the serialized verification record filling the shipping/fulfillment slot.

SAFE (Services to Acquirers and Fraud Early-warning) reporting is Amex’s deadline-driven fraud-intel workflow — and missing the SAFE deadline forfeits the right to representment even when the evidence is overwhelming.

SAFE deadlines are the first thing that closes an Amex dispute window; once missed, even the best evidence package cannot recover the dispute. The deadlines sit ahead of the dispute itself, which is why LP, finance, and customer-care typically run them as a single shared workstream and not as separate filings.

A chargeback reason code is a network-assigned tag that names the dispute category — and the reason code is what tells the merchant which evidence shape the representment deck has to match to clear the dispute.

In the beauty lane the dominant three are Visa 10.4 (CNP fraud), 13.1 (non-receipt), and 4837 (no cardholder authorization); Amex’s taxonomy uses different categories ("goods not received", "significantly not as described", "fraudulent transaction") that map onto the friendly-fraud fingerprints the LP team has to defend.

Refund policy tightening is the post-chargeback practice of revising the refund policy to keep the recovered customers from reopening the dispute — and is what closes a representment program into a durable loop.

A representment program that recovers 60% of friendly-fraud disputes is materially less valuable without a refund-policy revision that keeps the recovered customers from re-disputing. The two programs ship together or not at all — and the C-suite framing is the share of friendly-fraud customers re-rated in the post-dispute workflow, not the share of new disputes opened.