The 36% Friendly-Fraud Number Is Measuring the Industry, Not the Buyer

The headline 2026 statistic — first-party fraud rose from 15% to 36% of all reported fraud — reflects better detection by acquirers, not worse behaviour by cardholders.

The most-shared chargeback statistic of 2026 is a single number. First-party fraud has risen from 15% of all reported fraud in 2023 to 36% in 2024, and the figure now anchors the talking points of every chargeback-tooling vendor on LinkedIn. The implied conclusion is that cardholders have become dramatically worse over two years, and that merchants need new detection tooling to keep up. That reading is wrong about which side of the equation moved.

The denominator is "reported fraud cases" and the numerator is "cases acquirers or processors labelled as first-party fraud after analytical review." Between 2023 and 2026, the analytical machinery on the acquirer side improved sharply. Visa rolled out Compelling Evidence 3.0 across the network in 2023. Mastercard followed with First-Party Trust the same year, and Verifi and Ethoca's pre-dispute deflection tools layered behaviour-detection on top of their alert flows. The major US acquirers retrained their internal classifiers on labelled data from the new frameworks. The share of cases that the industry's machinery can now classify as first-party fraud, rather than throwing into the generic "fraud" bucket, has risen significantly. The cardholder behaviour underneath did not triple.

The underlying first-party share has been roughly steady. Industry estimates and our own observations across thousands of card-not-present consumer disputes put first-party misuse, by which we mean disputes where the cardholder actually authorised the transaction and is now contesting it, at between 50% and 70% of CNP consumer disputes for the better part of a decade. That share has fluctuated within a band but has not undergone the step-change the headline number implies. What changed was the share of that band the industry's analytics can identify as such after the fact.

The operating consequence matters because the merchant's spending decisions follow the headline number. A merchant looking at "36% and rising" allocates budget to detection tools that promise to catch first-party fraud before it converts to a chargeback. The detection tools have a role, particularly on the recognition-failure cases that descriptor configuration can solve cheaply. But the bulk of the recovery is downstream, at representment, on the cases that arrive. A representment process that handles every dispute as if it were already labelled, with issuer-aware framing and evidence packets matched to reason code, wins recovery the detection tools cannot reach. Allocating budget around the 36% panic deflects spend from the work that actually compounds.

The work that compounds is the per-case quality of the letter that lands at the issuer, refreshed as outcome data accumulates and as issuer review patterns shift. The 36% headline measures the industry's labelling apparatus. The merchant's return is measured by the win rate on the cases that arrive, and the rule library representments.com maintains is built for that work, refreshed against observed outcomes so the next letter learns from the last.