Skip to content

CONCEPT Cited by 1 source

Fraudulent dispute prediction

Definition

Fraudulent dispute prediction is the architectural primitive of predicting whether a payment will result in a fraudulent dispute — a chargeback / dispute filed with adversarial intent (e.g. friendly fraud where the cardholder authorised the transaction but disputes it claiming non-receipt) rather than a legitimate complaint.

Distinct from early-fraud-warning prediction: EFW targets the card-network's pre-chargeback signal, which can fire on legitimate disputes too. Fraudulent- dispute prediction targets the adversarial subset specifically.

Stripe Radar's prediction

Per the 2026-05-27 Stripe Radar disclosure:

"Stripe can also predict whether a payment is likely to result in a fraudulent dispute. You can use this signal to issue refunds, gather evidence, or adjust your dispute strategy."

The signal is one of the two named members of multiprocessor fraud signal export (alongside EFW prediction) — usable on non-Stripe-processed transactions as well as Stripe-processed ones.

Action strategies

The post enumerates three merchant responses:

  1. Issue refunds — same response as EFW; most appropriate when contesting cost exceeds expected refund cost.
  2. Gather evidence — compile shipping records, usage logs, delivery confirmations, etc.
  3. Adjust dispute strategy — feed the prediction into Smart Disputes for evidence-tailoring; deploy higher-tier evidence on high-fraudulent-dispute-probability cases.

Composition with Smart Disputes

The natural composition with systems/stripe-smart-disputes:

  • Radar predicts fraudulent-dispute likelihood at transaction time.
  • If dispute is later filed, Smart Disputes uses the prediction to calibrate evidence-gathering depth — high-likelihood fraudulent-dispute cases get full evidence-library packets, low-likelihood cases get minimal evidence.

This is a lead-vs-lag composition: Radar's lead-time prediction at transaction time informs Smart Disputes' lag response after the dispute is filed.

Caveats

  • Distinction between fraudulent-dispute and legitimate- dispute prediction quality not disclosed.
  • No explicit signal-mix description — what does Radar use to predict fraudulent intent at transaction time?

Seen in

Last updated · 542 distilled / 1,571 read