Skip to content

CONCEPT Cited by 1 source

Fraudulent merchant detection

Definition

Fraudulent merchant detection is the architectural primitive of identifying whether a business account (existing or newly-onboarded) on a payments platform poses a fraud risk, based on Stripe-network-wide pattern analysis.

Disclosed by systems/stripe-radar-for-platforms at the 2026-05-27 Stripe Sessions roundup. Per the post:

"The fraudulent merchant signal identifies whether a new or existing account poses a fraud risk, based on analyzing patterns across the Stripe network, including bank account information, business details, transaction activity, and disputes."

Disclosed signal mix

Four categories named:

  1. Bank account information — recurring bank accounts, suspicious bank-account-sharing patterns across known-bad merchants.
  2. Business details — registered business info, address patterns, ownership graphs.
  3. Transaction activity — volume / pattern anomalies, known- bad-counterparty patterns.
  4. Disputes — dispute rate, dispute pattern, dispute reason-code distribution.

The architectural insight is that all four signals are Stripe-network-wide: a merchant's bank account that has appeared in known-fraud cases at other Stripe merchants is a red flag for this merchant. This is network-effect fraud detection applied to KYB rather than to transaction-tier fraud.

Platform action surface

The post enumerates platform actions:

  • Raise a review — human investigation.
  • Pause payouts — hold proceeds pending review.
  • Pause payments — stop accepting new transactions.
  • Reject the account — terminate the merchant.
  • Set reserves — hold a fraction of proceeds against predicted future losses.
  • Request identity verification — escalate verification depth.

The action surface is graduated, letting platforms calibrate response severity to their risk tolerance.

Distinction from sibling concepts

The three signals together form Radar for Platforms' KYB triad: website content + network-wide transaction pattern + forward-looking financial-loss prediction. Platforms can use any subset; high-confidence rejections come from cross-signal agreement.

Distinction from current-state-only KYB

Traditional KYB checks (business registration verification, beneficial-ownership disclosure, basic underwriting) check identity-validity at onboarding. Fraudulent-merchant detection layers on pattern-matching against the network's known-fraud merchant population — a merchant whose business details and bank account are clean but whose patterns match a fraud cluster Stripe has previously flagged elsewhere.

Caveats

  • No precision/recall envelope disclosed.
  • No latency-of-evaluation disclosed — at onboarding only, or continuously?
  • Cross-merchant-network privacy posture not described — the signal explicitly leverages patterns from other merchants' data; data-handling specifics undisclosed.

Seen in

Last updated · 542 distilled / 1,571 read