Fraud & Abuse Detection System

We are requesting a rule engine that scores incoming orders and flags the suspicious ones.

Possible signals to detect:

  • Multiple accounts/emails sharing an IP, device, address, phone, or payout handle

  • Disposable email domains, VPN/proxy/datacenter IPs

  • Same ISBN/SKU/IMEI submitted repeatedly, or resubmitted after a prior rejection

  • Order total or quantity above a store-set threshold

  • Orders composed entirely of top-of-price-list items (cherry-picking)

  • Velocity: N orders from one account or IP within X hours

  • Brand-new account submitting an unusually large first order

  • Same tracking number reused across orders

  • Customers with prior chargebacks, disputes, or high rejection rates

Actions

System can flag with a risk badge and reason, hold payout for manual approval, email the admin, or auto-reject.

Configurability

We can enable/disable each rule, set threshold and action, adjust overall sensitivity, and maintain allowlists for known high-volume sellers plus blocklists for emails, IPs, addresses, and payout handles.

This list is somewhat a comprehensive version but even a minimal list like IP/email linkage, repeat-item detection, and an order-value threshold with an email alert would cover most of what we are currently absorbing as a cost of doing business.

Please authenticate to join the conversation.

Upvoters
Status

Under Review

Board

Suggest a Feature

Date

12 days ago

Subscribe to request

Get notified by email when there are changes.