When acquirer A declines, send the same transaction to acquirer B in real time. The technical, regulatory and approval-rate detail you need to deploy it correctly.
Cascading is the practice of re-attempting a declined transaction on a different acquirer — same card, same amount, different network path — to recover an approval that the first acquirer's risk model rejected. Done correctly, it lifts authorisation rates 5–12 percentage points. Done incorrectly, it triggers issuer fraud rules and gets cards locked.
ApexPay FZ-LLC is a payments consultancy — we introduce merchants to licensed acquirers, gateways and alert networks, and we do not process payments or hold funds.
Decline codes split into two groups: issuer-side ('insufficient funds', 'do not honor', 'lost/stolen') and acquirer-side ('exceeds withdrawal limit on this MID', 'merchant velocity exceeded', 'risk decline at acquirer level'). Cascading recovers the acquirer-side declines almost completely. It does not recover issuer-side declines — and attempting to is what gets you in trouble.
Visa and Mastercard treat repeated authorisation attempts on the same card within a short window as 'card testing'. Cascading must be limited to one re-attempt on a different acquirer, with a different MID DBA where possible, and never cascade after issuer-side hard declines. Excessive re-attempts trigger network fraud rules and can blacklist your processing entity.
Deterministic rules cover the obvious cases. ML cascading uses historical approval data per BIN, MID, time-of-day, ticket size and decline code to predict which acquirer to attempt next — and whether to attempt at all. Our partner network runs both: deterministic rules as a floor, ML routing as a lift.
Yes, within scheme rules. The constraint is on excessive re-attempts and card-testing patterns. One cascade on a soft decline is well inside scheme tolerance.
No — recovered approvals are real customer transactions. CB ratio is calculated against approved transactions, and recovered approvals don't have a higher CB profile than first-pass approvals when cascading is rule-disciplined.
Yes. Cascading on the same MID achieves nothing — it's the same risk model. Real cascading requires at least 2 MIDs across different acquirers.
5–12 percentage points typical for high-risk merchants moving from single MID to 3-acquirer cascading. Largest gains on cross-border traffic.
Multi-MID architecture, orchestration, cascading, gateways and offshore vs domestic strategy.
Bring your current MIDs. We'll wire deterministic + ML cascading and benchmark approval lift in 14 days.
Get a free, no-obligation audit of your payment stack. We'll show you exactly where revenue is leaking and how to fix it.
