Cascade Routing / Failover Routing is a payment set-up in which a transaction that fails or is declined at the primary processor or acquirer is automatically resent to backup providers in a fixed order until one approves it or the list runs out. The customer sees one attempt, while two or three authorisation requests may run behind it under one order ID. In the narrow sense, failover reacts to technical errors and cascading to issuer declines; timeouts need a status check first to avoid a double charge. A second route only makes sense for soft declines, and card schemes cap reattempts and charge fees for excessive ones. Smart routing instead picks the first route per transaction.
How does cascade routing work
The merchant or its orchestration layer keeps an ordered list of routes: acquirer A, then acquirer B, then perhaps a local acquirer in the cardholder's country. When route A returns an error, a timeout or a decline, the engine checks the reason against an eligibility table. If the reason qualifies, it sends the same card details, amount and currency to route B under a new transaction reference with that acquirer, while keeping one order ID so the customer can be charged only once. Each hop adds latency, so most set-ups cap the cascade at two or three routes and a total time budget for the checkout.
Cascade routing vs failover routing
The names are used interchangeably but describe different triggers. Failover in the narrow sense reacts to technical failure: the acquirer's API is down or returns a server error, and no issuer ever saw the request. Cascading in the narrow sense reacts to declines: the issuer saw the payment and refused it, and the merchant tries again through another acquirer. Timeouts sit in between, because the first request may have been approved without the merchant hearing back; a status check or reversal has to happen before the next route, or the customer is charged twice.
Which declines can be cascaded
Only soft declines are worth a second route: acquirer-side risk filters, processing errors, issuer unavailable. Hard declines such as a stolen card or closed account fail everywhere. Visa's rules state that when an issuer returns a Category 1 response (the issuer will never approve), the merchant is not permitted to reattempt the transaction. Visa and Mastercard both cap reattempts on other declines and charge acquirers a fee per excessive attempt, which acquirers pass on to merchants. A cascade to a second acquirer is a reattempt on the same card, so it counts.
What cascade routing is worth
Illustrative example: a merchant makes 100,000 card attempts a month at an average €60, and its primary acquirer approves 88%. Of the 12,000 failures, 3,000 are technical errors or eligible soft declines. If the backup acquirer approves 40% of those, the merchant recovers 1,200 payments, or €72,000 a month, and its effective approval rate rises to 89.2%. The cost is a second set of authorisation fees on 3,000 attempts, plus scheme fees if the eligibility table is too loose.
A cascade is static: the order is fixed in advance. Smart routing picks the first route per transaction from BIN, country and live performance, and many merchants combine the two, using smart routing for the first attempt and a cascade for recovery. A circuit breaker that removes a failing acquirer from the list for a few minutes stops every payment from timing out on a dead route before reaching the backup.
For a buyer of a PSP or orchestration platform, cascade logs show how much of the reported approval rate comes from second attempts, and whether retry rules respect scheme limits. A platform that cascades hard declines carries fee exposure that its headline approval figures do not show.