PSP cascade is the routing layer that picks the payment provider for each transaction and defines what happens when an attempt fails. It looks like infrastructure; in practice it is one of the most direct levers on revenue.
The router decides based on method, amount, currency, player history, cost and provider availability. When a decline is retryable (timeout, momentary unavailability, provider limit), the transaction moves to the next provider in the queue. When it is hard (invalid data, suspected fraud, issuer block), retrying only adds cost and risk.
Common mistakes:
- Retrying without handling the decline code.
- Optimising for price instead of cost per approved transaction.
- A static order that ignores availability variation through the day.
- No observability per provider and per method.
An overly aggressive cascade also creates side effects in acquirer anti-fraud, because testing the same instrument in sequence is exactly the pattern those models look for.
Updated
All terms