1. Detect
Identify drops in authorization rate, errors, latency, timeouts, webhook failures, stopped payouts or settlement outside expectations.
PaymentsNova Operations Tool
Payment Incident & Failover Runbook
Redundancy only works if the team knows when to activate it, who decides and how to reconcile what happened afterward.
Signals that should trigger review
| Area | Signal |
|---|---|
| Authorization | Material drop versus the market/method baseline |
| Availability | Timeouts, errors or unavailable provider |
| Payouts | Stopped queue, abnormal rejects or undelivered funds |
| Settlement | Delay, unexpected amount or interrupted funds route |
| Reconciliation | Growing differences, inconsistent statuses or missing data |
Runbook sequence
2. Classify
Determine scope: provider, method, market, currency, cashier, acquiring, bank rail, wallet/network or internal dependency.
3. Contain
Avoid blind retries, duplicates or uncontrolled routing changes. Preserve IDs, timestamps, logs and evidence.
4. Decide failover
Activate an alternate route only if it is approved, tested and operationally available. Define who authorizes the change.
5. Communicate
Align operations, payments, support, risk/compliance and the affected provider. Maintain one incident timeline.
6. Protect funds flow
Review pending transactions, payouts, balances, settlement and treasury exposure before and after the change.
7. Reconcile
Compare the internal source of truth with provider reports to identify pending items, duplicates, fees, refunds and differences.
8. Recover
Restore the original route only when cause and stability are sufficiently verified.
9. Postmortem
Document cause, impact, recovery time, decisions, gaps and preventive actions with an owner.
Before activating failover
Confirm that the alternate route remains available for the applicable merchant, market and method. Ongoing provider approval, underwriting, compliance, limits and conditions can change; an existing integration does not guarantee permanent operational availability.
PaymentsNova
Turn the problem into a clear architecture.
Start with a stack review or describe the case directly.