Retry deliberately
Use documented retry timing and bounded backoff. Do not create an endless retry loop or ignore a limit response.
Control concurrency centrally
Several workers sharing one account can exceed a combined limit even if each worker appears conservative. Coordinate request budgets across the application.
Use bounded backoff
Follow documented retry timing when provided. Otherwise use a deliberately bounded strategy with jitter appropriate to the service contract. Set a maximum attempt count and surface exhausted work for review.
Measure useful throughput
Track completed records, latency, limit responses, and inconclusive results together. Sending more requests is not an improvement if it increases blocking or duplicate charges.
Before you enable this workflow
| Confirm | Why it matters |
|---|---|
| Production contract | Use documented fields, limits, and supported operations. |
| Failure policy | Decide what the application does when a check cannot complete. |
| Data boundaries | Keep credentials and customer data out of unnecessary logs. |