Production integrations fail in predictable ways. This page explains limits, error codes, and how to recover safely without creating noisy retry storms.

Rate limits and quotas

Minute-level protection: 10 requests/minute
Professional plan: 30 calls/day
Enterprise plan: 500 calls/day

Status code reference

Error matrix

Status	Likely cause	How to fix	Retry?
400	Malformed request body or invalid endpoint usage context	Validate payload assembly and endpoint path	no
401	Missing or invalid bearer token	Fix auth header and token source	no
403	Tier restriction or policy gate	Use eligible account tier and org context	no
422	Schema validation failed	Correct fields/ranges/types before re-sending	no
429	Minute-level throttle or daily quota exceeded	Apply backoff and queue lower priority jobs	depends
500	Unexpected backend failure	Retry with backoff and capture diagnostics	yes

Retry and queueing strategy

if (status === 429 || status >= 500) {
  retryWithExponentialBackoff();
} else {
  failFastAndFixPayloadOrAuth();
}

Daily quota handling

Prioritize urgent client-facing jobs first
Queue non-urgent matching requests
Surface clear user messaging on quota exhaustion

Avoid retry storms

Never blindly retry all failures. Treat 4xx as corrective actions and 429/5xx as controlled retry paths.

Production reliability checklist

Log request IDs and response status codes
Track 429 volume over time
Implement backoff with jitter for retriable failures
Add quota-aware queueing in automation workflows