What to monitor
- Run success: errored rows per run — trend, not just the latest run.
- Match rates: per provider, per workflow. A sudden drop means a provider or input change.
- Cost: Actions and credits by table; spikes indicate loops or new expensive columns.
- CRM sync: rejected writes and their reasons.
Error handling patterns
Retry transient errors (timeouts) with backoff and a cap — unbounded retries multiply Actions. Distinguish failure types in the workflow: empty result (data problem) vs error (system problem) get different handling. Route persistent failures to an exceptions table with the reason, reviewed weekly.
Alerting
Alert when: error rate exceeds threshold, a critical workflow fails to run on schedule, or credit burn runs ahead of budget pace. Route to Slack/email where the owning team actually looks. An alert nobody reads is decoration.
Need help applying this?
Describe your project and the ClayForEnterprises team will review it — the resources stay free either way.