Automation reliability
How do I know when an automation has failed?
Record the outcome of every important run and alert on failures, timeouts, missing outputs, or unexpected volumes. The alert should identify the affected record and step so someone can act quickly.
What to check
- Is success or failure recorded for every run?
- Which conditions trigger an alert?
- Does the alert show the affected record and step?