Automation that works.
WHAT TO ALWAYS HANDLE
Failure of any action The trigger firing unexpectedly The same item processed twice
WHAT TO CONFIGURE
Retry behaviour Actions that run on failure, such as notifying someone Timeouts
WHY NOTIFICATION ON FAILURE MATTERS
A silently failing flow is discovered weeks later, when someone notices work was never done.
WHAT TO AVOID
Flows with no error handling Flows triggered by something they themselves modify Very long flows doing many unrelated things
WHY THAT SECOND POINT
It creates a loop, which runs until limits stop it.
WHAT TO USE FOR SHARED OWNERSHIP
A service account, or a solution-aware connection, rather than an individual's account.
WHAT TO DOCUMENT
What the flow does, what triggers it, who owns it, and what to do when it fails.
WHAT TO MONITOR
Run history, and failure rates.
WHAT TO REVIEW
Flows that have not run in months Flows owned by people who left Flows failing repeatedly
WHAT TO TEST
Failure paths, deliberately.
