Retry-then-confirm logic means alerts are rare enough to trust, and an escalation path means a real incident never sits unread.
Two consecutive failed runs turn a suspicion into an incident — single blips stay quiet.
Sent straight to Slack or email, with the evidence attached, not just a link to dig through.
If nobody acknowledges it, it moves up the chain until someone does.
Incident timeline showing every retry and confirmation step
Evidence bundle attached — screenshot, logs and the failing step