WARD Ops

Most “failed escalations” aren’t technical.
They’re gaps no one admitted existed.

Last month we watched a critical alert hit three channels:

• Slack
• Email
• PagerDuty

All delivered.
No one responded.

Why?

• The rotation was outdated
• The Slack channel was archived
• And the team assumed the SRE lead “probably saw it"

Escalation didn’t fail.
Ownership did.

These metrics look great on a slide:
• MTTD (Mean Time to Detect)
• MTTA (Mean Time to Acknowledge)
• MTTR (Mean Time to Resolve)

But here’s what they don’t show:
• How was it detected
• How the escalation chain actually worked
• Whether the problem stayed fixed

We’ve seen teams hit a 45-second MTTA - and still need 20+ minutes to escalate.
Why? No ownership. Slack noise. Outdated runbooks.

Metrics ≠ resilience.
If your system can’t take a punch at 2 AM, the graph won’t save you.