The Production Readiness Playbook

A written readiness bar exists and the team agrees on it

Must-have criteria are separated from nice-to-have ones

The rollback procedure has been executed in staging, not just documented

Any non-reversible steps (e.g. destructive migrations) are identified in advance

At least one alert has been test-triggered end to end

Alerts cover the specific dependencies this launch relies on, not just generic uptime

A specific person or rotation is named, not 'the team'

That person knows they're on point and how to escalate