Incident Severity Classifier

P0 (Critical) incidents represent complete service outages affecting all users or causing data loss/corruption — requiring immediate all-hands response within 15 minutes. P1 (High) incidents are major degradations affecting a significant portion of users or critical business functions, requiring response within 30 minutes. The key distinction is scope and reversibility: P0 impacts are immediate, widespread, and potentially irreversible.

Blast radius measures how many services, teams, or customers are affected. A single microservice failure isolated to one team is low blast radius; a shared authentication service failure is extremely high. When blast radius is high, escalate the severity by at least one level regardless of other factors, because cascading failures can rapidly worsen the situation.

A workaround can justify reducing severity by one level only if: it is documented, reproducible, doesn't require special access, and the customer experience degradation is minimal. If the workaround requires customers to change their behavior significantly or affects SLA-covered functionality, it should not reduce severity.

P0 escalation should follow: on-call engineer → on-call manager (within 5 min) → VP Engineering (within 15 min) → CTO/CEO for prolonged outages (>30 min). A dedicated incident commander should be assigned to coordinate communication, separate from the engineers actively debugging. War room / dedicated channel should be opened immediately.

Revenue impact per hour = (annual revenue / 8760 hours) × affected percentage. For customer-facing services, also include indirect costs: SLA penalties, customer churn risk, support ticket volume. For internal tools, use productivity loss: (average engineer hourly cost × engineers blocked). Document your baseline so you can make consistent severity decisions under pressure.