Route alert episodes by severity in the experimental alerting system

When your rules in the experimental alerting system produce alert episodes at different severity levels, you can route them to different workflows by creating separate action policies that are scoped to specific severity values using match conditions.

For example, you might page an on-call team for critical episodes while sending lower-severity episodes to a Slack channel for async review.

Each action policy evaluates alert episodes independently.

• An episode with severity: "critical" matches Action Policy A but not Action Policy B.
• An episode with severity: "high" matches Action Policy B but not Action Policy A.
• If an episode's severity changes mid-lifecycle, the action policies that match it change accordingly. For example, if an episode escalates from high to critical, Action Policy A starts matching and Action Policy B stops matching. Action Policy A fires because it has no prior notification record for that episode.

• Manage severity escalation notifications explains what happens when an episode that has already matched an action policy changes severity.
• Action policy reference for match condition fields, grouping modes, and frequency options.
• Create and configure an action policy to set up an action policy with match conditions.