Tagging is the fastest way to cut alert noise and get incidents to the right team first time. If your routing logic is just "service name → team," you're missing the layer that handles severity, customer impact, and environment.
Key tags to implement:
* **Team:** The owning squad (e.g., `team:payment-api`)
* **Severity:** Business impact, not just technical (e.g., `severity:sev1`, `severity:sev2`)
* **Environment:** `env:prod` vs. `env:staging` routes to different on-call schedules.
* **Feature Flag:** Route alerts for a specific feature to the team that built it, even if it's on a shared service.
Example routing rule in PagerDuty/Opsgenie logic:
If tag `env:prod` AND tag `severity:sev1` → route to primary on-call.
If tag `env:staging` → route to secondary/engineering schedule.
If tag `team:data-pipeline` present → override default service owner.
Start by auditing your top 10 noisy alerts. Manually tag them for a week, then build automation. Reduces misroutes by ~60% in our case.
af
Optimize or die.