PagerDuty Notifications

The Statsig PagerDuty integration pages your on-call team when a topline alert fires, so metric regressions reach the people who respond to incidents. PagerDuty is a first-class entry in the Console Integrations catalog: you configure one or more named PagerDuty services once, then select those services in any topline alert's notifications. On-call responders and platform teams use it to turn Statsig metric alerts into PagerDuty incidents without copying routing keys into each alert.

How the PagerDuty integration works

When you add PagerDuty as a notification target on a topline alert, you pick from a dropdown of the services you've already configured rather than pasting a routing key. When the alert fires, Statsig pages every service you selected. You can select more than one service on a single alert.

Prerequisites

• Admin access to your Statsig project's integration settings.
• A PagerDuty service with an Events API v2 integration, and its routing key.

Configure a PagerDuty service

Add each PagerDuty service to the Integrations catalog before you reference it in an alert.

1. In your Statsig Console, go to Statsig Settings > Integrations.
2. Select PagerDuty.
3. Click Add Service.
4. Enter a name that identifies the service to your team.
5. Paste the PagerDuty Events API v2 routing key for that service.
6. Click Save.

Repeat these steps for each PagerDuty service you want available to alerts. Statsig stores each routing key as an obfuscated secret.

Add PagerDuty to a topline alert

1. Open the topline alert you want to page on, or create a new one. For details, refer to Topline Alerts.
2. In the alert's Notifications, find the PagerDuty section.
3. Select one or more configured PagerDuty services from the dropdown.
4. Save the alert.

When the alert fires, Statsig pages each PagerDuty service you selected.

Test the wiring

Use Test Notifications on the alert to confirm the connection. Statsig pages the selected service or services with an event tagged [TEST]. Test pages don't collide with real incidents, so you can verify routing without triggering an on-call response to a genuine outage.

Limitations