Add escalation steps to a rule
Make an alert reach a second and third person when nobody acknowledges it, or when the first phone call fails.
In the sidebar, open Critical Alerts, then Setup, then Rules. The page address is /critical-alerts/rules. Press Edit on the rule, then scroll to Escalation.
An escalation step is a follow-up delivery that runs while the incident is still active: call the manager 10 minutes after the technician, then the owner 10 minutes after that.
Before you start
- Your admin has to give your role permission to manage critical alert rules.
- Create a destination for each person or group in the chain. See Add a destination.
Steps
- Press Edit on the rule. The Edit rule drawer opens.
- Under Escalation, choose the Timing anchor. Incident created time counts every delay from when the incident opened; Previous notification sent time counts from the last message sent.
- Turn on Escalate on delivery failure if the next step should run right away when a phone call is busy, unanswered or fails.
- Press Add escalation step. A Step 1 block appears.
- Set Delay minutes (how long after the first delivery this step becomes due) and choose its Destination keys. Add a Webhook URL if another system should be told as well.
- Set Repeat count and Repeat interval minutes if this step should try more than once.
- Turn on Bypass quiet hours for steps that must deliver even during a destination's quiet hours; Continue after acknowledgment only if the chain should keep going after someone has claimed the incident; and Trigger on previous failure so this step runs when the previous call fails.
- Press Add escalation step again for each further person. Use Move up, Move down and Remove to arrange them.
- Save the rule with the button at the bottom of the drawer, then press Publish on the rule's row.
What you see
- Each step reads "Runs after the initial destinations when the incident is still active."
- With no steps, the block reads "No escalation steps. The rule will notify only the initial destination keys unless you add a follow-up step."
- The rule's row now shows the number of escalation steps after its destination keys.
If something looks different
- Steps never fire: the incident was acknowledged before the delay ran out, and Continue after acknowledgment is off. That is the usual and intended behaviour.
- A step fires sooner than expected: Timing anchor is Previous notification sent time, and a repeat notification reset the clock.
- Escalation runs on a timer. To push due steps through right now, see Run escalations and monitors now.