Escalation Policy Basics

Follow

Escalation policies make sure the right people are notified at the right time and is the connector between services and schedules — a service can only have one escalation policy, though an escalation policy can be associated with multiple services.

Creating an Escalation Policy

Escalation policies are found in the web UI under Configuration → Escalation Policies, where there will already be a Default escalation policy. By default, the Account Owner will be the only target on this escalation policy, though this can be changed at any time, of course.

To create an escalation policy:

  1. Go to Configuration → Escalation Policies.

  2. Click New Escalation Policy.

  3. Name your escalation policy and optionally add a description.

  4. Select the targets (on-call schedules and/or users) you want to notify immediately after an incident is triggered.

  5. If you would like to add a second level to the escalation policy, choose the number of minutes that should pass, without receiving an acknowledge or resolution, before the incident escalates (escalation timeout).

  6. Then click Add a new Escalation Rule to create the second level.

  7. Select targets for the second escalation rule.

  8. Configure additional escalation rules as required.

  9. Recommended: Check the box next to If no one acknowledges, repeat this policy __ and select the number of times to loop through the escalation policy (maximum is 9 times).

  10. When finished, click Save

Escalation Policy Example

In the example below, if an on-call user in the Primary On-Call Schedule schedule does not acknowledge an incident in 30 minutes, then it will escalate to Jason Collison. The escalation timeout at the second level is also 30 minutes, and Jason will have 30 minutes to acknowledge the incident before it escalates and starts over from the beginning of the policy. This escalation policy is set up to repeat 9 times.

Default_-_PagerDuty.png

Escalation Targets and Limitations

You can select users or schedules as targets on an escalation policy. When an incident triggers, we'll attempt to notify responders at the first level of the escalation policy. If the incident is not acknowledged within the escalation timeout period, then it will escalate to the on-call responders at the next escalation level, and so on down the line.

If nobody is on-call in the first level, then we will assign the incident to the on-call responder at the subsequent escalation level.

Important note: An incident follows the escalation policy as it was at the time the incident triggered. This means that if changes are made to an escalation policy, these changes will not affect any open (triggered or acknowledged) incidents associated with that escalation policy. 

Limits on escalation policies

  • You can add up to 20 escalation rules to an escalation policy.

  • You have the option to repeat an escalation policy 9 times. Once an escalation policy reaches the amount of loops configured, the incident will stay assigned to the last user and will not continue to notify once it has cycled through all of the responder's contact methods.

  • You can notify up to 10 users/schedules (25 on Standard and Enterprise plans) in a single escalation rule.

  • The minimum escalation timeout is 1 minute if there is a single target on an escalation rule. Alternatively, if there is more than one target in an escalation level the minimum escalation timeout is 5 minutes. These restrictions are displayed in the web UI.

Have more questions? Submit a request

Comments

  • Avatar
    Rishi Pande

    Can you please document a scenario here? It is very important to explain what the "Escalate after XX minutes" actually means.

    In the example figure above, what it means is that if the Primary On-Call Schedule does not acknowledge the incident in 30 minutes, it will get escalated to Jason Collison (this is based on the Escalate after field next to Primary On-Call). Since the figure above has 30 minutes as an example again, that means that if Jason doesn't respond in 30 minutes, it will get escalated - only this time, back to Primary On-Call.

    I would love the PD folks to post a better scenario so that people don't make any basic mistakes about this while setting  up escalation policies. :)

  • Avatar
    Dan Khersonsky

    Hi Rishi,

    Thank you very much for your feedback. I've made some changes to our guide based on your suggestions. Please let me know if you think this helps clarify the use of "escalation timeout" setting, or if you have any additional feedback.

     

    Best,

    Dan

  • Avatar
    Nicholas Harring

    It'd be incredibly helpful if escalation didn't have to reassign, for example I manage a technical team but do not directly work technical issues. I want to be notified if my on-call person hasn't ack'd the issue in 20 minutes, but I don't want ownership of the issue. Ideally I could have the on-call person retain ownership of the issue and they would receive additional notifications based on configuration and I would receive escalation notifications as the manager when my staff isn't responding. 

  • Avatar
    Ramin Surya

    I am trying to create two different escalation policies. Each would map to different on-call schedules (weekdays vs weeknights). I found out that I can only tie a service to one escalation policy. Is there another way to do this?

  • Avatar
    Ryan Hoskin

    Hi Ramin,

    Please contact us at support@pagerduty.com so that we can assist you with your configuration.