Docs / Alerts and incidentsBrowse

Escalation and on-call

Without a policy, everyone on the alert list hears about every outage once. An escalation policy pages people in order instead, until someone says they are on it. On-call schedules decide whose turn it is.

On this page
  1. Create an escalation policy
  2. Set up an on-call schedule
  3. Swap a shift
  4. Acknowledge an incident
  5. Escalate now

Create an escalation policy

Plan/Pro and up. Below Pro, every alert goes to everyone once

  1. 1.
    Open Alerting in the sidebar and press Create a policy (or New policy once you have one).
  2. 2.
    In the first step, set Notify after (0 means at once) and press Add who to notify.
  3. 3.
    Press Add a step for the next person, with a longer delay: “If still unacknowledged after 15 minutes”.
  4. 4.
    Switch on Repeat the last step if it should keep paging until someone answers.
  5. 5.
    Switch on Workspace default to use it for every site without a policy of its own (the editor says how many), then press Create as default. Left off, only the sites that pick it follow it.
The New escalation policy dialog with two steps
New escalation policy: each step pages its people if nobody has acknowledged yet.
Each step can notifyWhat they get
Site alert list
The owner, admins, the site’s site admins and its extra emails.
Whoever is on call
The person whose turn it is in a schedule.
People
An email, a push to each phone with the app (unless they turned push off for on-call pages), and an SMS or call if they turned those on.
Channels
A message in Slack, Teams, Discord, Telegram or a webhook.

A policy has up to 10 steps. Delays count from when the incident opened, up to 7 days. The last step can repeat every 1 to 1440 minutes, at most 100 times. To give one site its own policy, pick it under Escalation in the site’s settings.

NoteA site that follows a policy, its own or the default, is paged by its steps only: its email alerts and channel routing are not used for incidents. Add a channel to a step to keep it in the loop; the channel list and the site’s settings say which sites a policy took over.

Set up an on-call schedule

  1. 1.
    Open Alerting, then the On-call tab.
  2. 2.
    Press Create a schedule (or New schedule). Pick a Rotation: daily, weekly, or a custom number of days or hours.
  3. 3.
    Set when it Starts, the Handoff time and the Time zone.
  4. 4.
    Under Who takes turns, add people in order, then press Create schedule.
  5. 5.
    In a policy step, pick Whoever is on call and the schedule.

A shift lasts up to 28 days or 168 hours, with up to 50 people. Day shifts hand off at the same local time, across daylight saving changes. Each schedule shows who is on call now, who is next and the coming three weeks.

Swap a shift

For a swap, a sick day or a holiday, press Override on the schedule. Pick who is On call and the dates, up to 90 days. Where two overrides overlap, the newest wins. Anyone can put themselves on call; owners and admins can put anyone.

Acknowledge an incident

Acknowledging tells your team you are on it. The remaining steps and repeats stop. You can acknowledge from:

  • the Acknowledge link in the alert email, Slack, Teams, Discord or Telegram message. It works without signing in, for 7 days;
  • the Acknowledge button on a push, even on the lock screen;
  • the incident page, or the red banner on the site page;
  • the incident in the mobile app;
  • the API or the MCP server.

The others it paged, the people its steps name, who is on call and, without a policy, the site’s alert list then get a push saying who acknowledged it and when. The person who acknowledged gets none. Tapping it opens the incident. Turn it off with Acknowledged by someone else in your push settings.

Escalate now

If you need help before the next step is due, press Escalate now. It pages the next step at once; later steps keep their gaps. After the last step, it sends the next repeat.

On the web, the button is at the top of the incident and on the red banner of the site page; the incident then shows who the step reached. In the app, it is on the incident and on the push.

NoteEscalate now needs a policy, and an incident that is open and not yet acknowledged. See Incidents.

Still stuck? Write to us.