Blog ·
Incident management in Jira: the response, the review, and the routine
Incident management in Jira happens in three places, and most teams only set up the first one. The response, meaning alerts, on-call paging and the incident record itself, belongs in Jira Service Management. The review after the incident, with its action items, can live in any Jira project as ordinary work. And the routine that keeps the next incident small (runbook reviews, restore tests, the monthly postmortem review meeting) only exists if something puts it on a board on time. This guide walks through all three, says which Jira product does which part, and spends the most time on the part teams forget.
Statements about Atlassian's products and dates are as of October 2026, read from Atlassian's own pages.
The three parts of incident management #
An incident has a before, a during and an after. Jira handles each one with a different tool, so it helps to name them before choosing anything.
| Part | What happens | Where it lives in Jira | What goes wrong without it |
|---|---|---|---|
| Response | Alert fires, someone is paged, the incident is worked and resolved | Jira Service Management | Nobody is woken up, or everybody is |
| Review | The team writes up what happened and agrees on fixes | A post-incident review plus action items as work items | The same outage happens twice |
| Routine | Readiness work on a clock: drills, runbook checks, review meetings | Recurring work items in a team project | The fixes and drills quietly stop |
The response gets the tooling attention, the review gets the templates, and the routine gets nothing by default. That third row is where incident management decays, months after everybody agreed it mattered.
Part 1: incident response in Jira Service Management #
Plain Jira Software has no concept of an alert or of a person being on duty. If you want an alert from your monitoring tool to reach a human within minutes, the answer is Jira Service Management, which is Atlassian's product for that job.
This is also where the Opsgenie question lands. For years Atlassian's alerting and on-call product was Opsgenie, sold separately. Atlassian's migration page now says:
Now, Opsgenie's capabilities are being moved to Jira Service Management.
The dates that matter, from the same page:
| Milestone | Date | What it means |
|---|---|---|
| End of sale | June 4, 2025 | No edition upgrades or new sites for Opsgenie |
| End of support | April 5, 2027 | The deadline to move Opsgenie data to Jira Service Management |
Atlassian also states that "Opsgenie customers have until April 5th, 2027 to complete their move to Jira Service Management", and that the move itself is automatic once it is started by "an Opsgenie owner who is also a site admin of your account." If your team still pages through Opsgenie, the migration is a dated piece of work, and it belongs on somebody's board now rather than in March 2027.
For a new team the practical setup in Jira Service Management is:
- Connect your monitoring tools so their alerts arrive in one place.
- Build an on-call schedule with rotations, so an alert reaches the person on duty. We covered schedules and handover tickets in Jira on-call rotation.
- Define severity levels, and decide which severity counts as a major incident.
- Agree who declares an incident, who leads it, and who talks to customers.
Keep the response in the tool built for paging. A plain Jira board is a poor place to run a live outage, because nothing on it wakes anyone up.
Part 2: the post-incident review #
The review is where an outage turns into fixes. Atlassian's own guide puts it plainly:
An incident postmortem, also called a post-incident review, is a structured, blameless review held after an incident to establish what happened, why, what was done to resolve it, and what changes will prevent it from happening again.
The same guide is specific about timing: the review meeting is ideally "held within 24-48 hours of the incident resolving", with a firm outer limit of under a week. It also suggests using severity as the trigger, for example "any incident Sev-1 or higher triggers the postmortem process", with the review optional for smaller incidents.
A useful postmortem in Jira has four pieces:
- A timeline: first alert, first announcement, each remediation attempt, resolution.
- A root cause and an impact statement: what broke, for whom, and for how long.
- Action items as work items, each with one owner and a due date.
- A link from every action item back to the incident, so a later reader can follow the thread.
Atlassian describes Jira Service Management doing some of this for you: it "creates a post-incident review from the incident record, with its timeline and responders already attached" and "assigns action items as tracked work." If you are not on Jira Service Management, a review template in Confluence and a work item type called "Postmortem action" in your team project gets you most of the way. For a template you can reuse, the patterns in Jira task template apply directly.
Where action items go to die #
The weak point is not writing the action items. It is finishing them. An action item from a Sev-1 review is urgent for about a week. After that it competes with planned work, and planned work has a sprint, a product manager and a deadline. The action item has a ticket and a good intention.
Atlassian's guide answers this with a meeting:
Schedule a recurring meeting with engineering (and anyone else who may have an interest, like customer support or account managers), at least monthly, to review incident postmortem reports.
That meeting is the review of the reviews. It is also exactly the kind of work that slips, because it is never urgent on the day it is due.
Part 3: the routine that keeps the next incident small #
Most incident readiness is boring and recurring. None of it is triggered by an event, so none of it is created by an alert or a rule that listens for one. Here is a typical calendar for a team that runs a production service:
| Routine | Cadence | Why it matters |
|---|---|---|
| Postmortem action review meeting | Monthly | Open action items get an owner and a date again |
| Runbook review for the top services | Monthly | A runbook written for last year's architecture misleads at 3 a.m. |
| Backup restore test | Monthly or quarterly | A backup nobody has restored is a hope, not a backup |
| Alert noise review | Monthly | Alerts nobody acts on train people to ignore alerts |
| Incident drill or game day | Quarterly | The first time you fail over should not be the real outage |
| Escalation contact check | Quarterly | Phone numbers and owners change when people move teams |
| Opsgenie migration milestones, if you use it | Once, before April 5, 2027 | A dated deadline with a fixed end |
Each of these is a ticket you can create by hand on the first of the month. That works until the month of a big incident, which is exactly the month the restore test and the action review matter most.
Three ways to make the routine appear on its own #
| Approach | How it works | Good for | Watch out for |
|---|---|---|---|
| Clone a master ticket | Keep one ticket per routine and clone it each cycle | Two or three routines, one reliable owner | It depends on the person who remembers |
| Jira Automation scheduled rule | A scheduled trigger plus a Create work item action | Teams already writing Automation rules | Rule runs count toward your plan's usage, see Jira Cloud automation limits |
| A scheduling app | Write the schedule once per routine, the work item appears on time | Many routines, holidays, time zones | It is one more app to install |
We compared these in depth in Jira recurring tasks, and the cron side of Automation is covered in Jira cron expression.
Recurring Work Items for Jira is our app, and this routine is the job it was built for. You write a rule once in the team's project: a monthly or quarterly schedule, a template with summary, description, assignee and priority, and subtasks for the checklist. For the restore test, the subtasks might be "pick a backup", "restore to a scratch environment", "verify the data", "record the time it took". Titles can carry dates through tokens, so each month's ticket is named for its own month.
The parts that matter for incident routines:
- Real time zones. A rule runs at a wall-clock time in a real time zone, so a 09:00 drill reminder stays at 09:00 through a clock change.
- A working-day calendar per project. A run that falls on a holiday can run anyway, move to the next working day, move to the previous one, or skip.
- A due date stored as an expression, such as a number of days after creation, so an unfinished drill shows up as overdue.
- A run history that records every run as created, skipped or failed, with a reason.
- Scoped to one project, so the incident team's routine does not create tickets across every project on the site.
- No Jira Automation quota. Rules run on the app's own scheduler.
It does not page anyone and it does not replace Jira Service Management. It makes sure the readiness work exists on the board when it is due.
Putting it together #
A small team can run sensible incident management in Jira with a few decisions:
- Response in Jira Service Management, with an on-call schedule and severity levels. If you are on Opsgenie, put the migration on the board with its date.
- A review for every incident above your severity threshold, held within a few days of resolution, with action items as work items that link back to the incident.
- A recurring monthly review of open action items, created automatically, assigned to whoever runs the incident process.
- The readiness calendar as recurring work items: runbook review, restore test, alert review, drills and contact checks, each on its own cadence.
For the change side of the same picture, see Jira change management template. For the ops routine beyond incidents, such as patching and access reviews, see Jira for IT operations and Jira patch management.
When the simple way is enough #
Not every team needs the whole structure. Keep it light if:
- Your service has no users who notice an outage within minutes.
- Incidents happen a few times a year and the same two people handle all of them.
- A shared calendar reminder for the monthly review has never been missed.
In that case, a Confluence page for each review and a label on the action items is enough. Add paging when an alert has to reach a human fast. Add automatic routine tickets when you notice that the restore test has not happened since spring, or that last quarter's action items are still open.
Frequently asked questions #
Can you do incident management in Jira Software alone? You can track incidents and action items as work items, but Jira Software does not page anyone. For alerts and on-call you need Jira Service Management.
What is happening to Opsgenie? Atlassian is moving Opsgenie's capabilities into Jira Service Management. End of sale was June 4, 2025, and end of support is April 5, 2027, per Atlassian's migration page as of October 2026.
How soon should a post-incident review happen? Atlassian's guide suggests a review meeting within 24 to 48 hours of resolution, and well inside the following week.
How do I stop postmortem action items from going stale? Give each one owner and a due date, link it to the incident, and hold a recurring review of open items at least monthly. Creating that review ticket automatically removes the step that people forget.
Summary #
Incident management in Jira is three jobs. Jira Service Management runs the response. A blameless review turns the incident into action items with owners. And a recurring routine (the monthly action review, runbook checks, restore tests and drills) keeps the fixes moving and the team ready. The first two get set up because an incident forces them. The third has to be put on a clock on purpose.