Blog ·

Jira SLA: set one up, find the breaches, and review them on a schedule

A Jira SLA is a timer Jira Service Management puts on each request: it starts on one event, pauses on another, stops on a third, and compares the time in between with a goal such as "resolve in 8 working hours". You build one under Space settings, then Request management, then SLAs. Plain Jira (the software project kind) has no SLA feature of its own, so the SLAs in this guide are a Service Management feature. Below: how to set one up, the parts that cause most wrong numbers, the JQL that finds breaches, and the review that keeps the targets honest.

Statements about Jira Service Management are as of October 2026, from Atlassian's own Cloud documentation. Atlassian now calls a project a space and an issue a work item. This guide uses both, because most admins still search for the old words.

What a Jira SLA is made of #

Atlassian's own summary is short:

Project admins can create SLA goals that specify the types of requests you want to track and the time it should take to resolve them.

Every SLA has four parts, and each one is a place it can go wrong.

Part What it decides Where you set it
Name What the timer is called in queues and JQL Add SLA, next to the clock icon
Conditions When time starts, pauses and finishes counting Conditions section of the SLA
Goals Which requests it applies to, and the time target Goals list, one row per goal
Calendar Which hours and days count Chosen per goal

The name is permanent. Atlassian says plainly that "You won't be able to change the name of your SLA once it is created", so choose one that says what it measures. The usual pair is Time to first response and Time to resolution, and those are the names Atlassian's own examples use.

How to set up an SLA in Jira Service Management #

  1. From your service space, select Space settings, then Request management, then SLAs.
  2. Select Add SLA and type the name.
  3. Under Conditions, add what starts the clock (for example, Work item created), what pauses it (optional, for example a Waiting for customer status), and what finishes it (for example, Resolution: set).
  4. Add goals. Each goal has a JQL filter under Apply to work items, a Time target, and a calendar.
  5. Select Save.

Two limits are worth knowing before you start. You can add up to 90 goals to one SLA, and Atlassian notes that a new SLA "is automatically applied to all relevant work items, both open and closed". That second point means a new SLA can paint last quarter's closed requests as breached the moment you save it. Warn whoever reads the reports before you do.

Conditions: start, pause, finish #

Conditions are where most strange SLA numbers come from. Atlassian's rule for start conditions is that "If you add multiple conditions, the SLA measures the time when a work item meets either of the conditions", and the finish side works the same way: measurement stops "as soon as any one of the conditions specified here are met".

Goals: one row per kind of request #

Each goal uses JQL to say which requests it covers. A typical set looks like this:

Goal Apply to work items (JQL) Time target Calendar
Urgent incidents priority = Highest AND type = Incident 4h 24/7
Other incidents type = Incident 8h Business hours
Service requests type = "Service Request" 24h Business hours
Everything else All remaining work items 40h Business hours

Atlassian also lets you group goals by priority, so one goal can carry a different target per priority instead of needing a row each. If priority changes while the clock runs, the docs say timers "adjust the remaining goal time based on the new priority", and time already counted is kept.

Calendars: which hours count #

A goal of 8 hours means 8 hours on its calendar. A business-hours calendar with a 09:00 to 17:00 day turns that into one working day; a 24/7 calendar makes it a third of one. Calendars also carry holidays, and our Jira holidays guide walks through adding them and importing a holiday file.

Atlassian adds a detail that catches distributed teams: "SLA calculations are based on the timezone of the SLA calendar", while the tooltip shows times in each viewer's own zone. Two agents in different countries can read the same timer and see different clock times for the same deadline. Our Jira time zone guide explains which setting decides what.

How to read SLAs in the queue #

Agents see each SLA on the queue and on the request. According to Atlassian's agent documentation:

Hover for the detail. The tooltip shows the time remaining or past, and the percentage of the goal used. Selecting the hours shows which calendar the SLA is counting against, which is the first thing to check when a number looks wrong.

Jira SLA JQL: find breaches before the customer does #

Each SLA becomes a field you can query. Atlassian's JQL reference for SLAs lists these functions:

Function What it finds
breached() The last SLA cycle failed its goal
everBreached() Any cycle failed its goal
paused() The current cycle is paused
running() The clock is currently running
completed() The cycle has reached a stop event
withinCalendarHours() The clock is inside or outside calendar hours
elapsed("…") Time since the cycle started meets a condition
remaining("…") Time left to the goal meets a condition

Names with spaces need quotes, which is why the SLA field always appears quoted. Three filters worth saving:

About to breach. Atlassian's example finds requests that will breach their resolution target in the next two hours, including ones already breached:

"Time to resolution" < remaining("2h")

Add AND "Time to resolution" != breached() to keep only the ones you can still save.

Breached this month. For the review below:

"Time to resolution" = everBreached() AND resolved >= startOfMonth(-1) AND resolved < startOfMonth()

Waiting on us, unanswered. Atlassian's own example for requests waiting less than ten minutes for a first response:

"Time to first response" < elapsed("10m")

Flip the operator to > and raise the figure to find the ones nobody has picked up. Save each filter and put it on a dashboard; our Jira dashboard examples show layouts that fit them.

Editing an SLA without wrecking the history #

Atlassian is unusually candid about what an edit does. Completed cycles "stay as they are". Ongoing cycles are recalculated with the new setup, but some can be discarded: the docs warn that "Editing an SLA can cause some open cycles to close, restart, or disappear", most commonly when a stop condition is changed to an event that already happened.

Practical rules that follow from that:

The same page also says to "Review SLA configurations regularly to ensure goal definitions and conditions reflect your current workflows". That sentence is the part most teams never get round to.

The SLA review: the step Jira does not schedule #

Setting the targets is a one-off. Keeping them true is recurring work, and it is the step that quietly stops. A sensible cadence:

Cadence What to look at Who
Weekly Requests breached last week, and why Service desk lead
Monthly Breach rate per goal, pauses that look abused, goals nobody hits Lead and the team
Quarterly Do the goals still match what customers were promised? Calendars and holidays for next quarter Space admin and whoever owns the agreement

The review needs a ticket with the saved filters linked, the questions written down, and an owner. A good monthly one reads:

The checklist steps can live in the description or as subtasks; our Jira checklist template guide compares the options.

Three ways to make the SLA review appear on time #

Option How it works Watch out for
A calendar reminder Someone creates the ticket when the reminder fires Stops when that person is away
Jira Automation A Scheduled trigger with a Create work item action Each run counts toward your automation usage
A recurring-work app A rule in the project creates the ticket on schedule Another app to install and license

Jira Automation can do it with a cron schedule; our Jira cron expression guide has the monthly patterns, and Jira Automation pricing covers what each run spends.

Recurring Work Items for Jira, our app, is the third row. You write a monthly rule once in the service space: the title can carry the month through a date token, subtasks carry the checklist, and the rule can skip weekends and holidays. Every run, including a skipped or failed one, is written to the rule's history with the reason, so a missing review is something you can see rather than something you discover a quarter later. It uses no Jira Automation quota, is scoped to one project, and is free for up to 10 users. See how it works for the setup.

When you do not need any of this #

Summary #

A Jira SLA is four decisions: a permanent name, start and stop conditions, goals filtered by JQL, and a calendar. Get the conditions right, save a breach filter, and edit with care, because changing conditions can make open timers disappear. Then put the monthly review on a schedule, since the targets are only as good as the last time someone checked them.