How to forecast a release date in Jira
The honest answer to “when will it ship?” is a range with odds, not a single date. Here is how to get one from the data Jira already holds.
Why single dates fail
A single date hides the uncertainty everyone already feels. Work arrives, people go on leave, a bug eats two days. When the date slips, trust slips with it. A forecast that says “most likely 14 March, safe to promise 21 March” gives stakeholders something they can plan around.
What Jira gives you out of the box
In company-managed Scrum projects, the Version report predicts a release date from your velocity, provided every issue carries an estimate. Advanced planning in Jira Premium can also schedule work on a timeline. Both depend on estimates being complete and accurate, which is rarely the case in a live backlog.
A method that needs no estimates
Count how many work items your team actually finished each day over the last few months. That history already contains holidays, interruptions and bad weeks. Then replay it thousands of times against the work that is left:
- Pick a random day from the history and take its count of finished items.
- Subtract it from the remaining work. Repeat, one day at a time, until nothing is left.
- Note how many days that took. Do it 10,000 times.
Sort the 10,000 results. The middle one is your likely date. The one at 85% is the date you can promise: 85 simulated futures out of 100 finish by then. This is the Monte Carlo method, and it is the same idea weather forecasters use.
What to count
- Count stories, tasks and bugs. Leave out epics and subtasks so nothing is counted twice.
- Use the date an item reached a “Done” status, not when it was created.
- Look back 8 to 12 weeks. Shorter is noisy, longer includes a team that no longer exists.
Reading the result
Share the 85% date with stakeholders and keep the 50% date for the team. When scope grows, rerun the forecast: the dates move, and everyone can see why.
Doing it in two clicks
Clearlane Forecast runs exactly this method inside Jira, on the project, version or epic you pick. Your data never leaves your Atlassian site.