Limits and notices
10,000 work items per search
Each chart's search reads at most 10,000 work items. If more match, the chart shows what it has and says: "The search matched more than 10,000 work items, and the chart uses only the first 10,000. Choose a narrower source or a shorter window for complete numbers." Counts are then too low, and percentiles describe only the work items that were read.
What counts toward the 10,000 depends on the chart:
- Throughput, Cycle Time, Time in Status: the work items resolved in the window.
- Created vs. Resolved, Scope Change: the work items created or resolved in the window.
- Work Item Age: all unresolved work items, including those never started, in one search; the finished work items for the reference lines in a second.
- Work in Progress: all unresolved work items, including those never started, plus those resolved in the window. A large backlog can therefore reach the limit even when little is in progress.
- Monte Carlo forecasts: the resolved work items in the history window. Above 10,000, the forecast keeps the most recent ones, so the oldest periods are undercounted, and the chart says: "Partial data: forecast may be skewed".
Approximate counts
Three numbers come from Jira's count rather than from the work items themselves: the starting value of the Unresolved subplot in Created vs. Resolved, the Initial scope in Scope Change, and the open work items in the Monte Carlo "When" forecast. Each of them needs only how many work items match, not the work items themselves, so the chart asks Jira for the count in a single request instead of reading every work item. That is why the count is not limited to 10,000.
Jira takes the count from its search index. The index picks up changes shortly after they are made: according to Atlassian, most within seconds, while a change to many work items at once, such as a bulk edit, can take longer. A work item created, resolved or moved just before the chart loads can therefore still be missing from the count, or still be counted although it no longer matches. That is what makes the count approximate: it is off by the work items changed in that short time, usually a few at most, more after a bulk edit. The charts mark these numbers with "≈".
The work items a chart reads come from the same search, so a change from the last few seconds can be missing there too. The difference is that Jira returns the count as an estimate, and the charts say so.
Opening work items
A click opens up to 300 work items in Jira's search. For more, the tooltip says "Too many to open". Jira shows the work items as they are now, not as they were in the period you clicked.
Time windows
"Days previously" allows up to 720 days, 180 for the Work in Progress Chart, and 14 days with hourly grouping. The first and last periods of the period-based charts are explained on How the numbers are calculated.
Freshness
A chart loads its data when the dashboard loads. Charts on the same dashboard that need the same data share it, and your browser keeps it for five minutes, so a reload within that time can show the same numbers. "Refreshed …" shows the age of the oldest data on the chart.
Errors
If Jira cannot be read, the chart shows "Error:" followed by Jira's message.