Skip to content
Menu
HelpHow the numbers are calculated

How the numbers are calculated

Sumline adds up four fields from the bottom of the work item hierarchy to the top: subtasks, the stories, tasks, bugs and requests above them, epics and a level above epics: initiatives on Jira Premium, or, on any plan, a work type an admin rolls up through links. Children that live in another space are counted under their parent too.

The four fields

  • Story points, shown as done / total.
  • Original estimate: what was planned. Its done / total shows how much of the plan sits on finished work items.
  • Logged (Jira’s Time spent): what was logged, ever, so it only goes up. It is added up at every level, whatever else is estimated.
  • Remaining: what Jira currently thinks is left. It goes down as work is logged, but people also re-estimate it by hand, so it is not original minus spent. On a done work item it is not counted, by default; see “Remaining on done work items” below. Where Jira has none at all, see “Remaining missing in Jira”.

Sumline takes each field as Jira has it, with that one exception. Times are shown in hours and minutes. An admin can leave any field out; see Settings.

Your own number fields

An admin can also add up to three of the site’s own number fields, such as “Effort” or “T-shirt days”; see Settings. Each one is added up like story points:

  • shown as done / total, under the name the field has on your site;
  • counted once when a parent and its children both carry a value, by the same rule as the estimates below, with the values left out named;
  • a work item with a value in it counts as estimated.

They appear in the Sumline panel and its child table, the Details column, the tables (a column each after the points) and the Excel and CSV downloads, and each can be the “% complete by” basis on the tables. The gadget’s rows show only the progress bars and warnings, so it has no column for them. % complete by estimate is still story points, else time.

What counts as done

Any work item whose status is in Jira’s Done status category. Status names are never used, so renamed or custom statuses such as “Shipped” count as done as long as they sit in that category.

Each estimate counted once

When a parent and its children both carry an estimate, Sumline counts one level, never both. A story estimated at 8 whose subtasks also add up to 8 counts as 8, not 16.

By default the children win and the parent’s own value is set aside. An admin can choose “parent wins” instead, which sets the children’s values aside. Either way nothing disappears silently: the Sumline panel names each value it did not count, with its work item and field, the tables mark the row “Mixed estimates” in the Warnings column, and the downloads list the work items in their Exclusion note column. Hover the marker to see what was set aside and what was counted instead, for example “Child work items’ estimates are counted instead of the parent’s own, so nothing is counted twice: HLAB-1 original estimate 80h → 592h on 4 child work items.” Time spent is not affected: logged time is always added.

A 0 set aside changes nothing, so it is not named. Jira writes 0m remaining by itself when time is logged to the end, so a parent showing 0m under children that carry estimates is not marked Mixed estimates.

Under “parent wins”, a parent’s own original or remaining estimate of 0 counts as none, so its children’s estimates count instead: a story at 0m remaining over subtasks with 4h and 2h left shows 6h, not 0m. A parent’s estimate that is not 0 still wins.

When the value set aside is a row’s own, the tables say so under that figure too, Points, a number field, Original estimate or Remaining, for example “own 4h not counted”. Hover it for the same explanation, for that one value.

Remaining on done work items

Jira does not clear a work item’s remaining estimate when the work item is closed: it only goes down as time is logged. A subtask closed without logging its time keeps its whole estimate, and an epic whose work is all done would still show hours left.

So, by default, Sumline does not count the remaining estimate of a done work item, and says what it left out:

  • the Sumline panel names the work items and the amount in a grey line under its figures, for example “Remaining on done work items: SLC-81 and SLC-84. 7h is not counted”;
  • the tables and the gadget mark the row “7h remaining on done” in the Warnings column;
  • the Excel and CSV downloads add the same sentence to the row’s Exclusion note.

The work item still counts as estimated, and its original estimate and time spent are added as before. Only the remaining estimate, and the “% complete by time” worked out from it, change. A value already set aside because a parent and its children both carry an estimate is named there, not twice.

An admin can switch this off, under Settings, to count Jira’s remaining estimate as it is.

Remaining missing in Jira

When work is logged, Jira lowers the work item’s remaining estimate by the time logged, never below 0. Work items created through the API, imported from a CSV, or logged with “leave estimate unchanged” can end up with an original estimate and logged time but no remaining estimate at all. Counting that as 0 left would show the work as finished: an epic with 470h logged against 594h planned would read 100%.

So where a work item has an original estimate and no remaining estimate, Sumline uses what Jira would have kept: the original estimate minus the time logged on that work item, never below 0. A done work item gets 0. Each work item is worked out on its own, so one work item running over does not hide another’s remaining work. That epic reads 79%, with 123h 30m remaining. This happens only while Original estimate is one of the fields rolled up; without it, remaining is taken as Jira has it, empty included.

A worked-out remaining estimate takes part in “Each estimate counted once” like one entered in Jira. With the children winning, children whose remaining was worked out set aside a remaining estimate someone entered on their parent. Time logged on child work items does not lower the parent’s remaining estimate, the same as in Jira.

Sumline says where it did this:

  • the tables show “worked out” under the Remaining figure, and its tooltip names the work items;
  • the Sumline panel names them in a grey line under its figures, as “Remaining worked out: SLC-62 and SLC-69”, and the Remaining figure says “(worked out)”;
  • the Excel and CSV downloads add them to the row’s Exclusion note.

A work item whose remaining estimate is 0 in Jira keeps its 0. A worked-out value never makes a work item count as estimated: that still comes from its original estimate.

Work items with no estimate

A work item with no value in any estimate rolled up (story points, original estimate, remaining estimate, or a number field of your own) adds nothing to the estimate totals, and it is never counted as 0. Sumline names it instead:

  • the panel’s “Check these work items” box lists the unestimated work items, a sub-task with the row it sits under (“WEB-14 (under WEB-12)”), and its child table marks such a child “No estimate”, or “N unestimated” when only some of the child’s own subtasks are estimated, each naming the work items on hover;
  • the Details column names them in one small amber line, “No estimate:” and their keys;
  • the tables and the gadget show “N unestimated” in the Warnings column.

A work item whose parent’s estimate is the one counted is covered by it, so it is not called unestimated. Time logged on an unestimated work item still adds to Time spent.

A “Check these work items” box: one work item has no estimate, WEB-15. Its key is a link.
The panel’s “Check these work items” box: the work item it names is a link

% complete

Always two numbers, side by side:

  • By estimate: story points done ÷ story points total. When no story points are counted, time is used instead: time spent ÷ (time spent + remaining estimate), and only while a remaining estimate is known. Time logged with no remaining estimate to set it against says nothing about what is left, so it reads “—”, not 100%.
  • By work item count: done work items ÷ all work items, across every level below the parent.

A parent with nothing estimated below it shows “—” with no bar, never 100%.

When story points are the basis, work estimated only in hours cannot add to it: a done 3-point story beside a 40-hour story not started would read 100%. So Sumline names each branch that carries an original or remaining estimate and no story points anywhere in it: on the tables, hovering the percentage names them, and so do a line in the panel’s “Check the estimates” and the row’s Exclusion note in the downloads. % by work item count and the time columns still include them. Whichever basis a table shows, its percentage names the work it leaves out the same way: on Time, the branches with story points and no time. Sub-tasks with hours under a story with points are covered by the story’s points and are not named. On the tables, “% complete by” chooses the basis for every row: Auto (story points where a row has them, otherwise time), Story points, Original estimate, Time (time logged ÷ (time logged + time remaining)), or one of the number fields an admin added. The line under the control says what the chosen one measures.

The rows of a table

In short, a table counts the epics and initiatives that are open or were done in the last 90 days, and every work item under them, closed ones included.

A table has one row per epic, and per initiative where your site has that level, so an epic appears as its own row and again inside its initiative’s row. The Total line, on the tables and on the gadget, counts such an epic once, in its initiative (see Group by). An admin can limit the rows to initiatives; see Settings.

The rows come in Rank order, the order of the backlog, the board and the timeline: “Rank (backlog order)” in Sort by. A site without Jira Software has no Rank field, so its rows come oldest first. Sort by orders the rows by any column instead, ascending or descending.

Turn on Rank under Columns to keep each row’s place in the backlog on screen, whatever the sort. The number counts every row of the table, so a filter leaves it as it was. A work item under an opened row shows its place among its parent’s work items, as the board lists them; one a link rule brought under its parent has none (“—”). A site without Jira Software has no Rank column.

Every row with work items under it starts with a chevron. Opening it shows the work items under the row, indented, each with its own figures and warnings, in the order Sort by gives the rows: in Rank order, as the backlog lists them, until you choose another. Subtasks follow the same order under their story. They are worked out as part of that row, so each estimate is still counted once: under “parent wins”, a story whose epic’s estimate is counted has its own estimate left out there too. While the table is under five minutes old, opening a row reads nothing more from Jira.

Rows are the epics and initiatives that are open, or whose status category became Done in the last 90 days. One done longer ago is not a row, and the line above the table says so: “parents done more than 90 days ago are not shown”. It is still counted inside any row above it, such as an open initiative, and the Sumline panel on the epic itself still rolls it up. Inside a row, every work item counts, however long ago it was closed.

The Status filter matches the row’s own status category and the Assignee filter the row’s own assignee. The Sprint filter keeps a row when any work item in it is in the chosen sprint; “Backlog” keeps the rows with no work item in any sprint. The Sprint filter is not shown in a space where no work item is in a sprint, such as a service desk. The filters apply to the table’s own rows only: the work items under an opened row stay with it, sorted as the rows are.

Group by

Group by, beside Sort by, gathers the table’s rows by Assignee, Priority, Status, Type, Space (on the page across spaces), Sprint, Fix version (both below) or Team: Jira’s Team field on each row’s own work item, by the team’s name. A grouping that does not apply to the rows on screen is not in the menu: Type with one work type, Space with one space, Sprint with no work in a sprint, Fix version with none planned for a release, and Team on a site with no Team field or no team on any row. A shaded line over each group adds up its rows: points, each number field, original estimate, logged and remaining time and work items. Both percentages are worked out again from those sums, not averaged: a group of 5 of 8 points and 0 of 8 reads 31%, 5 of 16.

The groups come in their field’s own order: your site’s priorities from the highest, To Do before In Progress before Done, people alphabetically, initiatives before epics, spaces by key, teams by name. Nobody assigned, no priority and no team come last. The rows inside a group keep the Sort by order, and the filters apply first, so a group with no row left is not shown.

Each piece of work is counted once in a group. An epic is a row of its own and also part of its initiative’s row, so when both are in the same group the line counts the initiative and leaves the epic out; hover the group’s name to see which rows it counted once. When the initiative is in one group and the epic in another, say each with a different assignee, the epic’s work is in both lines, and hovering the initiative’s group says so. Lines from different groups are not meant to be added together.

By sprint and by fix version, a row stands under each group its work is in, with only that work’s figures: an epic with stories in two sprints is under both. Each work item counts in its current sprint, the active one, else the next one planned, else the last it was in, as Jira’s sprint report counts a work item carried over; so the sprint lines add up to the table, work items included: a story whose sub-tasks carry all its time still counts as a work item in its own sprint. A sub-task counts in its story’s sprint and release. A work item planned for two releases counts in both, and hovering either release’s name says so. Work in no sprint, or in no release, is the last group.

The Total line, the table’s first, adds up every row on screen by the same rule: an epic inside an initiative that is also on screen is counted once, in the initiative. Filter the initiative out and the epic counts on its own.

On the page across spaces, the address then holds the spaces and the group, such as ?spaces=APP,PAY,WEB&group=assignee; send it and it opens the same grouped table.

Delivery warnings

Besides the warnings about estimates, the Warnings column names what may hold the work up, for the row and everything under it:

  • Overdue: not done, and past its due date, as the day is where you are: a work item due today is overdue from tomorrow.
  • Over estimate: logged plus remaining estimate is more than the original estimate; the panel names each work item with where its work ends as a percent of the estimate (never 100% once over), such as “ACME-2 133%”. An original estimate of 0 counts as no estimate. Jira lowers the remaining estimate by what is logged, so on a work item nobody touched this means more time logged than estimated. It fires earlier where someone raised the remaining estimate by hand, and a done work item’s remaining counts as 0, as in the Remaining figure.
  • Closed early: done, with a work item under it still open.
  • Ready to close: still open, with every work item under it done.

Closed early and Ready to close read every work item under it, as Jira has them, even with Done only on: done only changes what is counted, not whether work is open.

Hover a warning to see its work items. The work item panel lists them first in its Check these work items box, and the downloads in a Delivery warnings column. None of them changes a figure. A Jira admin can switch each off, and switch on three more: no due date, unassigned, and time logged with no original estimate (Settings).

Without Jira Premium, Jira has nothing above epics. Many teams make a “Theme” or an “Initiative” of an ordinary work type and link its epics to it. A Jira admin can tell Sumline so (Settings): work items of that work type then roll up the work items their links point to, and everything under those, on the work item, in the tables and on the dashboard. Each work item is counted under one parent only: when two link it, it counts under the first, and the other names it (“counted elsewhere”), so nothing is counted twice. The downloads hold the group lines too; see Download, then Excel or CSV.

The size limit

A table, the Sumline tab, Sumline under Apps or the gadget, rolls up at most 2,000 work items, and the count is everything rolled up: the table’s rows plus every work item under them, children in other spaces included. Closed history does not count: an epic done more than 90 days ago is not a row, so it is fetched only when it sits under a row (an open initiative, say), and a work item under no epic is never fetched. A space with 30 open epics and 3,000 closed work items is well under the limit; 40 open epics holding 2,100 stories and subtasks are over it.

Above it Sumline says so, “Too large to roll up at once”, and rolls up each epic on its own instead, in the same table: every epic with up to 2,000 work items under it shows its real figures, and one with more reads “Over 2,000 work items” with a link to its work item, where the Sumline panel rolls it up: the panel has no work item limit. The Total line adds up only the epics rolled up and says how many (“Total of 5 of 8 rows”), never a partial number for the whole space. Roll up a narrower set opens Sumline under Apps with the spaces or filter as JQL, ready for you to narrow, by sprint or by epic, say, before you press Roll up.

How fresh the numbers are

Sumline rolls up when you open a view and keeps the figures for five minutes, per person; the next open in that time reuses it. Every view says how old its numbers are (“Updated just now”, “Updated 3 min ago”) and has a Recalculate button that reads Jira again straight away. The gadget has none of its own: Jira’s reload icon on the gadget, or the dashboard’s Refresh, recalculates it. When an admin saves the settings, every view rolls up again on its next open.