Delivery Overview

Every status the Jira workflow configures, in the order work walks them, for user stories only — Task and Technical Task run a workflow with no test gates, and mixing them made 88% of the apparent skips artefacts of the mix. A CIRCLE IS HOW MANY STORIES SIT THERE NOW. Filled means work is there; a tinted ring means work has been there and moved on; a dashed grey ring means no story has ever entered. That is why the release run reads zero and still shows traffic — nothing sits past QA test passed today, and stories have walked every status through to production. The chip beside each name is that status's own code, unique across the workflow so it identifies exactly one status wherever it is quoted, and coloured by its lifecycle phase so the eight groups still read at a glance. The legend below the diagram spells out all 28. A LINE TAKES THE COLOUR OF THE STATUS IT LEAVES, so a return is traceable to where it came from without following it. The main path is one constant width and the widest thing here; everything off it is thinner and weighted by how many moves took it. A dashed line with no motion is a route the workflow configures that no story has ever taken. Everything that fails, parks or reverts returns to In Triage or To Do, which is why those share one line back rather than six. No duration is shown anywhere on this diagram. 56% of status transitions in this project are zero-elapsed back-fill, so any median would measure when a status was typed rather than when work moved.

Reliability is of completed items carrying a due date, so its coverage sits beside it: the squads run from 78% to 96%, and the same percentage over different denominators is not the same claim. Blocked counts unfinished items on the receiving end of a Blocks or Depends link; Relates is association, not constraint.

Start plus added minus removed equals final, and it holds because added and removed are defined from the two boundary states rather than from raw join and leave events. An item that left and came back is committed scope that churned, not new scope. A build where they do not add up fails rather than renders.

Not the Sprint field. Stories are bulk-moved onto the active sprint, so the field would credit old completions to whichever sprint is running now.

An epic at 20% with four children is a rounding error. Furthest phase reached is computed from the children IN SCOPE, so narrowing to a release changes it. Childless epics are excluded: a list about where the work is cannot be padded with epics carrying none.

In the sprint’s final scope, not finished when it ended, and in a later sprint. Not “in more than one sprint” — an item can sit in two because someone corrected a field. Measured on completed sprints only: in a running sprint nothing has moved on yet and the figure reads near zero, which says carry-over is solved.

Of completed items carrying a due date, the share that finished on or before it; items completed without one are excluded from both halves rather than counted as met. The headline is every completed item to date and the line is each sprint’s own completions, so the endpoint and the figure above it are different measurements.

A move between two Development statuses is work continuing, not rework. The denominator is items that ever reached Development, so leaving work in Definition cannot improve the figure.

Median days in phase is not shown. 56% of status transitions in this project are zero-elapsed back-fill, so the figure would measure when a status was typed rather than when work moved.

Blocked means an unfinished item on the receiving end of a Blocks or Depends link; Relates is narrative association and is excluded. One item held by four tickets is one thing to unblock — an earlier draft counted edges and overstated it by half. A squad recording no links looks unblocked, which is why coverage is a column rather than a note.

The figures in the sections above cover the wider population, tasks included. Narrowing this register is display, never derivation: no figure above recomputes, and the count states how many of the whole population are showing so a filtered view cannot be mistaken for the programme.