Anatomy of a ticket
Anatomy of a ticket β GAM
Window: 30d Β· All 5 Β· AI 0 Β· Human 5
AI vs Human comparison
Bars proportional within each row (max = 100%). Ξ% lower-is-better β green = AI faster.
Not measurable (one cohort has zero samples):
- Β·Jira Flow Lead Time (raw / wall-clock) β no AI cohort samples (no completed tickets with an AI assignee, or no PRs by AI authors)
- Β·Lead Time (active, Γ· pauses) β no AI cohort samples (no completed tickets with an AI assignee, or no PRs by AI authors)
- Β·Time to Ready β no AI cohort samples (no completed tickets with an AI assignee, or no PRs by AI authors)
- Β·Cycle Time (raw / wall-clock) β no AI cohort samples (no completed tickets with an AI assignee, or no PRs by AI authors)
- Β·Cycle Time (active, Γ· pauses) β no AI cohort samples (no completed tickets with an AI assignee, or no PRs by AI authors)
- Β·Time to First PR β no tickets / PRs measurable in this window
- Β·PR Review Duration β no tickets / PRs measurable in this window
- Β·Lead Time for Changes β no tickets / PRs measurable in this window
Cohort breakdown
β marker when AI or Human cohort sample size is below 10.
| Metric | All p50 | n | AI p50 | n | Human p50 | n | Ξ (AI vs Human) |
|---|---|---|---|---|---|---|---|
| Jira Flow Lead Time (raw / wall-clock) | 12.2d | 5 | β | 0 β | 12.2d | 5 β | β |
| Lead Time (active, Γ· pauses) | 197h | 5 | β | 0 β | 197h | 5 β | β |
| Time to Ready | 10.1d | 5 | β | 0 β | 10.1d | 5 β | β |
| Cycle Time (raw / wall-clock) | 49h | 5 | β | 0 β | 49h | 5 β | β |
| Cycle Time (active, Γ· pauses) | 49h | 5 | β | 0 β | 49h | 5 β | β |
| Time to First PR | β | 0 | β | 0 β | β | 0 β | β |
| PR Review Duration | β | 0 | β | 0 β | β | 0 β | β |
| Lead Time for Changes | β | 0 | β | 0 β | β | 0 β | β |
All = every measurable ticket / PR in the window. AI = agent-owned cohort (ticket assignee or PR author is an active agent at completion). H = human cohort (everyone else, including unassigned). Ξ compares AI vs Human p50 β negative = AI faster.
Side metrics
π§ AI deep analysis
Generate an LLM-driven explanation of WHY the AI cohort is faster / slower on each lane. Cites specific tickets from the evidence pack, suggests next steps.
Active variants use business hours minus pause statuses (same formula as the per-project Cycle / Lead pages). Raw variants are pure wall-clock. AI cohort = ticket assignee (or PR author) is an active agent at completion time. Sample-size suppression marks values directional only when AI or Human cohort drops below 10 β the value is still shown for context but trust it less.