Metrics reference¶

Also in the app

Press ? in Fab Dispatch to read this in the help center, or / to ask the assistant.

Every view uses the same definitions. ↑ means higher is better, ↓ lower is better.

Coverage¶

Jobs covered (↑, %): share of known jobs assigned to someone.
Bottleneck downs covered (↑, %): share of priority-3 tool-downs (tool-downs on the bottleneck family) assigned. Shown as n/a while the shift has no priority-3 work, for example early in a live shift before any bottleneck down is reported: there is nothing to cover yet, so it is not 100%. The count of priority-3 jobs is reported alongside (critical_jobs).
Priority-weighted coverage (↑, %): coverage where each job counts by its priority (3, 2 or 1).

Response to tool-downs (↓, min): mean time from a tool-down's window opening to an engineer starting work on it, over the tool-downs that are served. A tool-down nobody takes is not averaged in; it is counted under Unserved tool-downs.
Response, p95 (↓, min): the slowest-but-one-in-twenty response among served tool-downs (nearest rank, so it is an actual job's response, not an interpolation). The mean hides a long tail; this shows it.
Unserved tool-downs (↓, count): known tool-downs (any priority) that no engineer takes.
Walking per job (↓, m): metres walked per job served.
Idle wait, all engineers (↓, min): total minutes engineers wait for a job's window to open.
Over-qualification (↓, levels): total certification levels above what jobs needed.
Engineer utilisation (↑, %): hands-on work as a share of total shift time.
Workload spread (↓, min): standard deviation of assigned service minutes per engineer, over every engineer on shift including those with no jobs. Lower is more even. This is what the UI shows (workload_minutes_std). The older jobs-based spread (workload_std, standard deviation of job counts over the same engineers) is still in the API for compatibility, but a 35-minute probe-card clean and a 240-minute chamber wet clean count the same there, so prefer minutes.

Operating cost (↓, pts): the objective every strategy minimises: walking, idle wait, over-qualification and workload costs, plus the priority reward of any unserved work. See Constraints and cost weights.
Solve time (↓, ms): how long the strategy took. Repeat plans come from cache.

▲ / ▼ vs Greedy shows each strategy's difference from the Greedy baseline, coloured by whether it's better or worse.