Dataset Methodology
Methodology
| Metric | Rule |
|---|---|
| Completion rate | Confirmed tasks divided by total tasks |
| Short pick rate | Tasks where picked quantity is below requested quantity |
| Exact-fill accuracy | Tasks where picked quantity equals requested quantity |
| Over-pick rate | Tasks where fill rate is above 1.0; shown separately from quality until classified |
| Median execution time | Median start-to-confirmation minutes |
| Median cycle time | Median creation-to-confirmation minutes |
| Backlog units in limbo | Open tasks multiplied by average units per task |
Interpretation Rules
| Decision | Reason |
|---|---|
| Prefer median over mean for cycle time | Long-tail open tasks distort average performance |
| Compare cold and ambient separately | Queue physics and labour conditions are materially different |
| Treat active tasks as operational risk, not completed work | They represent unresolved inventory and service exposure |
| Rank labour by both productivity and concentration | The top performer can also be a single-point-of-failure |
Gaps Still Requiring Joins
| Gap | Needed join |
|---|---|
| Revenue or customer impact of short picks | Order and shipment data |
| Product naming and category analysis | Product master |
| Replenishment vs true pick exception | Task classification or movement type |
| Labour cost per task | Payroll or labour standard tables |
| On-time fulfilment | Delivery schedule or carrier milestone data |