First-party measurement method
Benchmark the site against its own accepted baseline
This protocol defines how Fortune AI and an operations team can create a reviewable warehouse baseline from operating records and existing camera views. It is a measurement framework—not a customer case study, industry average, or performance guarantee.
Warehouse operations baseline worksheet
Record the scope, sample, exclusions, baseline, target, and acceptance decision for each metric.
Download CSV worksheetMetric definitions
Three bounded warehouse measures
| Measure | Start and end event | System record | Camera evidence | Reported result |
|---|---|---|---|---|
| Dock-to-stock | Unload begins to accepted putaway or receiving close | TMS arrival, ASN, receipt, location scan | Dock, staging, and agreed putaway handoff views | Median, 90th percentile, and exception clips by shift |
| Forklift utilization | Agreed shift window; classify active handling, loaded travel, empty travel, and waiting | Shift roster; optional telematics or task events | Only the zones covered by approved camera views | Minutes and share of observed time by activity state |
| Inventory handoff verification | Expected shipment or putaway event compared with observed physical handoff | Item, dock, vehicle, location, and event timestamp from WMS | Timestamp-aligned pallet, vehicle, dock, or location evidence | Match, mismatch, or uncertain—with footage and resolution state |
Evaluation sample
Agree on the test before reviewing the result
- Freeze the definition. Document eligible events, start and end states, zones, shifts, and exclusions.
- Select representative footage. Include normal volume, peak periods, changeovers, occlusion, and at least one known exception when available.
- Build the acceptance set. The worksheet starts with 100 eligible events across at least three representative shifts when volume permits. Lower-volume sites should review all available events and report the count.
- Compare independently. A team reviewer labels the acceptance set before comparing it with the camera-derived measure.
- Decide explicitly. Publish the event count, coverage, disagreements, uncertain cases, and the pre-agreed acceptance threshold.
WMS / TMS + camera
Join records only at an agreed operating event
Minimum system fields
- Stable event or shipment identifier
- Event type and timestamp with timezone
- Dock, lane, vehicle, or location identifier
- Expected state and later resolution state
Camera-side record
- Camera identifier and covered zone
- Observed start and end timestamp
- Measurement state and uncertainty reason
- Supporting clip reference and retention status
Operational output
- Matched event and evidence link
- Duration or activity classification
- Exception owner and review status
- API, MCP, report, or WMS/TMS write-back result
Reporting limits
State what the measure cannot see
- Camera-derived results cover approved views and time windows, not the entire facility by default.
- Occlusion, lighting, camera movement, timestamp drift, and incomplete source records can produce uncertain or excluded events.
- Forklift utilization is an activity measure within covered zones; it is not worker identity, productivity scoring, or continuous equipment telematics.
- Inventory handoff verification checks bounded state changes. It does not replace the WMS or claim continuous item-level tracking.
- Any public case study or performance benchmark must use verified customer permission, sample details, and measured results. None are implied here.