Define the business event before the reader
“We want RFID” is not a requirement. “We must know whether every expected carton crossed dock door 3 before truck release” is a requirement. The latter tells you where antennas belong, which items need tags and what the software must reconcile.
Test the ugly physical cases
RF performance changes with material, tag orientation, metal, liquids, stacking and reader placement. Your pilot set should deliberately include the cases most likely to fail rather than a clean demonstration pallet.
- Different carton orientations
- Dense pallet or tote packing
- Metal-adjacent products or assets
- Fast and slow movement through the zone
- People and forklifts near the read area
- Adjacent-zone tags that must not become false events
Measure more than read rate
A 99% raw read rate can still create a bad operational system if the 1% misses the wrong item or if adjacent-zone reads create false dispatches. Measure item-level recall, false positives, duplicate event handling and whether the operational decision is correct.
Design reconciliation before rollout
The useful software knows what EPCs represent, which shipment was expected, when a read becomes a business event, how long the read window stays open and what an operator sees when expected and observed inventory disagree.
Pilot exit criteria
Write the go / no-go thresholds before the test: acceptable business-event accuracy, maximum manual interventions, expected time saved per movement, tag cost, hardware cost and integration work. A pilot that cannot fail is a demo, not due diligence.