How to plan an RFID rollout in a warehouse

Warehouse RFID rarely fails because the radio does not work. It stalls because nobody agreed what question it should answer, the pilot was run on easy products, stray reads at the dock doors were never solved, or the reads never made it into the systems people use. The plan below is the order that avoids those traps.

1. 1
   1. Pick the question and the number

   Receiving accuracy? Shipping errors? Time to find a pallet? Cycle-count labour? Write down today’s figure; that is how the pilot will be judged.

2. 2
   2. Map the flow and the read points

   On the floor plan, mark every place a read should happen: dock doors, staging lanes, conveyors, pack-out, cages, exits. Each becomes a named location in the data.

3. 3
   3. Choose the tagging level

   Pallet (SSCC), case or item (SGTIN)? Who applies the tag — supplier, receiving, or a print-and-apply station? Tags applied at source are cheapest per tag but need supplier compliance.

4. 4
   4. Test tags on your worst products

   Metal, liquids, foil packaging, dense pallets. Test several tags on real products at real distances. See tags for metal, liquids and apparel.

5. 5
   5. Survey the site

   Metal racking, dock levellers, other readers, Wi-Fi and the country’s UHF band. Note where power and network can reach each read point.

6. 6
   6. Pilot one door or one zone

   Measure read rate with known pallets at real forklift speeds, and count stray reads from the neighbouring doors. Tune antennas, power and filtering until both are acceptable.

7. 7
   7. Fix the data model

   Location codes that match the WMS, EPC encoding scheme, which events middleware emits (arrive, depart, direction), and how often they are sent.

8. 8
   8. Integrate and show it

   Post events to the WMS for transactions and to a live map for people. Train supervisors on what the map shows and what it does not.

9. 9
   9. Scale and maintain

   Roll out door by door. Monitor every reader’s last-read time so a dead portal is noticed in hours, not at the next audit.

• Piloting on easy products. Cardboard boxes of T-shirts read perfectly; the motor oil aisle will not.
• Ignoring stray reads. At dock doors, reading the neighbouring door is as bad as missing a pallet.
• Raw reads into the WMS. Send events (arrived, departed, at lane 3), not every read; filtering is middleware’s job.
• Inventing location names. If the dock door is D07 in the WMS, it is D07 everywhere.
• No owner for reader health. Readers get unplugged, cables get crushed, firmware drifts.

Read-rate diagnostics have their own guide: RFID read rate: why tags get missed.

Step 2 is a map, and it should stay one. In Spatial Map the read points you sketch in discovery become shapes on a 3D floor plan; the pilot’s reads light them up; the rollout adds doors one at a time; and reader health is a colour, not a log query. The RFID warehouse management page shows the finished picture.

Dock doors and inbound lanes on a live map

Hardware choices are on RFID hardware (e.g. readers from Zebra, Chainway, SATO or Sanray — trademarks of their owners). Middleware, WMS integration or a pilot run end to end by our team is custom development.