GenFrontier homeScope a project
Menu
Reader questionFor which warehouse processes and SKUs does RFID justify its cost over barcodes, and what do retailer mandates change?

RFID or barcodes in the warehouse: where RFID pays back

RFID vs barcode, process by process: where RFID pays back in a warehouse, how to estimate payback with your own numbers, and what retailer mandates change.

field-notes / traceability-standards / rfid-vs-barcode-warehouse
A wooden-framed sieve tilted above a shallow stream, holding dozens of pebbles caught in one pass, with one pebble printed red.

RFID pays back over barcodes where you must identify many items without handling each one: counting tagged units inside closed cartons, verifying the contents of packed cartons at full line speed, and tracking returnable assets that come back again and again. Barcodes stay cheaper wherever a person handles each item anyway, where a tag would cost a large share of the item’s margin, and where the real problem is process discipline rather than data capture. A retailer tagging mandate changes the arithmetic, because the tag cost is spent either way; the question becomes whether you also use those tags inside your own four walls.

This field note works process by process, sets out the GS1 standards you need before buying tags, gives a payback method you can run with your own numbers, and lists what to confirm with a retail buyer before acting on mandate claims you read online. The studies cited are university field research, mostly from apparel, and each is dated so you can judge how far it carries to your operation.

What actually differs between a barcode scan and an RFID read

A barcode scan identifies one item, chosen by a person, with line of sight. A passive UHF RFID read identifies every tag in range at once, without line of sight, and each tag carries a serialized identity. Almost every cost and benefit difference flows from those two facts.

GS1 describes the tags most used in supply chains this way: “The most broadly implemented tags in our industries are UHF passive tags, also known as RAIN RFID tags.” It adds that “radio waves can be used to capture the unique identifiers at extremely high rates and at distances well in excess of 10 metres, without line-of-sight contact” (GS1, RFID standards, checked November 2026).

Characteristic Barcode (GS1-128, EAN/UPC, 2D) Passive UHF RFID (RAIN)
Line of sight Required Not required
Items per action One, chosen by the operator Every tag in the read field
Identity level Usually the product (GTIN); serial only if printed Always an individual item: a GTIN plus a serial (SGTIN)
Data carrier cost Printed label or packaging print A tag per item, bought and encoded every time
Infrastructure Handhelds and printers you may already own Encoders, readers, antennas, portals or tunnels, software
Typical failure No read, or a misread you notice at once Missed tags, stray tags and reads from the wrong carton, often silent
Proof that an item was handled Yes, the scan is the touch No, a read only proves the tag was in range

The last row is easy to overlook. A barcode scan doubles as a confirmation step in a process: the picker touched this item. An RFID read confirms presence, not intent.

RFID is strongest where presence is the question, such as “what is in this carton?” or “what is on this rack?”, and weakest where the process needs a deliberate human confirmation.

Every RFID tag is an instance, not a class

The GS1 General Specifications state that “An EPC/RFID data carrier can only encode globally unique instance level identifiers.” When a tag supplements a barcode that carries a GTIN, the serialized GTIN (SGTIN) “SHALL be used” (section 4.15.2, release 26.0). In practice, every tagged unit needs a serial number that nobody else has used for that GTIN, so serial assignment becomes a system you have to run.

That has consequences beyond the tag. Your encoding station must draw serials from a controlled range, your WMS or a middleware layer must store or at least tolerate them, and a reprinted or replaced tag must not reuse a serial. Barcodes let you avoid all of that by identifying the product, not the unit.

What GS1 EPC standards you need to know before buying tags

The data written on a tag is defined by GS1’s EPC Tag Data Standard (TDS, release 2.3, ratified October 2025), which “defines the Electronic Product Code and specifies the memory contents of Gen 2 RFID Tags.” How readers and tags talk is a separate GS1 standard, the EPC UHF Gen2 air interface (release 3.0.1, ratified February 2026). For warehouse work, five TDS points matter most.

  1. Pick the EPC scheme that matches the object. Each GS1 key has an EPC form. The table below lists the ones a warehouse meets.
  2. Know the serial limits of your encoding. For SGTIN-96, the TDS limits the serial to “Numeric-only, no leading zeros, decimal value must be less than” 2 to the power 38, which is “less than or equal to 274,877,906,943.” SGTIN-198 allows “All values permitted by GS1 General Specifications (up to 20 alphanumeric characters).”
  3. Match the barcode serial exactly. If the item’s barcode also carries a serial (AI 21), the General Specifications require the same serial on the tag, “taking care not to change any characters and neither add nor remove any leading zeros.” A serial with leading zeros or letters therefore cannot go into SGTIN-96.
  4. Do not use the filter value as business data. The filter value (for example 1 for a point-of-sale trade item, 2 for a full case for transport, 6 for a unit load) helps readers select tags. The TDS warns it “is not part of the Electronic Product Code” and “is not intended to provide a reliable packaging-level indicator for business applications.”
  5. Dates can go on the tag. TDS 2.0 introduced DSGTIN+, “a date-prioritised serialised GTIN” that GS1 expects to be “particularly useful for perishable goods, stock rotation and management of goods with limited remaining shelf life.”
Warehouse object GS1 key EPC scheme Typical use
Sellable unit or case GTIN + serial SGTIN (SGTIN-96, SGTIN-198, SGTIN+) Item-level counts, pack verification
Unit with a critical date GTIN + serial + date DSGTIN+ Selecting short-dated stock
Pallet or other logistic unit SSCC SSCC EPC Pallet reads at dock doors
Returnable tote, pallet or container GRAI GRAI EPC Asset tracking and pool reconciliation
Fixed or individual asset GIAI GIAI EPC Equipment, carts, racking assets
Location GLN with extension SGLN EPC Reference tags at doors or zones

Source: GS1 EPC Tag Data Standard 2.3, sections 7 and 10, checked November 2026. The “typical use” column is our summary.

Pallet-level RFID does not replace the GS1-128 label. The General Specifications make GS1-128 the mandatory carrier for the SSCC on a logistic unit, so a pallet tag is an addition to the label covered in our note on SSCC pallet labels.

Where RFID earns its cost, process by process

RFID earns its cost in processes where items are counted or verified in bulk without being handled one by one. It adds little where each item is touched anyway. The table gives our assessment for common warehouse processes, with the evidence behind each row.

Process What RFID changes Barcodes are usually enough when Main cost driver Evidence
Receiving tagged units in cartons Units counted inside closed cartons and compared to the ship notice Cartons are received by SSCC and contents are trusted Tags applied at source; dock portal or tunnel ITRI WP159 (2012)
Cycle counting units A bay or fixture of tagged units read in minutes Locations hold full cases or pallets, or few units Handheld readers; complete tag coverage ITRI WP146 (2009)
Picking Little: the picker handles the item anyway Almost always Not applicable Our judgement
Pack-out verification Every unit in every closed carton checked on a conveyor Low volume, full-case orders or low chargeback exposure Conveyor read points ITRI WP159 (2012)
Dock door and loading Pallet or case reads as loads pass Loads are scanned by SSCC at the door Portals; preventing reads from neighboring doors Our judgement
Returnable assets One durable tag read on every trip Assets are cheap or rarely come back Durable tags; read points where assets pass GS1 RTI guideline; our judgement
Customer returns The exact unit identified by serial Returns are few or low value Tags must survive the customer Our judgement

Receiving and pack-out verification

The strongest warehouse evidence comes from distribution centers, not stores. In a study funded by GS1 US and the American Apparel and Footwear Association, University of Arkansas researchers visited 17 apparel supplier facilities in the US (ITRI WP159, issued January 2012). Their central finding was that “for the supplier, inventory accuracy is high but they need to find a more efficient way of keeping inventory accuracy high.”

At one supplier, the researchers analyzed two conveyor read points, each a single reader with three antennas, carrying cartons of up to 72 tagged items at up to about 400 feet per minute. Each read point “was able to perform item-level quantity audits in excess of 99.99% of all cartons.” Compared with manual audits on the same lines, “The RFID system identified up to an additional 4.8% pick/pack inaccuracy” that would not otherwise have been found.

Read those numbers as what they are: apparel, tags applied at manufacture, and a 2011 study. The paper itself notes that passive Gen 2 UHF tags “have been found to work exceedingly well with apparel items.” The method carries over to other products; the read rates need to be proven on your own items.

Cycle counting

Counting is where RFID’s speed shows most plainly. In the Dillard’s study (ITRI WP146, 2009), researchers timed RFID and barcode cycle counts of the same items, on the same day, in the same store. For a mean of 1,500 items, the RFID count took 5.5 minutes and the barcode count took 2 hours and 18.5 minutes, which the paper reports as “a 96% labor reduction.”

Faster counting only matters if counting is your constraint. A separate field experiment (ITRI WP154, 2010) found “a significant decrease in IRI of approximately 26%,” where IRI is inventory record inaccuracy, in stores using RFID-enabled automatic adjustment with case-level tags. The authors add that “the effectiveness of RFID on reducing IRI varies by category.”

If your warehouse counts full pallets by license plate, the barcode count is already fast, and RFID has little to save. Our note on cycle counting versus a physical inventory covers what a counting program must prove before auditors accept it.

Picking

For most picking, a barcode scan is the right tool, because the picker handles the item and the scan confirms it. RFID can confirm a whole tote at the end of a pick path, but that is pack verification moved earlier. Treat RFID picking claims with care unless they describe exactly which handling step disappears.

Dock doors and loading

A portal at a dock door can confirm which tagged pallets or cases left on which trailer without anyone scanning a label. The difficulty is not reading tags; it is reading only the right ones. A reader at door 4 can see tags staged at door 5, or a pallet that was driven past the door and back again.

That makes dock door RFID a software and layout problem as much as a hardware one. Define each read zone, decide what sequence of reads counts as “loaded” (direction, time window, the open load at that door), and test with neighboring doors full. Where loads are already scanned by SSCC at the door, compare the portal against that scan for several weeks before letting it replace anything.

Returnable assets

Totes, pallets, roll cages and containers are where a tag’s cost is spread across many trips. GS1 identifies returnable assets with the GRAI. Its pallet tagging guideline notes that, “Depending on the placement of the tag, the load can have a significant influence in the reading performance of the RFID tag,” so tag placement on assets has to be tested with real loads (GS1 RTI (Pallet Tagging) Guideline, Issue 2, 2010).

Where RFID tends to fit: handling pattern against the value of item-level identity RFID strongest Carton content checks Unit cycle counts Returnable assets Tagged receipts Barcode plus serial scan Serial capture at pack Returns by unit RFID mainly if a customer requires tags Pallet and case level GS1-128 SSCC labels Pallet tags optional at busy dock doors Barcodes Picking Put-away to location Low-value each scans Items handled one at a time? No → Yes Value of unit identity: low → high
Decision framework by the GenFrontier Editorial Team, not the result of a study. Placement depends on your products, volumes and customer requirements.

Where RFID projects lose money

In Auburn University’s audits, the gap between RFID counts and perfect accuracy came mainly from what its authors call “execution failures”: tags that were never applied, tags that perform poorly on the product, and tags read where they should not be. The RFID Lab ranked the error sources it found in apparel audits, from most to least common (Key Considerations for RFID Pilots and Deployments, 2014):

  1. No tag on a product that should have one, about 2.3% in their audits.
  2. Poorly performing tags, about 0.3% with performance testing and compliance monitoring, but “30% or more” where neither is done.
  3. Double-tagged items, about 0.5%.
  4. Dead tags, about 0.35%.
  5. Incorrectly encoded tags, about 0.3%, mostly attempts “to match the UPC with the correct EPC, but getting it wrong.”
  6. Stray tags, about 0.1%: tags separated from their products that keep reporting themselves.

The same paper says “a more realistic inventory accuracy number when using RFID is 95%” in apparel. It also reports a warehouse-specific warning: “we have observed DC tagging execution failures in every pilot where tags are meant to be applied at the DC.” DC associates were often unsure which cartons to open and tag, and items assumed tagged arrived untagged.

Example (fictional). A fictional distributor pilots RFID pack-out verification and, to save time, tags 30 SKUs at its own receiving dock instead of waiting for suppliers. Associates are measured on cartons per hour, so some cartons pass the tagging station unopened. Four weeks in, the pack-out tunnel flags many cartons as short, but most of the “missing” units are untagged, not missing. The pilot is paused, tagging moves to two suppliers who can tag at manufacture, and receiving adds a tag-presence check on a sample of cartons.

Two lessons follow for a warehouse. First, tag at the source wherever you can, and audit tag presence at receiving. Second, test the tagged item, not the bare tag.

GS1’s Tagged-Item Performance Protocol exists for this: it provides “a standard test protocol to verify the performance of a tagged item (not a tag only)” between trading partners.

What RFID costs: the drivers to price

RFID cost has a recurring part that scales with volume and a one-time part that scales with the number of read points. Price both before comparing anything with barcodes. We give no industry price figures here, because tag and reader prices vary with volume, inlay type, form factor and date; get current quotes for your own items.

Cost element Recurring or one-time What drives it Questions to ask
Tags or inlays Recurring, per unit Volume, inlay type, label or hang tag format, special tags for difficult products Who buys them: you, the supplier, or the customer’s program?
Encoding and application Recurring Tagging at source, at your DC or at the store; print-and-encode equipment Can the supplier tag at manufacture?
Serial management One-time build, then upkeep Serial ranges per GTIN, reissue rules, storage of serials Which system issues and records serials?
Readers and antennas One-time, plus maintenance Number of read points: dock doors, conveyors, tunnels, handhelds Which processes need fixed reads, and which a handheld?
Software and integration One-time, plus support Filtering, event handling, WMS and ERP updates Where do reads become transactions?
Tag performance testing One-time per product family Product materials, packaging, read environment Has the tagged item been tested in your read points?
Process change and training One-time, then refreshers Number of sites, shifts and procedures that change Which steps disappear, and who confirms they did?

Price integration explicitly. GS1 publishes separate standards for reader protocols and for filtering reads into business events (the GS1 RFID page lists Low Level Reader Protocol and Application Level Events among them), but your WMS still has to decide what a read means. A read at a dock door is not a shipment confirmation until software ties it to an order, a load and a time window.

If your WMS or ERP needs changes to accept those events, price that work in the same model.

How reads become WMS transactions

A reader produces a stream of tag reads; a WMS needs transactions. Something in between has to turn one into the other, and its rules decide how much integration work you take on. Write them down before choosing software:

  1. Filter. Drop duplicate reads of the same tag in a short window, and ignore tags outside the zone or outside an open task.
  2. Decode. Convert each EPC into the GS1 identifiers your systems use, for example an SGTIN into a GTIN plus serial (AI 21). The TDS defines this translation, and GS1 publishes an EPC encoder and decoder.
  3. Associate. Tie the reads to a business object: a carton’s license plate, an order, a load, a location.
  4. Compare. Check what was read against what was expected: the ship notice at receiving, the pick list at pack-out, the system quantity in a count.
  5. Raise exceptions. Send shortages, overages and unexpected items to a queue with an owner, instead of adjusting inventory automatically.
  6. Record. Keep the raw reads for a period long enough to settle a customer claim.

Step 5 deserves care. Automatic adjustment from RFID reads can quietly import every stray and double tag into your on-hand balance. We recommend starting with exceptions reviewed by a person and automating only the cases that prove reliable in your own data.

How to estimate RFID payback with your own numbers

Estimate payback for one process at a time: compare the annual cost of doing that process today with its annual cost using RFID, then divide the one-time investment by the annual difference. The structure below follows the University of Arkansas cost comparison in ITRI WP159, which set the variable cost of manual audits (volume, scan rate, wage, number of audit points) and the cost of shipping errors against the per-item tag cost.

  1. Choose one process and one item population. For example, pack-out verification for one customer’s orders. Mixing processes hides where the savings come from.
  2. Cost today’s method. Labor hours equal units handled, times the share audited or counted, times the number of audit points, divided by units per hour. Multiply by your loaded labor rate. Add the error cost: units times the shipped error rate times the cost per error (chargebacks, re-shipments, write-offs).
  3. Cost the RFID method’s running costs. Tags (unless a customer program already requires them), encoding and application labor, support and maintenance, the labor left for exceptions, and the error cost that remains at the read rate you expect.
  4. Cost the one-time investment. Readers, antennas, portals or tunnels, handhelds, encoders, software, integration, tag testing and training.
  5. Calculate. Annual saving equals today’s cost minus RFID running cost. Payback in years equals one-time investment divided by annual saving. If the annual saving is zero or negative, the project does not pay back on this process.
  6. Test the sensitivity. Vary tag cost, error rate and volume. WP159’s authors note that tag cost “is, obviously, the biggest factor” and that “A 1-cent change in per tag cost can have a substantial impact.”
  7. Replace assumptions with pilot data. Read rate, residual error and exception labor should come from a pilot on your own items, not from a brochure or from this article.

Example (fictional). A fictional apparel distributor ships 3,000,000 units a year through pack-out. Today it barcode-scans 20% of units at a pack audit station at 600 units an hour, with loaded labor at $24 an hour: 1,000 hours, or $24,000 a year. Its shipped error rate is 0.8%, and each error unit costs $6 in chargebacks and handling: $144,000 a year. Today’s total is $168,000.

The RFID option uses four conveyor read points, encoders, software and integration for a one-time $210,000. Running costs are $20,000 a year for support, 100 hours of exception labor ($2,400), and residual errors at 0.1% ($18,000). Tags at a quoted $0.06 a unit would add $180,000 a year.

If the distributor pays for tags itself, RFID costs $220,400 a year against $168,000 today and never pays back. If its main customer already requires tags on these units, tags are bought either way and drop out of the comparison: RFID running cost is $40,400, the annual saving is $127,600, and the $210,000 pays back in about 1.6 years, or roughly 20 months.

Annual cost of pack-out verification in the fictional example $0 $100k $200k $168,000 Barcode audit today $220,400 RFID, tags paid by you $40,400 RFID, tags required by customer anyway Tags Shipping errors Support Audit labor
Fictional example data for illustration, not industry figures. The one-time RFID investment ($210,000 in the example) is not in the bars. Method structure follows ITRI WP159 (2012).

Run the sensitivity before you present it

The same fictional example shows how quickly the answer moves. Without a mandate, the annual saving before tags is $127,600, so each cent of tag cost on 3,000,000 units removes $30,000 of it.

Tag cost per unit (fictional) Annual tag cost Annual saving after tags Payback on $210,000
$0.00 (tags required by customer) $0 $127,600 About 1.6 years
$0.01 $30,000 $97,600 About 2.2 years
$0.02 $60,000 $67,600 About 3.1 years
$0.03 $90,000 $37,600 About 5.6 years
$0.04 $120,000 $7,600 About 28 years
$0.06 $180,000 Negative Never

The error rate matters as much. If the fictional distributor’s shipped error rate were 0.3% instead of 0.8%, today’s cost would fall to $78,000, the mandated-case saving to $37,600 a year, and payback would stretch to about 5.6 years.

A low error rate today can sink an RFID business case built on shipping accuracy, and that is a good problem to have. In that case, put the effort into the barcode process first.

When the case fails, fix the barcode process first

If the payback does not work, the problem RFID was meant to solve is still there, and a better barcode process may get you much of the way. The checks below cost little and make any later RFID case stronger, because the baseline becomes honest.

  • Scan, never key. Block manual entry of item and location numbers at the scan prompt, and report every override.
  • Label every location. Location labels with check digits stop confirmations to the wrong bin.
  • Count on a schedule. A cycle counting program by value and velocity finds errors while their cause can still be traced.
  • Code every adjustment. Reason codes on every inventory adjustment show which process creates the errors.
  • Audit by risk. Audit pack-out for the customers and SKUs that generate chargebacks, not a flat percentage of everything.

Run the payback method again after six months. If the barcode process has cut the error rate, the RFID case will look worse, and that is the right answer.

Which SKUs justify an RFID tag

Within a process that suits RFID, not every SKU deserves a tag. Decide by SKU family using five questions, and tag only the families that pass most of them.

Question Favors RFID Favors barcodes
Is the tag cost small against the unit’s margin? Tag is a small share of margin Tag is a large share of a low-margin unit
Are units counted or checked without being handled? Sealed cartons, mixed cartons, dense fixtures Each unit handled anyway
Is unit-level accuracy costly to get wrong? High chargeback exposure, high shrink, costly stockouts Errors are rare and cheap to correct
Does the tagged item read reliably? Passes a tagged-item test in your read points Packaging, contents or load prevent reliable reads
Will the tag be applied at the source? Supplier tags at manufacture Your DC would have to tag it

The fourth question cannot be answered from a datasheet. GS1’s pallet guideline notes that the load can significantly influence read performance, and Auburn’s error list shows poorly performing tags reaching “30% or more” without testing. Test each SKU family as a tagged item in the place it will be read.

Example (fictional). A fictional footwear and accessories distributor scores its SKU families. Boxed footwear passes all five questions: tags at manufacture, sealed cartons at pack-out, high chargeback exposure. Socks in multipacks pass three: low margin per unit, but counted in cartons. Metal belt buckles fail the read test at the pack-out tunnel. The pilot covers footwear only, socks wait for a lower tag quote, and buckles stay on barcodes with a serialized carton label.

What retailer RFID mandates change

A retailer mandate changes who pays for tags and what your shipment can be checked against. It does not, by itself, create a business case for using RFID in your own warehouse.

Mandates are not new. In June 2003, as a University of Arkansas paper records, “Wal-Mart asked its top 100 suppliers to begin using radio frequency identification (RFID) tags on pallets and cases beginning January 2005” (ITRI WP091, Is There a Business Case for RFID?, 2007). Today, searches for “Walmart RFID mandate” return mostly RFID vendor and trade press pages about the current requirements, not the retailer’s own documents.

Confirm with your buyer. We could not read Walmart’s own supplier specification for this note, and we have not verified any category list, start date, tag requirement or placement rule published by third parties. Treat every such claim, including on vendor pages, as unconfirmed until your buyer or the retailer’s supplier documentation confirms it in writing.

Questions to put to the buyer in writing

Topic What to ask
Scope Which departments, items and packaging levels need tags, and from which ship date?
Tag and inlay Is there an approved inlay list or a performance grade the tagged item must meet?
Encoding Which EPC scheme and encoding (for example SGTIN-96), and who manages serials?
Placement Where on the item, label or hang tag, and in what orientation?
Data Must EPCs appear in the ship notice or another message?
Compliance How is compliance checked, at which point, and what are the chargebacks?
Exceptions What happens to untagged or unreadable units already in the pipeline?

Four things a mandate changes inside your building

  1. Tag cost leaves your internal business case. You buy tags either way, as the fictional example above shows, so internal use cases only need to cover readers, software and process change.
  2. Encoding becomes a quality process. Incorrect encoding is on Auburn’s error list, and the General Specifications require the barcode serial and the SGTIN serial to match exactly. Verify encoded tags against the label before they leave.
  3. Your shipment becomes auditable at the customer’s dock. WP159’s authors expected that once suppliers receive claims based on the retailer’s RFID receiving reads, they “will want to ensure that they possess item-level RFID shipping data as a countermeasure.” Reading your own cartons at pack-out gives you that record.
  4. Where tags are applied matters. Auburn’s guidance is that “for RFID to be cost effective, tagging must occur at (or very near) the source.” Applying tags in your own DC is the pattern its authors saw break down in every pilot that relied on it. Push tagging upstream to your suppliers where you can.

Hybrid patterns: barcodes and RFID together

Adding RFID does not mean removing barcodes, and the General Specifications expect both to coexist: where EAN/UPC or ITF-14 carries the GTIN and an EPC/RFID tag carries more, “the same GTIN SHALL be encoded in all GS1 AIDC data carriers” (section 4.15.1). Choose each carrier for the step it serves.

Pattern Barcode does RFID does Watch for
Pick by barcode, verify by RFID Confirms each pick Checks the closed carton at pack-out Reads from neighboring cartons on the conveyor
Count by RFID, resolve by barcode Investigates and adjusts exceptions Finds discrepancies across a zone Stray and double tags inflating counts
Pallet label plus pallet tag GS1-128 SSCC for every partner SSCC EPC read at dock doors Label and tag must carry the same SSCC
Tagged totes, barcoded contents Identifies the products GRAI on the tote tracks the asset Tags must survive washing and handling
Dated tags for perishables Lot and date in a 2D code DSGTIN+ selects short-dated stock Same dates in barcode and tag

If you are also adding 2D barcodes for lot and expiry data, our note on GS1 Sunrise 2027 and 2D barcodes in the warehouse covers the scanner and parsing side.

Run the pilot to answer a business question

Auburn’s 2014 pilot guidance is blunt: “Don’t spend time on trying to answer the question: ‘Does the technology work?’” A pilot should test whether RFID solves a named problem in your operation. Its ten rules start with inventory accuracy, ask you to decide which use cases matter, and end with audits during the pilot. Those audits cover store or site execution, read rates, the sources of missed reads, and anomalies.

For a warehouse pilot, we would add one discipline: measure RFID results against a barcode count of the same items, on the same day, in the same place. That is how the Dillard’s study timed its counts, and it is the only way to separate a read-rate problem from a tagging or process problem.

What to do next

Start by measuring the process you want to fix, before any reader is bought. In our engagements we write the requirements first: which process, which items, which decision the data must support, and how success will be measured. Our software requirements checklist is a starting structure for that document.

  1. Measure accuracy today. Use the inventory accuracy calculator on a sample of locations, so the baseline is a count, not the system’s own figure.
  2. Cost your counting labor. The cycle count planner shows the hours your current program needs at the counts per hour your team achieves; compare that with what you would save.
  3. Pick one process. Receiving, pack-out verification, counting or returnable assets, not all of them at once.
  4. Run the payback method. Use your own volumes, error rates, quotes and wage rates, with the sensitivity table filled in.
  5. Confirm mandate requirements in writing. Use the buyer questions above before you buy tags or encoders.
  6. Test tagged items. Test each SKU family in the read point where it will be read.
  7. Pilot against a barcode baseline. Audit tag presence, read rates and exceptions throughout.

If you would like help writing the requirements, the payback model or the pilot plan, see our barcode, RFID and labeling service. Because we have no vendor partnerships, the specification states the requirement first and the device second.

Working on this problem?

If this field note describes a process you need to fix or build software around, tell us about it. We start with your process, your systems and your deadline.