Zebra Technologies announced on 4 November 2026 that it has helped JD Sports reach 98% on-shelf availability and 20% faster replenishment, using Android handheld computers fitted with passive UHF RFID sleds. Issued from Zebra’s Bourne End office in England, the release is the vendor’s account of a rollout RFID News has covered twice before, and it puts Zebra’s name, its kit and a headline figure on work previously told through the software side.
The figure is the awkward part. Six months earlier, in May 2026, JD Sports’ head of customer operations Daniel McGrath used the same number to describe the problem rather than the achievement. He asked Retail Technology Innovation Hub: “what’s the point in being 98% accurate, if an agent can’t trust that last 2%?” The vendor is now celebrating the number its customer singled out as the thing in the way.
Two different 98s
Those are not the same measure, and conflating them would be the easy mistake. On-shelf availability asks whether a size that exists somewhere in the building is actually out on the shop floor where a customer can find it. Inventory accuracy asks whether the count held in the system matches the stock physically present. Zebra’s 98% is the first, McGrath’s was the second, and a store can push stock out well while still holding a count it does not fully trust.
The distinction matters because of what McGrath says he is building towards. In that interview he argued that “Accuracy is a report. Confidence is a decision”, and that RFID is “a confidence engine for agentic commerce” in which stores become “real-time, queryable inventory nodes”. For a human browsing a shop, 98% availability is a good trip out. For software committing to a 30-minute delivery, the residual 2% is the risk.
The kit, and the bake-off that chose it
JD Sports is a UK-headquartered sports fashion retailer founded in Bury in 1981. Zebra Technologies (NASDAQ: ZBRA) is a US enterprise hardware maker whose range covers rugged Android handhelds, barcode scanners and passive UHF RFID readers.
The hardware detail is new. Stores use Zebra TC2 Series mobile computers paired with Zebra RFD40 UHF RFID sleds. A sled is a reader module that clips onto the bottom of a handheld the colleague already carries for barcode scanning, so one device does both jobs. Zebra specifies the RFD40 at an industry-best 1300-plus tag reads per second. The TC2 Series sits at the entry-level end of Zebra’s Android range, marketed as “smartphone styling, and pricing, with big business features”. JD did not buy the premium handheld, which fits McGrath’s framing that this was about behaviours, processes and mindsets rather than scanners, tags and technology.
Zebra’s accompanying case study adds a competitive detail absent from the press release: JD ran identical software on Zebra devices and on hardware from two other vendors across five pilot stores. Zebra’s read tags faster and at greater range, and were judged easiest for staff to pick up and use. Ease of use deciding it, rather than raw read rate, is worth noting for anyone specifying handheld RFID in retail.
JD also bought Zebra OneCare, which lets a store swap a broken device at no cost to that store. McGrath: “Stores used to fear that a broken device meant an unexpected bill they couldn’t cover… and that’s changed how confidently they run their stores.”
From a 90-day audit to 25 minutes
The strongest new number is not the 98%. It is the arithmetic of the count. Before RFID, JD ran a physical audit every 90 days that tied up ten people for a full eight-hour day. McGrath, in the case study: “Now we do a stock-count of an average store in 25 minutes, and our biggest store of 60,000 square feet at the Trafford Centre, in an hour. We can effectively count twice a week for an hour of someone’s time.”
Mechanically it is unremarkable passive UHF work done well. A colleague clips the sled on, walks the floor and stockroom sweeping rails and shelves, and the reader logs every tagged item in range while a live count builds on screen. Zebra says a whole store can sometimes be covered in a single pass.
The outcomes follow from that frequency. Colleagues move 95% more inventory from the storeroom to the shop floor than under the old pen-and-paper process, replenishment runs 20% faster, and on orders only one specific store can fulfil, completion rates have climbed by 4% to 8%. There is also a shrink angle neither of our earlier pieces covered: teams “can see where loss is happening rather than waiting for a 90- or 180-day audit”.
The suppliers, officially named
Zebra confirms on the record what we reported in September 2026 about who did what. Renovotec, a UK rugged hardware and mobility integrator and a Zebra Premier Solution Partner, led the device deployment. Checkpoint Systems, the loss prevention and RFID software firm behind the ItemOptix platform, is described as a Zebra Registered ISV and integrated its software for a single view of inventory data. JD’s own IT team built that into existing store applications using Checkpoint’s ItemOptix SDK, rather than handing colleagues a separate app. Checkpoint won the multi-year global software agreement in March 2026 after a 2024 pilot across five stores in the UK, France and Spain lifted on-shelf availability by 12%. Zebra gives no baseline for its 98%, so the improvement it represents cannot be derived from the release.
A snapshot of where things stood months ago
Zebra stops at 414 stores across the UK and Ireland, reached in roughly three months with the first store live in 76 days. That is the opening leg, not the current state. By the end of September 2026 JD had switched on RFID in Germany, the last country in its directly operated estate, completing Phase One at around 1,000 directly launched stores in fourteen months, with 2,700 live out of roughly 4,800 across the wider global group.
Two small discrepancies are worth recording for anyone citing these figures later. Zebra puts the pace at 26 additional stores a week on average, where our September 2026 reporting had 414 stores in 16 weeks, a pace of 25 a week. And Zebra describes the group as “nearly 5,000 stores across 46 countries”, where Checkpoint in March 2026 said 4,850 across 49.
The 27% no count can see
This is the part Zebra’s release does not touch. Source tagging means the RFID tag is applied at the point of manufacture by the brand that makes the product, so the item arrives at the distribution centre already readable. Where it is absent, someone tags the item downstream or it stays invisible. Across JD’s brand partners, source tagging rose from roughly 40% to 73% of product in the fourteen months to September 2026, so roughly a quarter of units still arrive untagged.
Both figures can be true at once, and it is worth being precise about how. On-shelf availability is measured on what the system can see, and the untagged population does not appear in an RFID sweep at all, so it cannot be counted as available or as missing: it sits outside the measurement. No source states that the 98% is calculated only on tagged stock, so this is an open question rather than an accusation of a flattering metric: neither company has published how a 98% availability figure reconciles with 27% of product arriving untagged. For a retailer whose destination is inventory that software agents can trade against, that reconciliation will matter more than the headline. Wider coverage sits on our retail RFID hub.
What JD is looking at next
Zebra’s case study sketches the roadmap: RFID-enabled self-checkouts, smart fitting rooms that add data where loss tends to go undetected, and overhead readers capable of converting unknown loss into known loss. The same visibility, it says, could give AI shopping assistants confidence that a product is actually on the shelf. That last item is McGrath’s agentic commerce thesis arriving in vendor marketing copy.
Our March 2026 report on the Checkpoint agreement covered the software decision, and our September 2026 piece on the end of Phase One set out the source-tagging ceiling. Zebra adds the hardware, the supplier roles and a number. Whether that number means what the headline implies depends on a gap neither company has quantified.
