JD Sports turned on RFID in its German stores in September 2026, making Germany the last country in the retailer’s directly operated estate to go live and completing Phase One of its RFID programme. Dan McGrath, JD Group Head of Customer Operations, set out the numbers in a LinkedIn post: 1,000 stores directly launched in fourteen months, and 2,700 stores live out of 4,800 across the wider global group over the same period, a run rate he put at 192 stores a month. JD Sports, founded in 1981, is a UK-headquartered sports fashion retailer; the Checkpoint Systems announcement of the programme in early 2025 described the group as 4,850 stores across 49 countries including franchises and joint ventures.
Store counts are the easy part of this story to report and the least interesting part of it. The number that matters sits in the middle of McGrath’s post, and it is the one JD had no power to instruct: over the same fourteen months to September 2026, source tagging across JD’s brand partners rose from around 40% of product to 73%.
The number JD could not mandate
Source tagging means the RFID label or hang tag is applied at the point of manufacture, by the supplier, rather than being stuck on later by the retailer in a distribution centre or a stockroom. For a retailer of JD’s shape it is the only version of item-level tagging that scales. JD is a multi-brand retailer: the shelves carry other companies’ products, made in other companies’ factories, under other companies’ tagging policies. A retailer can buy readers, write software and train staff entirely on its own authority. It cannot order a global sportswear brand to add an inlay to a garment tag at source.
So moving that figure from roughly 40% to 73% was a supply-chain persuasion exercise rather than a technology project, and it is the part of the programme least likely to be repeatable by anyone who assumes the hard bit is the hardware. McGrath’s own framing is that the work was about “changing behaviours, processes and mindsets” rather than “scanners, tags and technology”, and he credits the shift to brand partners increasingly recognising what RFID offers in their own retail formats and supply chains, not to pressure from JD.
That 27% gap is not a rounding error either. It is the ceiling on everything else the programme is being built towards. A store cannot behave as a reliable real-time inventory node if roughly a quarter of the units on its shelves carry no tag at all, because the untagged quarter is invisible to every count, every availability check and every downstream system that consumes the data. Read rates on passive UHF RAIN RFID in apparel and footwear are high enough that the tagged population can be trusted; the untagged population cannot be inferred.
What was actually deployed
The platform underneath the programme is ItemOptix, the RFID inventory management software from Checkpoint Systems, a supplier of retail loss prevention and RFID systems. JD signed a multi-year global software agreement covering its Sports Fashion Group stores, beginning in Europe. Notably, JD’s own IT team integrated the RFID processes into JD’s existing store software using Checkpoint’s ItemOptix SDK, rather than handing colleagues a separate application to learn beside the till system they already used. For a rollout measured in stores per week, that decision matters more than it sounds: it removes an entire category of training and change resistance.
Two other suppliers sit either side of the software, and neither has been named in the coverage of this milestone. Zebra Technologies, the enterprise mobile computing firm whose handheld computers are the default in warehouse and retail scanning, supplied the handheld RFID readers. Renovotec, a UK rugged hardware and mobility integrator, acted as the UK deployment and support partner. Between them they cover the two jobs that quietly sink RFID rollouts, which are getting working hardware into a thousand buildings and keeping it working afterwards.
In the distribution network, JD runs fixed tunnel readers, which are RFID portals mounted over a conveyor that read every tag in a carton as it passes through without anyone handling the goods. The Rochdale site near Manchester in the UK went in first, followed by Heerlen in the Netherlands serving continental Europe, with a stated throughput of up to 540,000 items per hour.
The results that justified all of it came from a deliberately small pilot: five stores in the UK, France and Spain in the second half of 2024, which returned a 12% increase in on-shelf availability, a 20% improvement in restocking speed and a 95% increase in items found. From that proof of concept to a full enterprise rollout took 76 working days, McGrath told Retail Technology Innovation Hub in January 2026. The UK and Ireland leg alone was 414 stores in 16 weeks, a pace of 25 stores a week, supported by a week-long immersive change and culture programme that trained 26 retail Change Champions.
Phase One complete is not rollout complete
It is worth being precise about what has finished, because the headline figures invite a bigger reading than they support. The 1,000 stores are the ones JD directly launched, which in McGrath’s wording covers “everything we directly control”. The wider group number he gives, 2,700 live of 4,800, leaves roughly 2,100 stores outside the programme, largely franchise and brand-partner estates where JD’s operating control is limited or indirect. Reaching 2,700 stores in fourteen months across that many territories is a genuinely large piece of work by any retail standard. It is also, by JD’s own accounting, a bit over half of the group.
Confidence rather than accuracy
What makes the source-tagging figure strategically loaded rather than merely operational is where McGrath says the programme is heading. In a May 2026 Retail Technology Innovation Hub interview he opened with “I hate RFID. Not because it doesn’t work. Not because it isn’t valuable. But because most of the industry is still looking at it through the wrong lens”. His objection is to the measure the sector reports. “Accuracy is a report. Confidence is a decision,” he said, and “what’s the point in being 98% accurate, if an agent can’t trust that last 2%?”
The destination he describes is machine-readable inventory. “Stores stop being places with stock and start becoming real-time, queryable inventory nodes,” he said, and “We’re building infrastructure for machines to trade physical goods.” His summary of the shift is blunt: “RFID isn’t a store ops tool anymore. It’s a confidence engine for agentic commerce”. He also noted that he chose Checkpoint Systems and ItemOptix specifically because it is “a fully serviceable, API first RFID platform”, which is a coherent choice if the long-term consumer of the data is software rather than a store manager.
He is dismissive of where the rest of the sector has its attention: “The industry is still talking about cycle counts”. The uncomfortable implication for anyone agreeing with him is that the agentic-commerce case rests entirely on tag coverage. An API-first platform, tunnel readers at 540,000 items an hour and 2,700 live stores are all necessary and none of them are sufficient while 27% of product arrives untagged.
The question left open
Which brand partners moved, and what moved them, is the part of this programme nobody has published. A jump from roughly 40% to 73% of product in fourteen months implies specific commercial conversations with specific suppliers, and the arguments that worked in those rooms are considerably more valuable to the rest of retail than another set of store counts. The same goes for the remaining 27%. Whether that tail closes through supplier economics, through JD tagging at its own distribution centres as a stopgap, or not at all, is the thing to watch in Phase Two.
Read more at https://www.linkedin.com/feed/update/urn:li:activity:7510307314281844736/
