Blog

Paper Price Tags Are Disappearing in 2026. The Label Is the Easy Bit.

Posted on by Support

Walk into any large grocery store this year and the paper price tag is on its way out. Walmart said in March that digital shelf labels were already live in 2,300 US stores, with the rest of the chain expected within the year (CNN). Morrisons is putting 10.8 million electronic labels into 497 UK stores starting early 2026. Co-op has 1,500 stores done and plans to finish the remaining 900 by December. In Norway, roughly 70% of all shelf labels are already electronic.

Those are good headlines. They also hide the thing that actually decides whether a rollout works.

An electronic shelf label is a 37-gram piece of plastic with a battery and a radio in it. It’s the simplest, cheapest, most commoditised part of the system. What decides whether 100,000 of them behave for five years is the wireless underneath, the rail they clip into, the battery arithmetic you did before you signed, and the software somebody else wrote. That’s the part that gets under-budgeted, and it’s the part you can’t fix after the boxes land.

The market stopped being a maybe

For years ESLs were the technology everyone praised at conferences and few deployed. Then the deployments got big enough that the forecasters had to redraw their charts.

Transforma Insights counts 1.2 billion connected ESL devices in 2024 and now projects 4.9 billion by 2034 — a 15% compound annual growth rate, and by 2034 that’s close to 5 billion individual displays (Transforma Insights, January 2026).

What’s odd about this market is how lumpy the growth is. It doesn’t climb like a subscription curve. It jumps whenever one buyer commits, because one buyer committing means millions of units. Morrisons alone is 10.8 million labels. Walmart’s rollout touched 2,300 stores before most people noticed it had started. That’s the pattern worth internalising: the average is boring, the order book isn’t, and capacity planning in this space happens in lumps.

Norway is the place to look if you want to see the end state. Pricer estimates around 70% of shelf labels in Norwegian stores are already electronic. High labour rates got them there faster than anywhere else.

The 2026 plot twist: the labels turned into a privacy argument

Here’s the story nobody predicted three years ago.

In September 2026, Walmart CEO John Furner published a letter promising the company does not use personal information — income, shopping history, or a customer’s willingness to pay — to set prices, and won’t in future. His wording was blunt: “We don’t set different prices based on who you are or the time of day, and we won’t.” Walmart also said it’s holding its AI shopping assistant to the same rule.

Why does the CEO of the world’s biggest retailer spend a public letter on price tags? Because shoppers, consumer advocates and lawmakers started asking what fast, centralised price changes could turn into. In August 2026 the US Federal Trade Commission issued a policy statement on personalised pricing, warning companies that the practice could violate consumer protection law and that they must disclose how personal information is used to set a price (FTC, August 2026).

Now translate that into hardware terms, because this is where integrators win or lose arguments.

The label does not set prices. It displays them. Every pricing decision lives in the retailer’s own ERP, POS or price management system — not in the tag, and not in the gateway. But the architecture you choose decides how defensible the whole thing looks when someone official asks a question. Per-label unique identity. A confirmation that the displayed value actually changed, not just that an update was sent. A timestamped audit trail of which labels received which content, when, and from which approved price list. Signed, encrypted updates so nobody spoofs a price over the air.

That isn’t marketing language. When a price dispute reaches a regulator or a reporter, the sentence that ends the conversation is: our system applied 4,381 approved changes at 06:12 from price list v44, and here is the acknowledgement from every label that received one. Hardware that can’t produce that sentence turns a pricing argument into a legal one.

The less glamorous benefit is the one that pays for the project: shelf price and checkout price finally agree, every time, without a human walking the aisles with a printed sheet.

What actually has to work in the field

This is the boring section, and it’s the one that decides your margins.

Take KKM’s Bluetooth ESL line as a concrete example, because the numbers on the datasheet are the numbers you’ll be arguing about in eighteen months.

2.1″ e-ink display tag (KESL021) — 74.0 × 40.0 × 12.3 mm, a 48.7 × 24.5 mm visible area, 212 × 104 pixels at 95 DPI, 37 g. Black-and-white, or black-white-red on the KESL021R. Runs on 2 × CR2450 with a quoted 3–5 years at three refreshes per day. Operating range 0°C to 50°C.

2.9″ e-ink display tag (KESL029) — 93.0 × 44.0 × 12.3 mm, 66.9 × 29.0 mm visible, 296 × 128 pixels at 112 DPI, 37 g, same battery and the same 3–5 year figure on the same duty cycle. Also available with a red channel (KESL029C-R).

Both are Bluetooth 5.0 at 2.4 GHz, bistable e-paper with a near-180° viewing angle, and support partial-area refresh — you can update a price without redrawing the whole label.

The access point in the same family is the KGateway01C: BLE 5.0/4.0, a quoted 30 m radius, listed at 500 labels per AP, 5V DC or PoE, backhaul over 4G, Wi-Fi or Ethernet, IP54, and a -10°C to +60°C working range.

Now the honest part, because datasheets all have the same blind spot.

The battery figure assumes three refreshes a day. That’s a steady-state retail week. It is not a promotion week, and it’s definitely not a store doing daily markdowns on fresh produce. Work out your worst week, multiply it across a year, and check the arithmetic before you promise a customer five years. A label that dies in month nineteen means a technician in an aisle with a ladder, and that cost arrives all at once.

0°C to 50°C is an operating range, not a suggestion. The -20°C to 70°C figure on the same sheet is storage. If a food retailer asks you to cover the chiller and freezer aisles, that’s a different spec conversation with the supplier, and it needs to happen before you quote, not after the pilot.

A 30 m radius is measured with nothing in the way. Shelf labels live inside racking full of product, surrounded by shoppers, and 2.4 GHz does not care about your floor plan. Metal and liquid are the two worst offenders and a supermarket is full of both. Design coverage at a fraction of the datasheet number and then walk the aisles with a scanner to confirm.

500 labels per AP and “up to 1,000” are both true. KKM lists 500 per access point in the spec table and suggests planning around 1,000 per gateway in its guidance. Both numbers are real; the variable between them is how long a full-store update takes. If every label in the store has to be refreshed before the doors open at 8am, that timing constraint sets your device count, not the coverage map.

E-paper holds its image with zero power, which cuts both ways. A label that never receives an update keeps showing the old price — silently, forever. It looks identical to a label that’s working. That’s why per-label acknowledgement matters more than the transmission log.

The rail is doing more work than people give it credit for. KKM’s KESL-B shelf rail has a 22 mm fixing hole pitch, 40 mm height and weighs 135 g per metre, with a theft-proof option and a matching removal tool. In a store where people peel labels off, the rail and the tool are the difference between a managed asset and a consumable.

Colour costs you something. Black-white-red gets attention on a promo, but red-channel e-paper refreshes more slowly and reads differently under store lighting than the plain black-and-white version. Pick per aisle, not per store.

This stopped being a retail-only story

The most interesting deployments in 2026 aren’t in supermarkets.

Volkswagen has installed ESLs at its Dresden plant, its Wolfsburg headquarters and other facilities in Europe and the US, mainly to help workers find locations on the line. Samsung deployed LED-blinking labels in a warehouse for real-time inventory accuracy. Bosch and DAU D&D are on the same list, and pharmacy and clinical settings are starting to use digital labels for medication information, storage requirements and expiry dates.

That’s the shift worth noticing. In a plant or a pharmacy, the label isn’t merchandising — it’s an instruction surface. Wrong content on a shelf is a margin problem. Wrong content on a pharmacy shelf or a production line is a compliance and safety problem. Which is exactly why “did the label actually update” goes from a nice-to-have to a hard requirement the moment you leave the grocery aisle.

A 12-store chain, 108,000 labels, and 216 boxes on the ceiling

Here’s a scenario that shows up regularly enough to be worth walking through.

A regional grocery group runs twelve stores, and its pricing department publishes a weekly promotion cycle plus mid-week markdowns on fresh. Across the estate that’s roughly 2,000 price changes a week, printed onto paper tags and swapped overnight by staff who have better things to do. Two of the stores keep failing internal price-integrity checks — the shelf says one thing, the till says another — and the group has been fined once for it. Their estimate was around 400 staff hours a month spent on tags across the chain, which is the number that actually got the project approved.

They sized it at roughly 9,000 labels per store, which lands at about 108,000 labels across twelve stores, and eighteen access points per store — 216 gateway boxes in total, mounted above the shelving rather than on it.

The hardware that went in:

  • KESL029 for the standard dry-goods shelving, where a 2.9″ display fits the existing rail width and the extra pixels make room for unit pricing and a barcode.
  • KESL021 for the narrow strips — the small-format aisles where a 93 mm label overhangs the shelf edge.
  • KESL-B rails with the theft-proof option in the two aisles that kept losing tags, plus the removal tool for staff.
  • KGateway01C access points on PoE where the store already had cabling, and on 4G backhaul in the two older sites where it didn’t.
  • A phone app and a Raspberry Pi rather than an access point for two small convenience-format stores, because at that size an AP is an unnecessary box. That’s a real option, not a downgrade.

The integration is the part worth being precise about. The chain’s existing price management system — the one already feeding their tills — pushes approved changes out through KKM’s open API to the gateways, which write to the labels. No new price engine, no second source of truth, no cloud account in the middle. The retailer kept the logic, the audit trail and the responsibility for what a price should be.

The rollout that kept it from becoming an expensive pilot: two stores first, timed across a full promotion week, measured on how long a complete estate-wide update actually took end to end and how much battery was consumed doing it. Then ten more stores using the real numbers rather than the optimistic ones.

Two quarters in, the fill-in picture looked like this. The overnight tag labour mostly disappeared and got redeployed to shelf availability. Price-integrity failures dropped to the point where the manual audit became a spot check instead of a weekly routine. The markdown workflow on fresh got faster because a partial refresh could change one price on a big label instead of reprinting a whole sheet. And when a customer complained about a price, the store manager could pull up what the label had been told to display and when — which ended more disputes at the service desk than it ever reached head office.

None of it was exotic. Labels, rails, access points, and an API into software the retailer already owned and paid for.

Before you spec a hundred thousand labels

Seven questions that separate a deployment from a very large pilot:

  1. What’s your real refresh rate, not your average one? Do the battery arithmetic on your worst promotion week and your daily markdown aisle. Multiply by the number of labels and the price of a service visit.
  2. What’s the operating temperature range, and where does it fail? Chillers and freezers need an answer before you quote, not after.
  3. How many labels per access point, and how long does a full update take? Coverage gets you 500 labels per AP. Update windows get you your device count.
  4. What happens when a label misses its update? Look for per-label acknowledgement, not just a transmission log. Silent failure is the expensive kind.
  5. Who owns the price logic? If the answer isn’t “our system”, you’ve bought a platform subscription by accident.
  6. Can you prove what a label was showing on a given day? Unique IDs, timestamped updates, signed content. This is the part that answers regulators.
  7. Does the rail suit the shelf? Hole pitch, height, theft protection, removal tool. The mounting is where rollouts go sideways at 2am.

Where KKM fits

KKM builds the hardware: BLE e-ink labels in 2.1″ and 2.9″, the KGateway01C access point, the shelf rails, and the OEM and ODM work behind them — firmware customisation, private labelling, packaging, new product development from mechanism design through PCBA to casing. The factory is ISO9001 certified, the products carry CE, FCC, BQB, RoHS and REACH, and the company has been doing this since 2008 with 50-plus patents on the design and algorithm side.

What KKM doesn’t sell you is a store platform. No monitoring dashboard, no cloud account holding your price data, no SaaS seat you rent back every month. You get the development kit — the template tooling, its source code, and the API — and you build the price engine, the alerting and the audit trail inside the software you already run. For a retailer with a working ERP, or an integrator delivering into one, that’s the right shape. Your pricing logic and your compliance evidence are your IP, and they belong in your stack.

The label is the cheap part. Getting the radio, the rail, the battery and the audit trail right is the work. In 2026, with labels going onto shelves by the million and regulators asking what they’re allowed to display, that work is the whole difference between a rollout and a write-off.

Sources

Support