Guide · Retail technology
What the connected floor actually returns.
Most retail still happens in the store, on technology a decade behind the website. The fix is real, and so is the way it gets bought wrong. Single-capability tag projects rarely pay back; the floor returns when one shelf-edge network does several jobs. Here is what actually pays, and what to refuse.
This guide, in 6 parts
The floor is where the money still changes hands.
E-commerce has been measured and rebuilt for twenty years. The store floor, where roughly five sales in six still happen, runs on paper, guesswork and systems no one owns. That is the gap, and the shelf-edge spending above is the market trying to close it.
The kit we build to close it lives in The Connected Floor. This guide is the part before the purchase order: how to make that spend return, and how most retailers accidentally make it fail.
« The floor is retail's last unoptimized surface. »
Why single-capability tag projects fail.
Five failure modes, and not one of them is the hardware. They are what happens when a floor project is bought as a device for one job.
The price-change-only business case never clears
Costed against the labour saved swapping paper tags alone, electronic shelf labels rarely pay back inside a hold period. Finance discounts the soft benefits to zero, and the project dies in committee. The number works when the hardware carries more than one job, not when it carries one.
One capability, one cable run, paid for once
A tag rollout buys the hardest, most expensive part of any floor project: the install, the network, the integration into pricing and inventory. Spending all of that to win a single use case throws away the four others the same infrastructure could have carried for almost nothing more.
It was scoped as hardware, so nobody owned the outcome
A tag is a device with a price on it, easy to buy and easy to forget. The moment it ships, the question of what number it was meant to move has no owner. Hardware projects close on installation; operating changes close on a result, and these get scoped as the former.
It never touched the systems that hold the work
A shelf label is only as good as the pricing, promotion and inventory feeds behind it. Wired to a clean export instead of the live systems, the floor shows yesterday's price and last week's stock, and staff go back to trusting paper. The data plumbing is the project; the tag is the easy part.
Staff were trained on the device, not the new routine
The label changes how a price is set, how a gap on the shelf is found, how an order is picked. Train the floor on the gadget and not the routine it replaces, and the old habits survive next to the new kit. Adoption is the work, and a single-capability rollout almost never budgets for it.
Bundle the use cases that share one install.
The shelf-edge network is bought once: the install, the wiring, the integration. Load it with four jobs and the same infrastructure that wins the price case carries the rest for almost nothing more.
Centrally pushed prices and promotions, live at the shelf in minutes, not over two days of staff walking the aisle with a label gun. This is the labour case finance already understands, and the reason the infrastructure gets approved. It is the floor, not the ceiling.
Lights at the shelf edge guide in-store picking for click-and-collect and ship-from-store, cutting pick errors from the 3 to 8 percent of manual picking toward under half a percent. The same shelf network that shows a price can show a picker exactly where to go.
Self-scan and scan-and-go let a shopper check out at the shelf or on their own phone, pulling people out of the queue at peak. The shelf-edge tags carry the codes and the live price the scan reads, so the two systems lean on one source of truth.
A code on the shelf opens the full range, the reviews, the size that is out of stock here but in the warehouse, and lets the shopper buy it on the spot. The floor stops being limited to what fits on it. One tag estate, and the aisle reaches the whole catalogue.
Each job leans on the same source of truth, the live price and the real stock, which is why they belong on one network and not in four separate projects. Pick-to-light and ship-from-store in particular only work when the floor knows what it actually holds.
Do not buy electronic shelf labels just to change prices.
This is the most common way the spend is wasted. The labour saved by retiring paper tags is real, but on its own it seldom clears a business case, and finance is right to discount it. Four things to refuse.
- Do not buy electronic shelf labels just to change prices. The labour saving alone rarely clears the business case, and you will have paid for the expensive part to win the cheapest use case.
- Do not let it be scoped as a hardware purchase. A device with an install date and no owner for the result will close on installation and move no number.
- Do not wire it to an export. If the tags do not touch the live pricing, promotion and inventory systems, the floor shows stale data and staff go back to paper.
- Do not skip the routine. Budget to retrain how prices are set, gaps are found and orders are picked, or the old habits survive next to the new kit.
Walmart is putting digital shelf labels into about 2,300 US stores, going shelf-wide by the end of 2026, and the reason it pencils is that the labels do pricing, inventory accuracy and staff tasks at once. Take that lesson, not just the headline: buy the network once, and make it earn across several jobs.
Start from the number, not the catalogue.
Pick the one floor problem with a cost on it: slow price changes, pick errors, queues at peak, sales lost to out-of-stocks. Design the shelf-edge install so it can carry that case and the adjacent ones on the same network, prove it in a handful of stores, and measure adoption rather than installation.
Which floor problem is actually worth solving first, and how the store and the website should share one truth behind it, is its own question. We keep our map of it private; if that is the call in front of you, it is the conversation to have, and it is where the Hub Map starts.
Frequently asked.
Are electronic shelf labels worth it?
Only if they carry more than price changes. Costed against the labour of swapping paper tags alone, the business case for electronic shelf labels rarely clears a hold period, which is why so many proposals die in finance. The same hardware and the same install can also drive pick-to-light fulfilment, self-checkout and QR commerce off the shelf edge. Bundle those and the infrastructure pays back; buy it for price changes alone and it usually does not.
Why do in-store technology projects fail?
Most fail for one of two reasons that have nothing to do with the hardware. Either they were scoped to win a single capability, so the most expensive part, the install and the data integration, gets spent on the smallest return; or they were treated as a device to buy rather than an operating change to run, so no one owned the number the project was meant to move. The tag is the easy part. The data plumbing and the new floor routine are the project.
What returns does in-store retail technology actually deliver?
Real return comes from a connected floor where one shelf-edge network does several jobs: price and promotion automation that saves the labour finance already counts; pick-to-light that cuts in-store pick errors from 3 to 8 percent toward under half a percent; self-scan checkout that shortens the queue at peak; and QR commerce that opens the full catalogue from the aisle. With roughly 83 percent of US retail still physical, the floor is the surface with the most unworked return left in it.
Should we follow Walmart into digital shelf labels?
Take the lesson, not just the headline. Walmart is rolling digital shelf labels across about 2,300 US stores and aims to go shelf-wide by the end of 2026, but at that scale the labels are a platform for pricing, inventory accuracy and staff tasks, not a price-changing gadget. The transferable move is to buy the shelf-edge network once and load it with several jobs. Copying the rollout without the bundle copies the cost and leaves most of the return on the table.
Where should we start on the connected floor?
Start from the number, not the catalogue. Pick the one floor problem with a cost on it, slow price changes, pick errors, queues at peak, lost sales to out-of-stocks, and design the shelf-edge install so it can carry that case and the adjacent ones on the same network. Prove it in a handful of stores, watch adoption rather than installation, then extend. The infrastructure is bought once; the discipline is making it earn across more than one job.
Much of this is yours to scope. The floor problem worth solving and the number it should move belong to the people who run your stores, and they should name them first.
Bring us in when the shelf-edge network has to carry more than one job and stay working through a peak, from price automation to pick-to-light to checkout to QR. We build it and we stay until staff and customers use it. See The Connected Floor.
Weighing a floor-technology spend?
No deck, no gate. A working session on your real floor and your real numbers, with the operators who would make it return.