Electronic shelf label pilot: scope, KPIs and duration

An electronic shelf label pilot is a bounded trial, usually one store and one or two departments, with indicators measured before installation and success criteria written down. It should run long enough to include at least one promotional cycle and one bulk price change. Without both, it is a demo, not a pilot.

This page is the practical half of phase 3 in the implementation guide. That page explains why a pilot exists; this one covers how to set it up, what to measure and when to stop. It does not describe any vendor’s pilot offer, including Keonex’s, which publishes this site. It is a method you can apply with anyone.

What a pilot is for (and what it is not)

A pilot answers questions that brochures and slide decks cannot. Does the signal reach the back shelf of the chiller? How many codes in the product master end up unpaired? How long does a price change take to show on the shelf, end to end? What does the team do when a label goes blank on a Saturday?

It is not a way to prove return on investment for the whole chain. One store will not give you a reliable labour saving, and anyone who extrapolates one is selling, not measuring. Use the total cost calculator to line up the overall numbers, and leave the pilot to prove what it can: that the system works in your conditions.

Scope: what goes in and what stays out

A good rule is to choose by difficulty, not convenience. The calmest, tidiest store makes the pilot go well and the rollout go badly.

Decision	Useful criterion	Common mistake
Store	A typical format for your estate, with an engaged store manager	Picking the new or refitted store, which does not represent the rest
Departments	One fast-moving with frequent promotions, plus one hard zone (chilled, back of store, dense display)	A single uniform department
Products	Include awkward cases: weighed items, multipacks, dual pricing, unit pricing	Selecting only the clean codes
Integration	The real one, even if partial. If the rollout will read from the ERP or EPOS, so should the pilot	Manual file uploads that will not exist later
Templates	The ones planned for rollout, with the longest descriptions in the range	Shop-window designs with short product names

Integration is where people mislead themselves most easily. A pilot with hand-loaded prices proves that labels show what they are told to show, which nobody doubts. What you need to see is the whole chain: change at source, propagation, label, till. The detail of that chain is in the ERP and EPOS integration guide.

Write down what is out of scope: other stores, a rework of buying processes, changes to pricing policy. If the pilot is used to fix everything else, you will not know what failed.

Duration: a criterion, not a number

No law or standard fixes how long a pilot should be, and you should distrust anyone who quotes a figure without knowing your store. What you can reason about is what has to happen inside the window:

1. An initial installation and pairing phase, where the faults are data and fitting problems.
2. One full promotional cycle, with offers added, changed and withdrawn. This is where gaps between the label system and the till show up.
3. One bulk price change, such as a tariff review of a whole product family. It measures the real load on the radio network and the platform.
4. A stability period after fixes, with no changes while you measure again. Without it you do not know whether the fixes hold.

Added together, that does not fit into a few days. As an editorial guide, not drawn from any source, plan for several weeks and set an interim review date. As a public reference that real pilots run longer than a demo, Consum stated that it had kept five pilot stores since 2019 before its current rollout, according to Alimarket (in Spanish). That is a large Spanish grocer, not a norm.

Baseline: measure before you install

This is the step almost everyone skips, and without it the KPIs mean nothing. If you do not know how many price errors you had with paper, you cannot say they fell.

For one or two weeks before installation, in the same departments:

• Price changes. How many were made, how much staff time each batch took (printing, cutting, placing, checking) and how late they were compared with the change at the till. Ask the store to time real batches rather than estimate.
• Shelf-to-till mismatches. A random sample each week: take the shelf price, scan it at the till and note whether they match. It is a simple check, and it is the one that counts in a dispute; what the rules say about a mismatch is in shelf price different from till price.
• Ticketing problems. Fallen, illegible or out-of-date offer tickets.
• Customer complaints about price, if you log them.

Keep the measurement sheets as they are. When the pilot ends, the people deciding will want to see the before measured the same way as the after.

Pilot KPIs

Pick few. Each indicator needs a formula, a data source, someone who collects it and a frequency.

KPI	How it is calculated	What it reveals
Time per price change	From approval at source to the label showing it; separately, staff minutes per batch	Whether the system shortens propagation and manual work really falls
Till price errors	Weekly sample of products: shelf against EPOS, as a percentage of matches	Data quality and sync; the headline indicator
Label faults	Labels not communicating, blank, showing the wrong image or broken, per hundred installed per week	Hardware reliability, battery life and in-store handling
Gateway coverage	Share of labels confirming the latest update; and those confirming late	Whether the coverage survey was right, especially in hard zones
Pairing	Share of products paired without manual intervention on first load	How clean the product master is
Staff effort	Store minutes per week spent on label issues	The hidden cost that is not on the datasheet
Open issues and vendor resolution time	Log with opening and closing dates	The real quality of support

Two points of method. First, separate label faults from data faults. A label showing the wrong price because the ERP held it wrong is not a system failure, but it is a problem the rollout will inherit. The diagnosis by cause is in why an electronic label does not update. Second, if any promotion in the cycle includes a price cut, check that the previous price displayed meets the reference-price rules. In Great Britain a “was” price is judged under the misleading-actions provisions of the Digital Markets, Competition and Consumers Act 2024 (section 226) and the trading standards guidance on pricing practices; the EU requires the lowest price of the previous 30 days (Directive (EU) 2019/2161). If you sell on both sides of the Channel, decide which rule your labels follow before the pilot starts. Either way, the platform must hold that data; see the 30-day reference price guide. This is not legal advice for your case.

Success criteria: write them first

A success criterion works if two different people looking at the data reach the same conclusion. “It works well” does not qualify. Examples of how to phrase them, with thresholds you must set yourself because they depend on your case:

• Shelf-to-till match at or above an agreed percentage over the last two weeks of measurement, with no product family below an agreed floor.
• No approved price change takes longer to appear than the defined limit, with 100% of labels inside coverage in the hard zones.
• Label faults below a set threshold per hundred units per week, and no label left without an identified cause.
• Automatic pairing of at least an agreed share of the pilot range on first load, with a dated plan for the rest.
• Resolution time for the vendor’s critical issues within the contract deadline.

There is no universal percentage. If a vendor offers a threshold, ask them to put it in writing and say how it is measured. And decide in advance who draws the conclusion, from which data, and at which meeting.

Exit terms

Negotiate the exit before installing, because afterwards your position is weaker. Ask for these to be in the agreement:

• What happens if the criteria are not met. A limited extension to fix things, termination or a change of scope, at your choice.
• Who removes the hardware and at whose cost, by when, and what state the shelves are left in.
• What happens to your data: catalogue, pairings, platform history. Return or deletion, with confirmation.
• Exit costs, including any charge for cancelling a subscription or a lease already started. If the pilot is leased, read lease or buy first.
• No implied commitment to roll out. The pilot must not turn into an order just because you never said no.
• Mutual confidentiality. What the vendor may say about the trial, for example naming your business as a reference.
• Liability for price errors during the trial. If a label from the system shows a wrong price and the till undercharges, who covers the difference.

Pilot checklist

Copy it as a worksheet.

Before

• One owner for the pilot in store and one at head office.
• Written scope: store, departments, number of products, hard zone included.
• Coverage survey reviewed and test labels in the worst spots.
• Trial catalogue import done and errors sorted by cause.
• One to two weeks of baseline measurement, sheets kept.
• KPIs, formulas, sources and frequency agreed.
• Success criteria and thresholds signed by both parties.
• Exit terms reviewed by whoever signs.
• Contingency plan: printed fallback for one department if communication fails.

During

• Weekly shelf-to-till sample, always by the same method.
• Issue log with date, cause and who resolved it.
• At least one full promotional cycle and one bulk price change.
• Mid-point review meeting, with decisions recorded.
• Final period with no configuration changes, to check stability.

At close

• Report with KPIs against the baseline.
• Explicit decision: continue, fix, renegotiate or exit.
• Lessons for the rollout: data to clean, templates, training.
• If exiting: hardware removal and data handling as per the contract.

Mistakes that invalidate a pilot

• Measuring afterwards without having measured before.
• Choosing the easiest store and the easiest products.
• Changing the configuration mid-trial and comparing as if nothing happened.
• Letting the vendor collect and present the data without your own check in store.
• Dragging it out until “it’s already installed” makes the decision for you.

What comes next

If the result is favourable, the next step is the rollout sequence in the implementation guide. If it is not, the pilot report is the document that justifies stopping or renegotiating with evidence.

This site is published by Keonex, which develops electronic shelf label solutions for retail. The method on this page is general and does not describe any Keonex offer.