Stock counts fell from two closed nights to three hours, and shrinkage halved.
22 storesOne distribution centre1,400 staff
The situation
What we found
Stocktaking closed two stores every weekend on a rolling cycle, and the count was already wrong by the time trading resumed.
Cages left the distribution centre full and arrived short. Nobody could say whether the shortfall happened at loading, in transit or at the goods-in door.
Shrinkage was measured once a year, as a single number, with no way to tell theft from miscounting from stock written off informally.
Shift attendance was recorded by a supervisor on a clipboard, and payroll disputes were settled by whoever argued hardest.
Where the margin was going
One trading year across all 22 stores, as a share of total stock loss.
Cages short between the depot and the store38% of loss
Never proven at either end, so never recovered from anyone.
Theft from the shelf31% of loss
Counted wrong at stocktake19% of loss
Two closed nights per store, and the count still disagreed with the system.
Damage and expiry written off informally12% of loss
Under a third of the loss was actually theft. The rest was measurement — and measurement problems are cheaper to fix than people problems.
Illustrative scenario
The service blueprint
Everything the customer sees, and everything they do not.
A service blueprint lays the whole service out at once: the touchpoints across the top, and underneath each one the evidence, the customer’s own actions, the staff they meet, the work happening out of sight, and the systems holding it up. The line of visibility is the one that matters — almost every delay lives below it.
A shopping trip, blueprinted end to end
The customer sees a shop. Below the line of visibility sits the machinery that decides whether the shelf is full, the price is right and the alarm is fair.
Arrive
Choose goods
Checkout
Leave the store
Return an item
Physical evidence
ArriveThe entrance, trolleys, and a shelf that is either full or is not.
Choose goodsThe product, its price label and its security tag.
CheckoutThe till, the screen, the receipt.
Leave the storeThe detection gates at the door.
Return an itemThe service desk and a refund receipt.
Customer actions
ArriveWalks in. Forms no opinion yet, unless something is missing.
Choose goodsPicks up goods, including the high-value lines that carry a tag.
CheckoutPays, and expects the price to match the shelf.
Leave the storeWalks out, and expects nothing to beep.
Return an itemBrings an item back, usually without the original receipt.
Line of interaction — the customer touches the organisation here
Onstage — staff the customer meets
ArriveNobody, ideally. A greeter only where footfall justifies it.
Choose goodsShelf staff replenishing, and answering “where is…” questions.
CheckoutThe cashier — and a supervisor if an override is needed.
Leave the storeA guard, who now intervenes far less often.
Return an itemThe service desk assistant, who can now see what was actually sold.
Line of visibility — everything below this line is invisible to the customer
Backstage — work they never see
ArriveOvernight replenishment driven by a count that is actually correct.
Choose goodsTag application at the depot, not the store, so the shelf is never the bottleneck.
CheckoutOverrides and voids recorded against a fingerprint, not a shared supervisor code.
Leave the storeThe gate logs which item triggered it, not merely that something did.
Return an itemThe item’s history is retrievable, so a refund decision is evidence-based.
Line of internal interaction — where staff hand work to systems
Support systems
ArriveStock system fed by handheld UHF counts
Choose goodsUHF tags on cages and pallets; EAS on high-loss lines
CheckoutFingerprint confirmation at the till
Leave the storeEAS detection, and hybrid EAS/RFID on high-value stock
Return an itemItem-level history from the RFID record
Look at the fourth column. The difference between a gate that says “something left” and one that says “this item left” is the difference between accusing a customer and knowing what happened.
Illustrative scenario
The workflow
The same service as a workflow, exceptions and all.
The depot-to-shelf workflow
The single change that recovered the most margin: making a cage account for itself.
1
Cage is loaded and sealed at the depot
Contents written to the cage tag as it passes the dock portal. No paperwork, no counting.
Depot pickerUHF cage tagAutomatic
2
Cage passes the depot dock portal
Departure time and contents recorded automatically as the cage rolls through the door.
No staff involvementDoorway reader portalAutomatic
3
Cage arrives at the store goods-in door
Arrival read happens as it comes through, before anyone signs anything.
No staff involvementDoorway reader portalAutomatic
If the contents differ from departureA discrepancy is raised at that moment, with a departure time, an arrival time and a vehicle. Previously this was a monthly argument with no evidence.
4
Goods received are photographed and confirmed
A photograph attached to the receipt, so a disputed load has evidence rather than opinion.
Goods-in staffPhoto validation
5
Stock moves to the shelf and the count updates
The stock figure reflects what physically arrived, not what the delivery note claimed.
Shelf staffStock system
6
Category counts run during trading hours
A handheld reader counts a category in minutes. Stores stopped closing to count.
Store teamHandheld UHF reader
If the count disagrees with the systemThe gap is investigated the same day, while the cause is still recent enough to find.
Illustrative scenario
The work
What we put in
What we put in, and in what order
Piloted in three stores and the depot, then rolled out over two quarters.
1
Stage one
Cages that check themselves in and out
UHF tags on every cage and pallet, with reader portals at the depot dock and each store’s goods-in door. A cage that leaves full and arrives short is now a recorded fact with a time and a place attached.
38% of loss became visible
2
Stage two
Stocktaking with a reader instead of a clipboard
Handheld UHF readers count a category in minutes without touching the stock.
Two nights to three hours
3
Stage three
Protection on the lines that actually walk
EAS tags and door detection on spirits, infant formula, razors and small electronics — the four categories carrying most of the shelf loss, rather than tagging everything.
4
Stage four
Hybrid tags where both jobs are needed
On high-value stock, one tag that both triggers the door alarm and carries an RFID identity, so a store knows not only that something left but exactly which item it was.
5
Stage five
Attendance that settles its own disputes
Facial recognition at the staff entrance, replacing the clipboard.
6
Stage six
Every till override has a name on it
Fingerprint confirmation on price overrides, refunds and voids, with photo validation on goods received. Both were previously supervisor discretion with no record.
Illustrative scenario
The result
What changed
One trading year later
Same stores, same suppliers, same product range.
Stock countTwo closed nights per store
Accuracy against the system after countingAbout 82%
Depot-to-store shortfalls resolvedAlmost none
Total stock lossBaseline
Payroll disputes each monthAround 40
Everyone suspected everyone, and nobody had evidence.
Stock countThree hours, store open
Accuracy against the system after countingAbove 99%
Depot-to-store shortfalls resolvedTraced to a dock and a time
Total stock lossRoughly half the baseline
Payroll disputes each monthTwo or three
The largest saving came from counting properly, not from catching thieves.
Illustrative scenario
The technology
What each piece is actually for
The technology, in plain words
Retail is where the difference between the two tag families matters most.
UHF RFID on cages and palletsLong-range tags read automatically as a cage passes through a doorway. This is what makes depot-to-store loss provable.Depot dock and store goods-in
Reader portals, in and outFixed readers at the door record both directions, so stock has a departure time and an arrival time rather than just a delivery note.Goods entrances
Handheld UHF readersCounts a shelf or a category without touching it. This is what ends the closed-night stocktake.Stock counting
EAS tags and door detectionThe simple anti-theft tag. It knows something is leaving; it does not know what. Cheap enough to put on a whole category.High-loss lines
Hybrid EAS and RFID tagsOne tag doing both jobs — it triggers the alarm and identifies the exact item. Used only where the stock justifies the cost.Spirits and small electronics
Facial recognitionClock-in that cannot be done by a colleague on your behalf.Staff entrances
Fingerprint recognitionA deliberate confirmation attached to a specific transaction.Till overrides, refunds, voids
Photo validationA photograph captured and checked at the point a delivery is accepted, so a disputed load has evidence.Goods received
Illustrative scenario
About this case study.
Retail is a fictional organisation. The situation, the figures and the outcomes are
illustrative — they show how BTS Consulting approaches a problem of this shape and
what the technology involved actually does. They are not a record of a delivered project,
and nothing here should be read as a client result.
More case studies
The same approach, different sector
A state university
Higher Education
A university where every department held its own version of the student, and nobody could say where the laboratory equipment was.
Tell us where performance is being slowed, hidden or lost. We will help you define the problem, identify the priority actions and build a practical path forward.
