A single shop is coherent because one person can hold it in their head. The owner knows what is in stock, what a thing costs, what sold yesterday and which supplier is late.
A chain is not that shop repeated. It is a set of outlets that will steadily diverge — on price, on product codes, on how a return is recorded, on what counts as a sale — unless something actively holds them together. Divergence is the default state, not a failure of discipline.
The question for any growing retailer is what that something is, and at what point informal coordination stops working. In practice it stops somewhere around the third or fourth outlet, and the symptoms are always the same.
The four things outlets disagree about
Price. The same item at different prices in different outlets, usually because a promotion ended in one place and not another, or because a manual override became permanent. Customers notice, and staff cannot defend it.
Product identity. The same item under two codes, or two items under one. This is the most corrosive divergence because it silently invalidates every consolidated report. Chain-wide stock figures for an item that exists under three codes are not wrong by a little — they are meaningless.
Stock position. Head office believes an outlet holds forty units; the outlet holds twelve. Transfers between outlets are the usual cause, recorded at one end and not the other.
Performance. Two outlets reporting sales computed differently — one including tax, one excluding, one counting returns against the day of return and the other against the day of sale. Comparison becomes impossible, and comparison is the entire point of having a chain.
Why the numbers stop reconciling
The mechanism is nearly always the same: each outlet has a local record, head office has a consolidated one, and reconciliation is a manual step performed by someone at month end.
Manual reconciliation has a predictable lifecycle. It is done thoroughly for two months, partially for the next four, and then it becomes a formality — the consolidated number is accepted because checking it takes three days nobody has.
At that point the chain is running on numbers nobody has verified, and decisions about purchasing, promotion and expansion are being made on them. This is the specific failure examined in why multi-outlet businesses lose money between systems.
What "one truth" actually requires
It is not one screen. It is a small set of structural commitments, and each one is a decision rather than a feature.
One product master, centrally controlled. Outlets may not create products. This single rule prevents more downstream chaos than anything else, and it is the one most often conceded to under operational pressure.
One price list, with overrides that are visible and expiring. Local pricing flexibility is legitimate in some chains. Permanent invisible overrides are not, and the difference is whether the override has an end date and a name attached.
Transfers recorded at both ends, or not permitted. A transfer is two events. Systems that record dispatch and trust receipt to follow produce the largest and most persistent stock discrepancies in retail.
One definition per metric, written down. What a sale is, when a return is counted, whether tax is included. Boring, unglamorous, and the reason two outlets can finally be compared.
Sales recorded at the point of sale, not summarised and entered later. Summarised entry loses the detail that makes analysis possible, and it introduces a transcription step.
What genuinely should stay local
Centralising everything produces its own failure — outlets that cannot respond to anything without a call to head office, and staff who work around the system to serve a customer standing in front of them.
Reasonable local authority: accepting a return within stated limits, applying a defined discount band, ordering fast-moving stock within a threshold, correcting an obvious billing error.
The design rule is that local action should be recorded rather than prevented. A system that blocks a legitimate local decision creates a workaround; one that permits it and records it keeps the chain informed. This is the configuration question addressed in configuring software to the business.
Adding the next outlet
The test of a chain's systems is what happens when outlet eleven opens.
If opening requires a new installation, a fresh product list, local decisions about codes and prices, and a period during which the new outlet's numbers cannot be trusted, then every expansion carries a fixed cost and a data cost. If it requires creating an outlet in an existing system and stocking it, expansion is an operational event rather than a project.
Truffaire built a multi-ERP deployment for a retail chain running over thirty outlets, and this is the property that determined the architecture: outlets are configuration, not installations. What that costs to maintain over years is the subject of what it takes to maintain software for years.
The count that keeps it honest
Every structural commitment above degrades silently unless something checks it.
A rolling count — a small number of lines counted in every outlet every week, rather than a full count annually — is what surfaces divergence while it is still small. The full annual count tells you the year was wrong; the rolling count tells you which outlet, which product and roughly when. That mechanism is covered in detail in why the count never matches.
Frequently asked questions
At what size does this become necessary?
Usually the third or fourth outlet. Below that, one attentive person can hold the chain together; above it, informal coordination fails quietly rather than visibly.
Can outlets keep their own systems?
They can, and then the reconciliation between them becomes the job. Every chain that has tried this ends up employing someone to do manually what a common product master does structurally.
What about franchised outlets?
The commercial relationship differs; the data requirement does not. Common product identity and common metric definitions are what make a franchise network measurable at all.
Does every outlet need to be online continuously?
No, and it should not depend on it. Outlets must keep selling through a connectivity failure and reconcile when it returns. A chain that stops trading when the link drops has a design fault.
How do we handle an existing mess of product codes?
Deliberately and once, as a mapping exercise before any consolidation is trusted. Doing it gradually means running two identity schemes concurrently, which is worse — see moving years of records without losing them.
Where to start
Pick one fast-moving product and ask every outlet what it costs and how many they hold.
If the answers require checking, or if they disagree, you have found the reason head office reports feel approximate — and product identity is where the fix begins.
Whether a chain needs retail-specific software or a configured general system is the question in choosing technology for your sector.
If you want a chain running on one set of numbers, get in touch.