The ceiling
How a warehouse discovers the spreadsheet has stopped scaling
Spreadsheet-and-scanner is a legitimate way to run a small operation, and the teams that run it well are resourceful. These are the signs the operation has outgrown it — not of a bad team.
- 01
Stock exists only where one person remembers it
The count in the sheet and the count on the shelf drift apart, and the gap is bridged by the supervisor who knows where things really are. The operation now has a single point of failure with annual leave.
- 02
Picking time grows faster than order volume
With no system assigning locations or sequencing picks, every new SKU and every new hire adds walking. Throughput per person falls exactly as the business grows — the curve that eventually forces the decision.
- 03
The ERP and the floor tell different stories
Finance sees stock the floor cannot find and sells stock that is already promised. Every discrepancy becomes an investigation, and the month-end count becomes an event the whole operation braces for.
- 04
Peak season is survived, not managed
The busiest weeks run on overtime, heroics and quiet error rates nobody measures. When the plan for peak is “everyone works harder”, the operation is telling you its coordination layer is full.
The vocabulary
Locations, waves and picking paths — what a WMS actually manages
Strip the feature lists away and a WMS manages three things. Locations: every bin and rack position is addressable, so stock is where the system says rather than where someone put it, and putaway becomes a directed task instead of a judgement call. Waves: orders are grouped and released to the floor in planned batches — by carrier deadline, by zone, by priority — so the work arrives in an order the floor can flow with rather than as it happens to come in. Picking paths: within a wave, the system sequences each picker’s tasks along the geometry of the floor, which is where the walking — most of a picker’s day in an undirected warehouse — is taken out.
Everything else on the brochure — slotting, cycle counting, labour tracking, yard management — is an elaboration of those three. The reason the vocabulary matters is that it turns the selection from comparing feature counts into asking which product’s model of locations, waves and paths fits the floor you actually run: its order profile, its mix of case and unit picking, its physical shape. A mid-market product whose model fits beats an enterprise suite whose model must be fought.
The standing decision
Change the process or change the product: pricing the real trade
Every gap between the product and your process forces this choice, dozens of times per project. The honest default is configuration; customisation is a debt taken on deliberately for the few processes that genuinely differentiate you.
| Option | What it means on the floor | Cost over the system’s life | What happens at upgrade time | Best for |
|---|---|---|---|---|
| Configure and adapt the process | The floor adopts the product’s standard flow; supervisors retrain the habit, not the software. | Paid once, mostly in training and change management during the project. | Upgrades install cleanly; the vendor’s improvements arrive for free. | The default for every process that does not win you customers |
| Customise the product | The software bends to a process you keep — justified when that process is genuinely how you compete. | Paid forever: every customisation is code someone must maintain, test and re-justify. | Every upgrade becomes a project; enough customisation and the system quietly freezes at its current version. | The two or three flows that are truly your edge — priced knowingly |
| Fill the gap by integration | The special logic lives in a separate service beside the WMS, talking to it through its standard interfaces. | A middle cost: owned code, but isolated from the product’s core and replaceable on its own. | The WMS upgrades cleanly; the side service is retested against stable interfaces. | Differentiating logic that does not need to live inside the product |
The implementation
Migrating a live warehouse without gambling the peak on one weekend
The two risks that sink WMS projects are dirty data arriving in a clean system, and a cutover scheduled against the operation’s calendar instead of around it. Both are managed by sequence.
Clean the data where it lives
Stock records, SKU masters, locations and units of measure are reconciled in the current system first — because migration copies whatever exists, and a wrong count moved into a new system becomes a trusted wrong count. A physical count the new system can inherit with confidence.
Rehearse the migration
The load is run into a test environment, timed, and checked by reconciliation reports — repeatedly, until the team knows exactly how long it takes and what breaks. The real cutover should be a performance of a rehearsed script. A timed runbook with a rollback decision point.
Choose the window from the calendar
Cutover lands in the operation’s quietest period, far from Ramadan peaks, seasonal campaigns and contractual surges — and never in the run-up to the busiest quarter on the theory that the team will manage. A date the operation agreed to, not one it was given.
Run the seam deliberately
For a defined period the old records stay readable and the reconciliation reports run daily, so a discrepancy is caught in days. Only when the counts have held through real operation does the old way get switched off. An operation that never had a day without a trustworthy stock record.
Our role
A selection with no product on our side of the table
- 01
Selection from your order profile
We derive the requirements from your actual order lines, SKU mix and floor geometry, and score candidate products against them — we resell nothing, so the recommendation has no commission behind it.
- 02
The ERP and transport joins
We build the integration with the ERP and transport systems so orders, stock and shipments reconcile across all three — the join most projects underestimate and most go-live pain comes from.
- 03
The migration and the cutover
We run the data-cleaning, the rehearsals and the cutover runbook, with the rollback decision defined in advance — so the go-live is boring, which is the highest compliment a cutover can earn.
Which processes adapt to the product and which justify customisation are operating decisions we argue openly and you make; after go-live, the configuration, the integrations and the runbooks are documented and handed to your team.
Common questions
What buyers ask
Is our operation big enough to justify a WMS?+
Size is the wrong measure; strain is the right one. A compact warehouse with high order lines, many SKUs and tight carrier deadlines can need a WMS before a much larger one moving pallets of few products ever does. The signals that matter are the ones on this page: stock accuracy depending on individual memory, picking productivity falling as volume grows, the ERP and the floor disagreeing, peak survived on heroics. If none of those are present, we will say so and recommend spending nothing yet — sometimes with small fixes that extend the current setup another season.
The vendor says implementation takes a few months. What makes it overrun?+
Three things, in a stable order of frequency: data that turned out dirtier than anyone admitted, customisation that grew because nobody priced the configure-versus-customise decision honestly, and integration with the ERP discovered late to be more than a file transfer. None of these are software problems, which is why they do not appear in the vendor’s estimate. The counters are equally unglamorous: reconcile the stock records before the project starts, hold a hard line on customisation, and build the ERP join early enough to find its surprises when they are cheap.
Can we go live in one warehouse first, or does it have to be everywhere at once?+
One site first, almost always. A pilot warehouse absorbs the lessons — the configuration that did not fit, the training that needed a second pass, the integration edge cases — at a scale where they are corrections rather than crises. The rollout to the remaining sites then reuses a proven configuration and a rehearsed runbook. The exception is an operation whose sites share one stock pool so tightly that split-brain inventory would be worse than a coordinated cutover; that is discoverable in advance, and it changes the plan rather than the principle.
In this cluster
A readiness review that maps your order profile, stock accuracy and seasonal calendar, and returns a recommendation on whether to implement now, fix the data first, or scale the current setup another year.
Put it against your operation
