A warehouse works on knowledge before it works on systems. Someone knows where things are, which supplier is reliable, what moves fast, and which pallet at the back has been there too long.
That knowledge is genuinely valuable, and it scales badly. It works at one warehouse with a stable team. It degrades when volume rises, when staff turn over, when SKU count grows, and when a second location opens — and the degradation is gradual enough that nobody identifies the moment it stopped working.
This article is about recognising that point, and about what a warehouse system has to do beyond holding stock numbers.
Inventory accuracy and warehouse operation are different problems
Worth separating, because they get conflated.
Inventory accuracy asks: how much of this do we have? That is a stock variance problem, with its own six causes.
Warehouse operation asks: where is it, how quickly can someone retrieve it, and did the right thing leave the building?
A business can have accurate totals and a badly run warehouse — correct quantities, but pickers walking twice the necessary distance and dispatching the wrong variant. The second problem costs labour and customer trust, and it does not appear in a stock report.
The signals that knowledge has stopped scaling
Picking depends on who is working. If order accuracy varies by shift, the process lives in people rather than in the system.
New staff take months to be useful. Long ramp-up is a measure of undocumented knowledge, not of complexity.
Items are found by searching. If retrieving something involves looking in the likely places, there is no location system — there is a habit.
The same item lives in several places. Partly received stock, returns not put away, overflow that became permanent. Each duplicate location is a place a count goes wrong.
Returns pile up. Returns are slow to process because they need inspection and a decision, so they get deferred — and become stock that exists physically but not in the record, or vice versa.
Dispatch errors are found by customers. The most expensive detection mechanism there is.
Any one of these is manageable. Three together means the operation has outgrown its knowledge.
What a warehouse system actually has to do
Know where things are
Location tracking is the core function, and it is what distinguishes a warehouse system from inventory software. Every item has a place; the system knows it; retrieval does not depend on memory.
The practical consequence is that a new picker can be productive on day one — because the system tells them where to go rather than requiring them to have learned it.
Direct the pick, not just list it
A picking list is a document. A picking route is a sequence ordered by location, so one pass collects an order rather than crossing the floor repeatedly.
For multi-line orders at volume, this is usually the largest available labour saving, and it requires nothing but the system knowing locations and the floor layout.
Handle receiving properly
Goods in is where accuracy is won or lost. Recording against the purchase order, capturing shortfalls and damage at the point of receipt, and directing put-away to a specific location — rather than "somewhere in aisle four" — determines whether everything downstream is reliable.
Make transfers two-sided
Stock leaving one location and arriving at another must be a single linked transaction with a dispatch and a receipt, and in-transit stock must stay visible. Anything else produces the discrepancy-in-two-places problem described in where multi-outlet businesses lose money.
Treat returns as a first-class flow
Returns need inspection, a disposition decision — resaleable, repair, scrap — and a route back to a location or a write-off. Where this flow is undesigned, returns become an informal pile, and the pile becomes variance.
Support rotation where it matters
For anything with expiry or batch traceability, the system has to enforce the rotation rule rather than rely on a picker choosing correctly. If first-expiry-first-out depends on someone reading dates under time pressure, it will not hold.
What we have seen
Warehouse, inventory, delivery and logistics systems are among the ten we have delivered, alongside distribution operations where the same failure appears in different vocabulary.
Two findings recur.
Put-away discipline decides picking accuracy. Businesses focus effort on picking because that is where errors are noticed. But most picking errors originate at put-away — something placed in the wrong location is then reliably picked wrong by someone following the system correctly. Fixing receiving and put-away improves picking without touching picking.
The floor decides the software, not the reverse. A location scheme that does not match how the physical space is actually organised gets ignored within weeks. We map the floor as it is before designing anything, which is the same principle as documenting an operation before specifying software — covered in how Truffaire builds software.
Frequently asked questions
At what size does a distributor need a warehouse system?
Volume matters less than SKU count and staff turnover. A high-volume operation with fifty stable SKUs and a long-serving team can run on knowledge. A lower-volume operation with two thousand SKUs and rotating staff cannot.
Do we need barcodes and scanners?
They help substantially with receiving and picking accuracy by removing identification and transcription errors. They are not a prerequisite — a location system with disciplined put-away delivers much of the benefit. Scanning without locations mostly speeds up finding the wrong thing.
Is this different from inventory software?
Yes. Inventory software answers how much. A warehouse system answers where, in what sequence, and did the right thing leave. Many products do the first and describe themselves as doing the second.
How do we handle stock that is partly received?
It needs a real state in the system — received-pending-inspection or similar — rather than being either fully in or invisible. Undefined intermediate states are where the "same item in several places" problem starts.
What about integrating with couriers and transport?
Useful, and secondary. Dispatch accuracy is decided inside the warehouse. A courier integration on top of an unreliable pick just tracks the wrong parcel more precisely.
Where to start
Before evaluating software, do two things. Map where items are actually stored, including the informal overflow locations. And record your last month of dispatch errors — what went wrong, and at which step.
Those two documents will tell you whether your problem is location, put-away, picking, or returns. Most operations find it is put-away, which is both the cheapest to fix and the least likely to have been the original complaint.
SPEXA covers warehouse alongside inventory and delivery in one record, so stock does not exist differently in two systems. If you want a read on where your accuracy is actually being lost, get in touch.