The most expensive part of a WMS is not the software
A warehouse management system does exactly what its data tells it to. What has to be in order before you switch it on
In July 2019, online retailer ASOS cut its profit guidance for the year to £30–35 million. In its trading statement it said the equipment in its new automated warehouse in Berlin had been installed on plan, but problems appeared as more stock went through the systems: in the interaction between the automation and the warehouse software. Transition costs rose from around £35 million to around £47 million, and customers in Europe and the US saw less choice.
ASOS is large and its warehouse is highly automated, so the case is not directly comparable with a manufacturer turning over €15 million. The pattern is the same, though: the hardware was installed on time, and things got stuck where the system meets the real flow of goods. In smaller companies that point is almost always the same: inventory accuracy, master data and the habits of the people in the warehouse.
What a WMS actually does
A warehouse management system (WMS) is an application that takes orders from the ERP and runs the warehouse's daily work by predefined rules: where to put stock away, which location to pick from, when to replenish the pick face. The key word is predefined. A WMS does not know where the goods are. It knows where the data says they should be.
If the data is right, a WMS speeds the warehouse up and cuts errors. If it is not, the WMS does not fix the errors; it executes them faster and with more confidence. The forklift driver is told to pick from a location that is empty. The picker waits for a replenishment the system never triggers because it thinks the stock is still there. Sales promise delivery from stock that exists only in the system.
That is why the licence and implementation price rarely tells you what the project will really cost. The biggest cost is the work before go-live, and the week after go-live when the warehouse does not work.
Four conditions to meet before go-live
1. Inventory accuracy. ASCM gives 95–99% of items within tolerance as the usual target, depending on item value. For a WMS, accuracy by item is not enough. Location counts too: an item that is in the warehouse but in the wrong place is lost to the WMS.
2. Master data. Units of measure and conversions (piece, box, pallet), dimensions and weights, packaging, batches and shelf life, and in manufacturing the bills of materials. A wrong box-to-piece conversion is a nuisance in the ERP. In a WMS it means a wrong pick on every order for that item.
3. Locations and layout. Every location must be labelled, recorded in the system and physically used the way it is recorded. ASCM explicitly recommends mapping the warehouse layout, data capture equipment and goods movements before implementation, down to which forklift carries a scanner.
4. Process and ownership of exceptions. Who fixes a posting that fails? Who decides when physical stock differs from the system? If this is not defined, exceptions pile up, data drifts from reality, and within a few months the WMS is running a warehouse that does not exist.
| Condition | How to check | Who owns it |
|---|---|---|
| Inventory accuracy by location | Cycle count a sample of locations for several weeks in a row | Warehouse manager |
| Master data | Review units, conversions, dimensions and packaging for the most-moved items | Purchasing and engineering, with IT |
| Locations | Walk the warehouse with the system's location list | Warehouse manager |
| Exceptions | Number of unposted or wrong movements a month and time to fix | Process owner appointed by management |
From practice: a problem software did not solve
A metalworking company with €5.4 million in revenue ran SAP. Its annual stock count showed a shortfall of around €150,000. The cause was not the system but how people used it: around 150 operations a month were stuck in an error queue because the system found too little stock when issuing material. The errors were ignored because nobody owned them. Consumption was not posted, EBITDA looked better than it was, and the company was about to be sold.
In four weeks the company introduced daily cycle counting, fixed the posting errors and corrected the bills of materials. It bought no new system. Had it installed a WMS on that data instead, the shortfall would have been the same, only found later and at greater cost.
When a WMS really pays off
A WMS is not a necessary step for every manufacturer. The warehouse module in the ERP is often enough. A dedicated WMS is worth considering when several of these apply:
- several thousand active items or locations, and searching for stock takes measurable time;
- picking errors that customers complain about;
- customer or regulatory requirements for batch and expiry traceability;
- several shifts or warehouses, where knowing what is where can no longer be passed on by word of mouth;
- stock that grows because you cannot find it and order it again.
If only one applies, fixing the process and data in your existing system is usually cheaper.
Decisions before choosing a vendor
ASCM advises basing WMS selection mainly on three functions: directed put-away (the system says where), directed picking (the system says from where) and directed replenishment. Before you approach vendors, also decide:
- Which system owns inventory. ERP or WMS. Two systems that both "know" how much stock there is will sooner or later disagree.
- Where scanning happens. Every movement you do not scan is where data starts to drift.
- Cloud or on-premise. ASCM notes that WMS moves to the cloud more slowly than other systems because the warehouse needs very fast scanner response.
- Who leads the project. If IT leads it, it will succeed technically. If the warehouse leads it with IT support, it will be used.
A go-live without surprises
Do not switch the system on until inventory accuracy by location is stable within the target range. Count every location you migrate just before go-live. Plan the first weeks with fewer orders or more people, not the same volume as before, and measure three numbers every day: picking errors, unposted movements and order-to-dispatch time.
Five questions before you decide
- What is your inventory accuracy by location today, not just by item?
- How many movements a month stay unposted or wrong, and who fixes them?
- Does every high-movement item have correct units, conversions and dimensions?
- Which problem should the WMS solve, and how will you measure it before and after?
- Will someone from the warehouse lead the project?
If you cannot answer three or more, the first step is fixing data and process. That step is worth taking even if you never buy a WMS.
Where to start
For four weeks, count 20 to 30 locations every day, starting with the most-moved items. Record every difference and its cause. After four weeks you will know your accuracy, which causes repeat, and whether you need a new system or different rules in the one you have.
Sources: ASOS plc, Trading statement, 18 July 2019. WMS definition, selection advice and inventory accuracy targets: ASCM (APICS), CSCP 2022 learning system (APICS Dictionary, 16th edition). The metalworking company's figures are from case study CS01 and are measured on actual results.
Case study: Inventory control (PDF) ↗
Inventory management: from master data to excess and obsolete materials
Why the warehouse fills up again after a one-off inventory write-off, and six steps that prevent it: master data, posting movements, cycle counting, BOM standards, S&OP and excess and obsolete stock.
Less inventory, more cash, but not more profit
Reducing inventory releases cash, but EBITDA usually falls in the year of the reduction. Three traps, a case from practice, a board table and six steps to keep the warehouse from filling up again.

