Home / WMS software

Logistics — warehouse software selection

WMS software: how to run a selection you will not regret

Most WMS selections are decided by a demo that the supplier controls, on data the supplier chose. This page is about taking that control back: how to score a shortlist, what to insist on in a demo, the questions that expose a weak fit, and the costs that appear after the quote is signed. If you first want to know what a WMS actually does, read our explainer on that — this page assumes you already do.

By Tom Joseph · Last updated: 18 August 2026

Free consultation Pricing

In short

A WMS selection is won or lost before the demo, in two decisions: which of your own order profiles the vendors must run in front of you, and how the shortlist is scored. Vendors demo a warm, tidy operation with clean master data; your Tuesday afternoon is not that. Insist they run your top three order profiles — including the awkward one — with your own article file, and score against weighted criteria written before the first call. We start with a €2,500 audit that produces exactly those inputs: the profiles, the weighted criteria, and an honest read on whether a WMS is the right spend at all.

The flow through the warehouse

Where a WMS takes the decision over

Five moments where the system decides what happens — and what a selection has to test.

Receiving
Arrival
A receipt that arrives short is the first place system and reality diverge.
Locations
Where it sits
The ERP knows the quantity. The WMS knows location, condition and unit — only that makes picking possible.
Picking
Directed work
If the pick list comes out by article number rather than walking route, you do not have a WMS.
Master data
What delays it
Slotting needs dimensions and weights across the range. This work starts before you choose a supplier.
Go-live
One flow first
Start with outbound picking. That is where the measurable gain and the fewest upstream dependencies are.

Differentiators

What actually separates one package from another

Every WMS on a Dutch shortlist will handle receiving, putaway, picking and shipping. Feature checklists therefore tell you almost nothing — everyone ticks everything. The differences that matter appear in four places, and they are the ones to score:

How much changes without a developer. Adding a picking rule, a print layout or a validation step: configuration by a trained key user, or a paid change request with a lead time? This single answer determines your cost of ownership more than the licence does.
The depth of the unit-of-measure model. If you sell in eaches, inners, cases and pallets, and buy in a different hierarchy, ask them to model your actual conversion. Some packages handle two levels comfortably and bend awkwardly at four.
What happens on the exception path. Short picks, damaged stock, a customer changing an order after release. The happy path looks identical everywhere; the exception handling is where packages diverge sharply.
Who does the implementation. The vendor, an implementation partner, or you. This changes the price, the timeline and — most importantly — who is still reachable in month nine when something breaks.

Scoring

Write the scoring before you take the first call

A shortlist scored after the demos scores the demos, not the packages. Write weighted criteria first, circulate them internally, and do not change the weights once you have seen a vendor — change them only if you learn something about your operation.

A workable split for a single-site Dutch operation:

Fit to your order profiles (35%) — measured on the demo you scripted, not on claims.
Configurability without development (20%) — how much your own team can change in year two.
Integration effort with your ERP (20%) — including who has done that specific ERP before, with references you may call.
Total cost over five years (15%) — licence plus implementation plus the annual change budget, not the licence alone.
Supplier viability and support (10%) — support hours against your shift pattern, and what a Saturday incident costs.

The demo

The demo script: your orders, your data, your awkward Tuesday

A vendor demo is a rehearsed performance on data engineered to flatter the product. It is not dishonest, but it is not evidence either. Replace it with a scripted session and the shortlist reorders itself surprisingly often.

Send every vendor the same pack at least two weeks ahead: an extract of your article file including the messy SKUs, three real order profiles, and this sequence to perform live:

Receive a purchase order that arrives short, then receive the remainder two days later against the same order.
Pick your highest-line-count order and show the walking sequence, then pick a single-line order and show how the system avoids sending someone across the building for one item.
Break something mid-pick — mark a location empty that the system believes is full — and show what the picker sees and who gets told.
Change an order after release and show whether the change reaches the floor or arrives too late.
Show the same flows on the actual handheld your staff would use, not on a laptop browser simulating one.

Vendor questions

Five questions that expose a weak fit

Ask these of every vendor, in writing, and compare the answers side by side. Vague answers are themselves an answer.

"Which of your last five implementations were on our ERP, and may we call two of them?" — a supplier who has done your ERP three times prices the interface differently from one discovering it on your project.
"What in this demo was configuration and what was development?" — asked immediately after the demo, while the screens are fresh.
"What is your change-request rate and lead time in year one?" — the honest ones have a number; it tells you what to budget.
"What does master-data migration cost, and who does the measuring?" — see the next section.
"If we outgrow this in five years, how does our data leave?" — export formats and cost of exit, agreed while you still have leverage.

The quote

What the quote does not include

WMS quotes are usually accurate about what they cover. The problem is the scope boundary. Four costs land after signature with dependable regularity:

Master-data preparation. Slotting needs dimensions, weights, units per case and stackability across the whole range — not the 200 SKUs someone measured once. Whoever does that measuring, it is weeks of work and it is rarely in the quote.
Hardware and the wireless survey. Handhelds, printers, and a site survey that regularly finds dead spots at exactly the racking heights you pick from.
The second half of the interface. The WMS side is quoted; your ERP partner's side often is not, and the two are quoted by different companies who each assume the other handles the edge cases.
Backfill during go-live. Your best warehouse people will be in training and testing rather than on the floor, in a period when throughput must not drop.

Phasing

What "go live on one flow" actually means

Everyone agrees a phased go-live is wiser and then plans a big-bang weekend anyway, because phasing means running two truths at once. It is worth the discomfort.

The usual first phase is outbound picking, because it produces the most measurable gain and the fewest upstream dependencies. Receiving stays on the old process, stock is handed over at a defined boundary, and one person owns reconciling the two pictures daily until inbound follows.

Judge the phase on a small number of things: pick accuracy against your baseline, lines per hour once the learning curve flattens rather than in week one, and how many exceptions needed a supervisor override. That last number is the one that predicts whether phase two goes well.

Not yet

Signals you are not ready to buy yet

Some operations should buy a WMS but not this quarter. These are the signals that a selection started now will produce a bad decision:

Nobody can state your current pick accuracy. Without a baseline you cannot evaluate a business case or prove a benefit afterwards.
The ERP is also being replaced within a year. Two moving systems means the interface is built twice.
Locations are not yet labelled and fixed. A WMS on informal locations automates the confusion.
No named internal owner with time. Not a sponsor — someone with hours in their week for master data and exception rules.
If several of these are true, barcode scanning on your existing stock module buys accuracy now and makes the later selection sharper. It is also a legitimate final destination, not only a stepping stone.

FAQ

Frequently asked questions

How long should a WMS selection take?

Six to ten weeks from writing criteria to signing, for a single site. Faster usually means the demo script was skipped.

Two weeks to write weighted criteria and assemble the demo pack from your own data.
Three to four weeks of scripted sessions, with at least a week between them so scores are recorded while fresh.
Two to three weeks for references, contract and the exit clause — the part most often rushed.

How many vendors should be on the shortlist?

Three for a scripted demo, from a longlist of six to eight. More than three and the scripted sessions stop being comparable because your own team tires.

Filter the longlist on two hard criteria — proven work on your ERP, and support hours matching your shifts — before any demo.
Include one supplier outside the obvious Dutch set if your profile is unusual; it calibrates the others' pricing.
Tell all three they are one of three. It changes what appears in the quote.

Should we ask for a proof of concept before committing?

Only where a specific risk cannot be settled by a scripted demo — an unusual unit hierarchy, a difficult interface, or automation already on site.

A proof of concept with a vague goal becomes an unpaid extended demo and tells you little.
Define the pass criterion numerically in advance, on your data.
Where the approach is genuinely uncertain we run this as a €20,000 proof of concept before production work, precisely so the risk is priced rather than discovered.

What should we prepare before the first vendor call?

Three things, and all of them are useful regardless of which supplier wins.

Your three real order profiles, including the awkward one you would rather not show.
An article-file extract with the messy SKUs left in — odd units, missing dimensions, anything that has caused an argument.
Your current pick accuracy and lines per hour. Without a baseline no business case can be tested afterwards.

What does a WMS project cost with Crux Digits?

We work in fixed steps so the risk is priced before the commitment rather than after.

A €2,500 audit producing the order profiles, the weighted criteria and an honest read on whether a WMS is the right spend.
A €20,000 proof of concept where a specific risk needs settling on your own data first.
Production from €50,000. Package licences are separate and paid to the vendor you choose — we do not resell them, which is why we can tell you not to buy one.

Running a WMS selection this year?

The €2,500 audit produces your order profiles, weighted scoring criteria and demo script — and says plainly when scanning on your current system is the better spend.

Book a conversation