Custom or off-the-shelf software in logistics: how to choose

For your WMS and your accounts a standard package is usually the right call. For the work between your WMS and your clients often not, and then you keep a custom project small by starting with one process.

Custom software is built for your own processes, while off-the-shelf software is bought the way other companies buy it. Whether you need custom software in logistics comes down to one question. Does your warehouse do this work the way almost every other warehouse does, or in a way that belongs to you and your clients alone? In the first case you buy a package. In the second you have it built, one process at a time.

When is off-the-shelf software the better choice?

When the process runs the same everywhere. A WMS or an accounting package has been built and tested many times by others, and the vendor keeps it up to date when the law or the market changes, so rebuilding it as custom software costs you more and gets you less.

We say that even when it costs us the job. We do not replace a WMS and we do not rebuild planning or routing software. A vendor who has worked on that for years knows the edge cases better than we would after a few months, and you buy that experience along with the package.

A package has one more advantage. You see it running before you sign, and you can ring another warehouse that uses it. With custom work you cannot. So the approach has to make up for that with a measurement before anything is built and a written agreement on what the system must do.

Where does the work around your WMS need custom software?

In the seam between your WMS and the rest of the supply chain. That is where the work sits that differs per client. No package can even it out for you, because the difference comes from your client, and that client will not change its files because your WMS expects a different column.

Take a client who emails orders as a PDF. A second sends Excel with its own column names. A third sends EDI, but its article numbers differ from yours. The WMS expects one format, so someone in the office types the difference away every morning. How a system reads those orders and gets them ready is in orders from email into the WMS.

Rates that differ per contract work the same way. So does extra handling that never reaches an invoice, and so do the staffing agency invoices you want to check against your own clock data. The WMS does its job. The manual work sits around it.

That is where custom AI makes sense. A system reads what comes in, checks it against the agreement and gets it ready in the package you already run. Your people keep the screens they know, with no new platform or login, and an order the system is unsure about goes to a queue where someone in the office looks at it before it reaches the WMS.

Why is a big platform riskier than a small system?

Because a big platform changes everything at once. Then you can no longer see which part works and which part does not. With a small system around one process you see that quickly, and you see it on your own orders.

A platform rollout touches order intake, invoicing, reporting and everyone’s login in the office in one go. The team has to learn new screens while the warehouse keeps running. The data has to move, and every client has to be connected again. If one part goes wrong, the rest waits. Six months in, the conversation is often about when it will be finished, and hardly about what it delivers.

A small system does not carry that weight. It reads the orders of one client, or it checks the invoices from your staffing agencies. You know up front what it has to do and you measure it. If it does not work, the damage stays inside one process and you also know why it failed, which you use straight away on the next attempt. If it works, you have proof.

Your operations manager can judge a system like that without help, by checking in the WMS whether the orders that used to be typed by hand are now waiting there. With a platform you need a project organisation to know where you stand.

How do you run an AI project one step at a time?

Start with one process and put in writing what the system must do, then measure how that process runs today. The next step only comes once this one works and the measurement shows it. With us that follows the order of how we work.

Free call

In a free 45-minute call we go through where your hours go on manual work: orders, hours, invoices or documents. And which systems run today. Often it is clear afterwards which process comes first. If it is not a fit, you hear so by the end of that call.

One process and a fixed price

Before the build we write down what the system has to do. It is a definition you can measure again later: which orders it reads, what it puts in the WMS and what it always leaves to a person. You approve it before we build.

A baseline measurement goes with it. It records how the process runs today before anything is built, so that after the build you do the same count again and see the difference on paper instead of by feel.

The price is fixed up front. If the system does not meet the agreed definition, you do not pay the final instalment.

Build and roll out

We build in the systems you already have and roll it out with the people who do the work today. A person checks every correction and every flag first. That stays the case until the system has proven itself on your own data.

If the system touches the attendance, conduct or performance of employees, the works council has a right of consent (article 27 WOR). You arrange that before the build. We supply the documents for it.

Ongoing maintenance

After that you pay a fixed monthly fee that you can cancel every month. Each month you get a report of what the system did and found. We maintain it and change it when your work changes, for example when a new client comes on board.

Only when the first system works and the measurement shows it do we talk about the next process. That might be invoicing extra handling, or checking staffing agency invoices. Sometimes there is nothing left to do after that. Then we say so.

What do you keep if you stop?

Everything that was built. You can take what we build for you with you, even if you stop working with us. Your data and the output can always be exported.

You pay no licence per user. If ten more people join the office, the system costs no more. With a standard package you often do pay per account.

Custom software does have one downside: you depend on whoever maintains it. That is why you get documentation, and why your data and your accounts stay yours.

Package or custom: the test per process

  • Does this process run the same at almost every warehouse? Then a package is usually the better buy.
  • Does the difference come from your clients, who will not change their files for your WMS? Then the work sits in the seam, and that is where custom software fits.
  • Can you write down what the system must do, so that you can measure it again after the build? If not, start with the baseline.
  • Can you start with one client or one process? A plan that replaces everything at once is a platform rollout.
  • Do you know who maintains it, and can you take your data with you if you stop?

In one warehouse orders from email pass that test first, in another the invoices from agencies or work that was never invoiced. What we do shows the processes we most often build first, and AI in the warehouse helps you prepare that choice.

If your process falls between two answers, bring it to an introductory call.

Does this sound familiar?

In a 45 minute first call we look at your own process and tell you honestly whether AI pays off there.