Why nobody likes answering "how much does it cost"
When a company asks what automation costs, it usually hears an evasive "it depends." That's frustrating but also true — and the problem can be worked around. The price doesn't come from some secret rate card; it comes from two simple things: how many distinct steps you want to automate, and how many systems have to start talking to each other.
Automating one clear step — say, copying the figures from an incoming invoice into your accounting software — is cheap and quick. Wiring your shop, warehouse, accounting and email into one smooth flow costs more, because there are more places where something can go wrong. This article won't give you exact figures for your company — no honest person can do that through an article — but it will give you a framework to estimate the cost yourself and spot when a quote makes no sense.
Small modules versus one big build
There are two ways to approach automation. The first is the big project: a few months of analysis, then months of development, and at the end you switch on a large system meant to solve everything at once. The second is modular: you pick one specific problem, solve it in a few weeks, measure the gain, and only then decide on the next step.
For small and mid-size companies we almost always recommend the second approach. Not because it's fashionable, but because it carries less risk. With a big project you pay years in advance for something you won't see working until the end — and if it turns out you understood the brief differently along the way, fixing it is expensive. With small modules you get a first tangible result quickly, you learn as you go, and you fund each next step out of the savings the previous one produced.
Ballpark numbers — and why they're only ballpark
To give you something to hold on to, here are typical ranges. Treat them as illustrative, not as a price list — the real figure depends on your specific situation.
A smaller module that handles one clearly bounded task (say, automatically processing incoming orders or sending customer notifications) typically lands somewhere in the hundreds to low thousands of euros. A mid-size project that connects two or three systems and carries its own logic tends to run into the higher thousands. A large custom system spanning the whole company gets into the tens of thousands. On top of the one-off cost, budget for an ongoing one too — running it, small tweaks, and fees for any AI services. This range is genuinely wide precisely because "automation" covers everything from a single script to an enterprise system. You'll get a concrete number only after a short conversation about your process.
How to think about payback without spreadsheets
Payback sounds like a topic for an accountant, but it can be sketched on a napkin. Ask: how many hours a month does someone currently spend on the task you want to automate? Multiply that by that person's hourly cost. That's what you spend on manual work each month.
If the automation cost, say, the equivalent of three months of that manual work, it pays for itself in a quarter and then simply saves you money. If it would take three years to recoup, it's a weak idea — either the wrong process was chosen, or it was over-engineered. To the pure time saving, add things that are hard to quantify: fewer typos, invoices sent on time, a customer who gets an answer at midnight. Don't overstate them, but don't forget them either — they often decide whether the investment was worth it.
Build custom or buy off the shelf
Every project starts with a choice: use an off-the-shelf tool you subscribe to, or have a custom solution built? The answer is almost always a mix, governed by a simple rule. Wherever you do what thousands of other companies do — invoicing, email campaigns, attendance tracking — an off-the-shelf tool almost certainly exists, and building your own would be a waste.
You build custom only what is unique to your company: the specific way you process orders, your discount rules, the link between systems that otherwise don't talk to each other. In practice most good solutions assemble a handful of off-the-shelf tools and write a thin custom layer between them that teaches them to cooperate. Anyone offering to build from scratch even the things you can rent for a few euros a month is selling you work, not a result.
One cheap question that saves thousands
Before you sign anything, ask the supplier a single question: "What's the smallest step that can deliver a result we'll measure within a month?" A good answer is concrete and modest. A bad answer is a sprawling six-month plan that has to be paid for in full up front.
Automation isn't an expense that either lands entirely well or entirely badly. It's a series of small decisions, each of which can be small, measurable, and paid for out of the savings of the one before it. Split it up this way and the price stops being a bogeyman and becomes an ordinary line item you decide on with the same calm as a new company car.