SME digitalization benchmark: how is the Netherlands doing?

Layer 1 — Getting started: this is what we’re good at

The figures everyone brandishes—and they’re correct, too.

The basic level means using at least four of the twelve technologies measured by the EU. That's a low bar, and almost all of us meet it.

Layer 2 — Moving forward: this is where progress stalls

Same sources, different question. Not do you use it, but how deep does it go.

The difference between layer 1 and layer 2 is the whole problem: 89% use something; 24% use it well. Between those two figures are sixty-five percent of Dutch small and medium-sized businesses, and little is written about them because it's not a flattering figure.

Layer 3 — Results: this is where we’re lagging behind

The studies that don’t ask what you have, but what it delivers.

Layer 4 — The reason: it’s not the technology

Ask why things are getting stuck, and you’ll hear the same answer everywhere—and it’s never about technology.

What it costs: the figures that are actually available

There’s a lot of talk about how much time is lost to searching, retyping, and asking around. More has been measured than you might think — it’s just rarely brought together.

And it’s not just about time. More than a quarter redo work that’s already been done — not because they don’t care, but because they couldn’t find that it had already been done.

The AI paradox

Another figure that is rarely compared with the previous one. Knowledge workers say they save up to eleven hours a week with AI. But only 13% see that reflected in company performance (Glean, 6,000 knowledge workers, 2026).

The explanation lies in the checking work: 42% spend more time checking AI output than the technology saves, and 49% say mediocre output delays projects (Adaptavist, 2,500 knowledge workers).

It's the same contrast as above, one layer deeper: time is saved at the individual level and lost at the company level. Put AI on shaky data, and you speed up searching while slowing down trust.

Why this matters more in the Netherlands than elsewhere

There’s a figure that’s rarely mentioned in this conversation, and it changes everything: 42% of employed people in the Netherlands work part-time. That's the highest in Europe. Germany is at 28%, Belgium at 24%, and the EU average at 18%.

Elsewhere, knowledge held in people’s heads still works to some extent because everyone is there five days a week. Not here. The question comes up on Wednesday, and the person who knows the answer is here Tuesday and Thursday. So you call someone on their day off, or wait a day and figure it out yourself.

Both take time, and neither is tracked anywhere.

The Netherlands needs a working business memory more than the countries around us — and demonstrably has it less well organized. What exactly a business memory is, has its own page.

What this means if you’re making the decision yourself

If 89% meet the basic level and 21% of AI projects deliver something, then the question isn't should we be doing something with AI too. That question has already been answered: your colleagues are already doing it, and it usually doesn’t help.

The question is: how do you get into that 21%? The figures consistently point in the same direction:

• Start with a task, not a tool. Seven to twelve separate subscriptions deliver less than one complete workflow.
• Calculate the return in advance. Not because the amount is right, but because then you can establish afterwards whether it worked. Without measurement, there's no result to measure.
• Make sure your data is in order before using AI. 18% of SMEs do something with data analysis. That’s the real ceiling.
• Expect resistance, not a need for training. Adoption is a matter of trust. People need to know what it does, why it helps them and how they can stay in control.
• Doing nothing is a valid decision. For most tasks, it’s actually the right one.

The fifth layer: what’s missing above

All figures on this page come from surveys. They ask people how far along they are, and then you get an answer to a different question: how far they think that they are. People who know their processes well rate themselves lower than those who have never given them any thought.

That's why we build a fifth layer: figures from scans where companies enter what actually happens. How many hours does the task take? Does it recur often? Is the information stored somewhere? These are questions with answers, not an estimate of your own maturity.

Take part. At the bottom of the AI scan you can participate anonymously: industry, size category and your result. Nothing else — no company name, tasks, amounts or contact details. Once there are enough responses, the figures will appear here, with sources and ranges, for everyone to use.

We publish nothing with fewer than fifty scans. Then every figure is pure chance, and that’s no use to anyone.

Sources

• CBS & Eurostat — ICT use by businesses, Digital Intensity Index, March 2026
• CBS — Digitalisation and the knowledge economy 2025, published February 2026
• European Commission — DESI report, 2025–2026 edition
• Frends / Sapio Research — State of Integration & AI 2026, 611 decision-makers from six countries, April 2026
• IDC commissioned by SAS — more than 1,600 SMEs in 28 countries, 2026
• AI Platform MKB — state of Dutch SMEs, June 2026
• Sharp — AI adoption research among SMEs, 2026

Figures from these sources are presented in our own words. Where figures differ — AI adoption ranges from 42% to 71%, depending on what is counted — the range is shown instead of the most favorable figure. Last updated: 16 August 2026.