95% of Dutch organizations say they use AI. Five percent actually get value from it [3].
Read that again. Ninety-nine out of a hundred companies say "yes, we're doing something with AI." And then, when you dig deeper, most of them turn out to have a ChatGPT subscription that three people occasionally open to rewrite an email. That's not applying AI in your business. That's having a tool that happens to contain AI.
We've worked with enough SMEs by now to recognize the pattern. The companies that actually get results do three things the rest skip: they start with a specific process (not a tool), they measure what it delivers from day one, and they work with someone who knows which approach fits. That last part sounds like a sales pitch, but honestly: the numbers are there. 84% of failed AI projects fail due to organizational reasons, not technical ones [4]. The technology works fine. It's the implementation that trips companies up.
The timing is right for SMEs. In the Netherlands, where 84% of SMEs plan to invest more in AI within three years — the highest percentage in Europe [2] — and 81% already operate in the cloud [2], the infrastructure is ready. And the results from companies that do take the step? 88% report higher revenue, 87% lower costs [5]. Those aren't projections. Those are reported figures.
Why AI is now within reach for SMEs
Two years ago, this was a different story. AI was something for Unilever, ING, the big players with entire IT departments and budgets that SME owners don't need to think about. That's no longer true.
The cost of AI models dropped by more than 90% in 2025. Let that sink in. Ninety percent. Tools that used to require weeks of implementation time now run in days. Harvard Business School showed in a controlled study that employees with AI complete tasks 25% faster, with 40% higher quality [8]. McKinsey measures 5.7 hours of savings per week per person among their own consultants [9].
Let's do the math. Twenty employees, all saving 5.7 hours per week. That's 114 hours. Nearly three full-time positions worth of capacity freed up without hiring anyone.
The Netherlands sits in an unusual position. The ambition is there, the infrastructure is in place, but adoption lags behind. Only 8 to 13% of companies with 10 to 50 employees actually use AI [7]. For companies with 500+ employees, that's 48% [7]. This gap can't be logically explained by budget or access — the tools are affordable and the cloud is already running. It comes down to knowledge and guidance. AI adoption among smaller businesses is growing at 72% per year [1], so it's accelerating fast. The question is whether you're leading or catching up.
Government support is helping too. Through the Netherlands AI Coalition, €276 million has been invested in AI adoption [7]. The EU AI Act takes full effect on August 2, 2026, and gives SMEs priority access to regulatory sandboxes and proportional fines [12]. A small nuance: most SME applications fall into the low-risk category, so the practical impact on your day-to-day operations is limited. But good to know.
Start with your business, not the technology
The most common mistake? "We need to do something with AI."
Sounds proactive. It's not. It's the equivalent of "we need to do something with social media" in 2014. It leads to a tool nobody uses and a feeling of disappointment that AI "isn't all it's cracked up to be."
The companies that successfully apply AI in their business start with a different question. Which process costs us the most time? Where do we make the most mistakes? What costs us the most money without anyone ever really looking at it?
This doesn't have to be a big analytical project, by the way. Two questions. That's all.
Where's the repetition? Every organization has it. Entering invoices, forwarding emails, updating customer records, assembling quotes, running reports. The kind of work nobody wakes up thinking "yes, that's what I'm excited about today." Employees spend an average of 62% of their workday on this type of task. Two-thirds of their time. In a previous analysis of business processes, we described how to map this out systematically — but honestly, most business owners know exactly which processes these are. They just never looked at them seriously.
And where are the errors? Manual work is error-prone. Every time someone retypes data from one system to another, there's a chance of mistakes — and they compound. Koninklijke Dekker, a 140-year-old Dutch timber company, had exactly this problem. Their order processing ran on Excel files, PDFs, and emails that the sales team processed manually. After automating order intake: better data quality, fewer errors, a sales team focused on client relationships instead of administration [7]. No rocket science. Just a process that hadn't kept pace with how the business grew.
The point isn't finding the perfect AI project. Perfect projects don't exist. It's about knowing where your productivity leaks, so you can choose a solution that addresses something that actually hurts.
Choose your first project wisely
Your first AI project determines whether your organization gets excited or gives up. That sounds dramatic, but we see it happen again and again. A company that starts with "we're going to make all our customer communication AI-driven" and ends up empty-handed after four months, versus a company that starts with invoice processing and has concrete numbers after six weeks.
What should you look for?
Volume. The more often the process runs, the faster you see results. A task that occurs every day simply delivers more than something that comes up monthly. Which daily tasks lend themselves best to automation varies by industry, but the patterns are surprisingly similar.
Complexity — or rather, the lack of it. The first project should be rule-based. If X, then Y. Not a process with lots of exceptions or one that requires strategic judgment. That comes later, once your team has built experience and confidence in the approach.
And measurability. Not "improve our communication" but "reduce invoice processing time by 60%." Companies that define concrete KPIs upfront succeed 2.4 times more often [4]. It sounds so logical you wonder why everyone doesn't do it. Yet the majority skips this step.
Good first projects? Email triage and automated responses, invoice processing, lead qualification, inventory management. Processes with high volume, clear rules, directly measurable results.
Where AI delivers the fastest results
Not all AI applications are equal. Some show measurable difference within weeks. Others take months before you notice anything. In our experience, these are the areas where SMEs see returns fastest.
Customer service and communication
AI-powered classification and automated responses are the first project with visible results for many companies. It makes sense: the volume is there, the questions repeat, and customers notice immediately that they're helped faster.
A solo consultant automated his lead qualification and scheduling and booked 40% more qualified appointments within three months [16]. No extra staff. No extra hours. AI qualifies incoming leads based on criteria you set yourself, schedules appointments automatically, and your team spends their time on conversations that matter instead of sorting emails.
Administration and document processing
Invoices, orders, reports. Processes that come back every day and where accuracy is everything. This is the terrain where AI makes the difference between two workdays and two hours. 62% of SMEs running their first AI tool see major productivity improvements within six months [13].
An EdTech company automated their complete onboarding: contract generation, account creation, appointment scheduling. The HR team now saves 2 to 3 hours per new employee [16]. Sounds modest. But with ten new hires per month, that's half a working week freed up for work that actually matters. (And honestly, if you've ever spent an afternoon manually creating accounts and sending welcome emails, you know how fast that "this has to be smarter" feeling builds up.)
Sales and marketing
AI-driven product recommendations, churn prediction, automated follow-ups: no longer reserved for the big players. An e-commerce company deployed an AI recommendation engine and saw 15% higher order values plus 12% better customer retention within six weeks [15]. Another company used AI for churn prediction and automatic reactivation emails. Result: 15% less customer churn within six months, 10% higher customer lifetime value [16].
Content and knowledge work
This is where the numbers get really interesting. A content marketing agency doubled their output from 80 to 160 articles per month and saved more than 85 hours — without extra people [15]. Research shows that AI triples productivity on roughly a third of all tasks, particularly in research, content creation, and analysis [8].
The pattern? The fastest results are where volume and repetition converge. How to calculate what that automation concretely saves you — we've done the math before, including specific amounts per process type.
What the leaders do differently
88% higher revenue [5]. 87% lower costs [5]. But there's a group of companies that gets significantly more out of AI than the rest. What are they doing that others aren't?
They define success upfront
Most companies throw an AI tool at something and "see if it works." Sound familiar? Companies that set concrete KPIs upfront (response time, error rate, hours per process, cost per transaction) succeed 2.4 times more often [4]. Sounds logical, right? And yet the majority has no measurable objective when they start. They know they "want to do something with AI." But not what success actually looks like.
They redesign their work processes
McKinsey discovered something striking. AI saves an average of 5.7 hours per employee per week, but only 1.7 of those hours are productively reallocated [9]. The rest evaporates into longer breaks, extra meetings, tasks that don't matter.
Honestly, we found that the most surprising figure in the entire study. You save nearly six hours, but four of them disappear as if they never existed. Companies that get the most out of AI don't just think about what work AI takes over — they also plan what their team will do with the freed-up time. That's where the real gains are, and almost nobody plans for it deliberately.
They keep it simple at the start
30% of generative AI projects stalled after the proof of concept, according to Gartner [6]. The cause? Not the technology. Too broad ambitions, unclear goals, poor data quality. Companies that do push through start with a pilot of four to eight weeks on one process. They measure. They only scale after that. Boring? Maybe. But it works.
They choose the right partner
And this is the point where it really tips. 84% of AI projects that don't deliver the desired results failed due to organizational factors, not technical ones [4]. The tool works fine. But the implementation — the integration with existing systems, training the team, redesigning processes — that's the part where expertise makes the difference between a pilot that fizzles out and a system that delivers structural value. Companies with ongoing guidance and sponsorship succeed 4.1 times more often [4].
From pilot to structural results
Your first AI project works. The team is excited. Now what?
Scaling up isn't "putting the same tool on more processes." It means AI becomes part of how your business operates — integrated with your CRM, your accounting software, your communication tools. That sounds big. It doesn't have to be.
McKinsey now runs 25,000 personalized AI agents alongside 40,000 employees, saving 1.5 million hours per year on search and analysis work [14]. That's an extreme example, especially for SMEs. But the principle applies at any scale: for them too, it didn't start with 25,000 agents. It started with one application that worked, and a plan to scale from there.
After a successful pilot, you pick two to three adjacent processes that run on the same data or systems. You integrate them so they function as a whole rather than isolated islands. And you make it stick: monitoring, documentation, someone on the team who knows when the system needs human attention. How to approach this per business function depends on where you can make the most impact with the least friction.
The investment and what it delivers
Now the part everyone actually wants to start with but nobody dares to ask about first.
Companies deploying AI achieve an average 5.8x return on investment in the first year [11]. Payback period for well-chosen processes: three to six months [11]. 30% of companies see revenue growth of more than 10% [5]. And 53% name improved productivity as the biggest impact [5].
What does it cost? Standard AI solutions start at €20 to €100 per month per user. Custom solutions that fully align with your processes and systems start at €20,000 [10]. That's a wide range, and it is. The right choice depends on how unique your processes are and where your competitive advantage lies. Standard tools work for standard things. But if your order processing or customer communication is truly what sets your business apart, you don't want a one-size-fits-all solution.
And then there's the cost of doing nothing. People forget about that one. Every month your team manually processes invoices, fails to follow up on leads, or leaves customers waiting is a month your competitor has already automated that. Adoption is growing at 72% per year among smaller businesses [1]. Those who wait now will be chasing a gap that gets wider every month.
Look. Applying AI in your business isn't an enormous leap. One well-chosen process. Four to eight weeks of piloting. Measure what it delivers. And then scale to what makes sense. 62% of SMEs that follow this path see major productivity improvements within six months [13]. The question is no longer whether it works. That's been answered. The question is when you start — and whether you get the approach right when you do.
Ready to apply AI in your business — the right way?
Nexaton helps SMEs choose, implement, and scale AI solutions that fit their processes and systems. From first pilot to structural results. Get in touch →
Sources
[1] OECD, "AI Adoption by Small and Medium-Sized Enterprises", https://www.oecd.org/en/publications/ai-adoption-by-small-and-medium-sized-enterprises_426399c1-en.html
[2] Wolters Kluwer, "Dutch SMEs Are Leading the Way in Europe in Terms of AI Ambitions and Cloud Infrastructure", https://www.wolterskluwer.com/en/news/dutch-smes-are-leading-the-way-in-europe-in-terms-of-ai-ambitions-and-cloud-infrastructure
[3] First AI Movers, "95% AI Adoption, 5% Value Creation in Dutch SMEs", https://www.firstaimovers.com/p/ai-adoption-netherlands-sme-2026
[4] Pertama Partners, "AI Project Failure Statistics 2026", https://www.pertamapartners.com/insights/ai-project-failure-statistics-2026
[5] NVIDIA, "State of AI Report 2026", https://blogs.nvidia.com/blog/state-of-ai-report-2026/
[6] Gartner, "30 Percent of Generative AI Projects Will Be Abandoned After Proof of Concept", https://www.gartner.com/en/newsroom/press-releases/2024-07-29-gartner-predicts-30-percent-of-generative-ai-projects-will-be-abandoned-after-proof-of-concept-by-end-of-2025
[7] Lleverage, "AI Automation in the Netherlands", https://www.lleverage.ai/blog/ai-automation-in-the-netherlands-how-dutch-businesses-are-leading-europes-automation-revolution
[8] Autofaceless, "AI Productivity Statistics 2026", https://autofaceless.ai/blog/ai-productivity-statistics-2026
[9] Fortune/McKinsey, "The AI Time Dividend", https://fortune.com/2026/02/27/erik-roth-mckinsey-ai-time-dividend-how-leaders-ceos-can-automate-more-work/
[10] SUCCESS, "The Real Cost of AI Tools for Small Business", https://www.success.com/the-real-cost-of-ai-tools-for-small-business-roi-calculator-2
[11] Versalence, "Small Business AI ROI Guide 2026", https://blogs.versalence.ai/small-business-ai-roi-guide-2026
[12] Harvard Business Review, "How SMEs Can Prepare for the EU's AI Regulations", https://hbr.org/2025/09/how-smes-can-prepare-for-the-eus-ai-regulations
[13] AIvenSoft, "AI for SMEs — Automation and Productivity Gains", https://www.aivensoft.com/en/blog/ai-sme-automation
[14] ByteIota, "McKinsey's Own AI Transformation", https://byteiota.com/mckinsey-25k-ai-agents-40k-by-2026-25-workforce-cut/
[15] DoneForYou, "Case Study: Small Businesses Winning with AI Tools", https://doneforyou.com/case-study-small-businesses-winning-ai-tools-2025/
[16] ActivDev, "Artificial Intelligence for SMEs: Case Studies & Examples", https://www.activdev.com/en/artificial-intelligence-for-smes-case-studies-examples/
