Most SME owners who search "chatbot laten maken" have already gotten at least one quote, and the numbers rarely agree with each other: one agency proposes a few thousand euro, another comes back with fifteen. The gap is not markup, it is scope. A scripted FAQ widget and a chatbot that reads your order database and hands off to a human agent are both "a chatbot", and they cost completely different amounts to build and run.
This guide breaks down the real 2026 price bands in the Dutch market, what you actually get at each level, what a chatbot can and cannot realistically do for a small or mid-sized business, and how to decide whether you need one at all.
How chatbot pricing works
There are two fundamentally different routes:
- Off-the-shelf SaaS tools. You subscribe to a platform (often built on top of an existing AI model), connect it to your website and knowledge base, and pay monthly. Low upfront cost, faster to launch, less control over exactly how it behaves.
- Custom-built chatbots. An agency or developer builds the conversation flow, integrations and, for AI-based bots, the underlying logic specifically for your business. Higher upfront cost, but it can be wired directly into your own systems (CRM, order database, booking tool) rather than working from generic training data.
Most Dutch SMEs start by asking for a custom quote, then discover partway through the conversation that a SaaS tool would have covered 80% of what they actually need.
Price bands in the Netherlands (2026)
| Chatbot type | What it does | Typical one-off cost |
|---|---|---|
| Simple FAQ chatbot | Scripted decision tree, answers a fixed list of questions, no AI | EUR 2,000 - 5,000 |
| AI chatbot with NLP and CRM integration | Understands phrasing variations, looks up real data (order status, availability) in your CRM or helpdesk | EUR 5,000 - 10,000 |
| Advanced chatbot | GPT-based, multilingual, hands off to a live agent when it cannot help | EUR 10,000 - 15,000 |
| Custom AI chatbot on your own data | Trained on your product catalogue, documentation or knowledge base, integrated into production systems | From EUR 15,000+ |
On top of the one-off build, budget EUR 50 - 500 per month for hosting, the underlying AI model or API usage, and maintenance, scaling with how much traffic the bot handles. If a fully custom build is more than your business needs right now, SaaS chatbot subscriptions with a built-in AI model start from around EUR 99 per month, with no upfront development cost.
What you actually get at each level
A scripted FAQ chatbot is a decision tree: click a button, get a pre-written answer, click another button. No language model, no understanding of free-text questions. It is cheap and predictable, and it is the right choice if your support questions are genuinely repetitive and few in number.
An AI chatbot with NLP understands the same question phrased ten different ways ("where is my order", "when does my package arrive", "has #12345 shipped yet") and, if it is connected to your systems, can pull the real answer instead of a generic one. This is where a chatbot starts doing actual work rather than just replacing a FAQ page.
An advanced, GPT-based chatbot can hold a more natural conversation, switch language mid-chat, and recognise when it is out of its depth and route the customer to a human, with the conversation history attached so nobody has to repeat themselves.
A custom bot trained on your own data goes a step further: instead of relying on general knowledge, it answers from your actual product documentation, pricing rules or internal knowledge base, and is wired directly into the systems it needs to act on (placing an order, checking stock, updating a ticket).
What a chatbot can actually do well
- Handle repetitive, well-defined questions 24/7 — opening hours, shipping status, return policy, pricing tiers — without a person having to answer the same question for the hundredth time.
- Qualify leads before a human gets involved, asking the right questions upfront so sales spends time on conversations that are actually worth having.
- Deflect first-line support tickets, freeing the helpdesk for issues that genuinely need a person.
- Work across languages for a customer base that does not exclusively contact you in Dutch.
What a chatbot cannot do
This is the part most sales pitches skip.
- It cannot replace judgement on complex or emotionally charged issues. A frustrated customer with a billing dispute or a damaged order needs a person who can make a call, not a bot working from a script.
- It is not "set and forget". A chatbot's answers are only as good as the content behind it. Product changes, price changes and new FAQs all need to be fed back in, or the bot starts giving outdated answers with total confidence.
- A GPT-based bot can hallucinate if it is not properly grounded in your actual data. Without guardrails, it will generate a plausible-sounding but wrong answer rather than saying "I don't know" — which is worse for trust than no chatbot at all.
- It will not fix a broken process underneath it. If your order status genuinely lives in three disconnected spreadsheets, a chatbot cannot magically produce a clean answer. Integration work has to happen first.
- It needs a real fallback to a human. A chatbot with no visible way to reach a person frustrates exactly the customers who needed help the most.
Recurring costs beyond the build
The one-off price is rarely the full picture. Plan for:
- Hosting and API/token usage, which scales with how many conversations the bot handles per month.
- Content maintenance: someone has to own updating the knowledge base, FAQ content and product data the bot draws from.
- Monitoring, especially for GPT-based bots, to catch wrong or off-brand answers before they become a pattern.
- Periodic reviews of what customers are actually asking, since the questions people type are rarely the ones you assumed they would.
Build vs. buy: when a SaaS tool is enough
A SaaS chatbot tool is usually enough when your questions are well-defined, your integrations are standard (a common CRM or helpdesk with an existing connector), and you do not need the bot to act deep inside custom systems. This mirrors the same logic as broader process automation: platforms cover most cases, and custom development only earns its cost once you go beyond what an off-the-shelf connector supports.
Custom development becomes the better route when the chatbot needs to read from or write to a system that has no standard integration, when the logic is genuinely business-specific (custom pricing rules, multi-step approval flows), or when you need full control over hosting and data residency, for example under NIS2-driven security requirements. Our AI & Automation team builds both, from SaaS-based chatbots wired into your existing tools to fully custom conversational systems, and our Web & Software Development team handles the integration work once a project goes beyond what a chatbot platform can do on its own.
Common mistakes
- Launching without guardrails. An AI chatbot with no boundaries on what it is allowed to claim will eventually promise something your business cannot deliver.
- No path to a human. Every chatbot needs a visible, working way to escalate.
- Treating the launch as the finish line. The content behind the bot needs an owner, or it decays within months.
- Building on a messy knowledge base. If your product information is inconsistent or out of date, the chatbot inherits that mess and repeats it at scale.
Is it worth it for an SME?
The honest answer depends on volume. If your team fields the same handful of questions dozens of times a week, even a simple FAQ chatbot pays for itself in support hours saved within months. If your support volume is low and highly varied, the build cost is harder to justify, and the time is better spent elsewhere. Before commissioning anything, pull a month of your actual support tickets or chat logs and see how many follow a genuinely repeatable pattern. That number, more than any vendor pitch, tells you which tier of chatbot, if any, makes sense.
FAQ
What does it cost to build a chatbot? In the Netherlands, a simple scripted FAQ chatbot runs EUR 2,000 - 5,000, an AI chatbot with CRM integration EUR 5,000 - 10,000, and an advanced GPT-based bot with multilingual support and live handover EUR 10,000 - 15,000. Fully custom bots trained on your own data start from around EUR 15,000, plus EUR 50 - 500 per month in ongoing hosting and API costs.
Can a chatbot replace my customer service team? No. A chatbot handles repetitive, well-defined questions well, but complex, emotional or exceptional cases still need a person, and every chatbot needs a working handover to a human.
Do I need AI, or is a simple FAQ page enough? If your questions are few and genuinely repetitive, a static FAQ page or a scripted decision-tree chatbot is often enough. AI earns its cost once questions vary in phrasing or need to pull real, personal data like an order status.
How long does it take to build a chatbot? A simple scripted chatbot can launch in days to a few weeks. An AI chatbot with real system integrations typically takes several weeks to a few months, depending on how many systems it needs to connect to and how much guardrail and testing work is required before launch.
Talk to us before you commission one
Not sure which tier of chatbot your business actually needs, or whether the questions your customers ask even justify one? Explore our AI & Automation service, or read our guide on automating business processes if a chatbot is one of several manual processes you are looking to cut down.