A good AI chatbot answers your recurring questions around the clock, from your own content, and hands over to a person when things get complex. Whether that is worth it for your website comes down to the structure of your enquiries: if many of them are repetitive, a chatbot takes a large share of them off your plate. If most enquiries are one-off and complicated, a chatbot is only a limited help.
In this article we break down the difference between a rule-based bot and an AI assistant with a knowledge base, which use cases actually work, what the costs look like, and what you need to keep in mind on data protection and transparency.
Rule-based bot or AI assistant with a knowledge base?
The word "chatbot" covers two very different generations of technology. The older generation is rule-based: every possible question has a predefined path and set of buttons, e.g. "Pricing" → "Package 1" → "Details". That is predictable and low on errors, but only as good as the list of cases someone thought of in advance. Anything outside those paths falls through the cracks.
The second generation is an AI assistant with a knowledge base: a language model answers questions phrased in natural language, "How long does delivery to Switzerland take?" by assembling answers from the content you have fed it. You supply website text, FAQs, documents and product information. The main advantage: visitors can ask in their own words, and the assistant does not have to think in menus. The trade-off: language models can also make things up if the knowledge base is thin or the question strays too far from it. An AI assistant is therefore not a set-and-forget product, but a system you build properly and keep an eye on.
What use cases does a website chatbot fit?
Not every use case is equally suitable. The following areas work most reliably in practice:
- FAQ automation: The most common questions, pricing, opening hours, processes, lead times, are answered around the clock. This is the classic starting point and often the single biggest lever.
- Lead qualification: The chatbot asks a few initial questions (topic, timeline, budget) and then passes a prepared contact to your team. You end up with an informative lead instead of a vague "maybe" ticket in your inbox.
- Appointment booking: For service businesses, practices and agencies, the chatbot can offer a slot directly or link into your booking system. That saves the classic back-and-forth by email.
Beyond that, there are limits. Complex comparisons, individual quotes, complaints or anything where a human has to decide belong in human hands, the chatbot is at its best there as a door opener, not a decision-maker.
What does an AI chatbot cost?
At WebDrift we plan a chatbot as part of our AI automation. Here is the cost framework for 2026 (all prices are final prices, no VAT charged under § 19 UStG):
| Variant | One-time | Ongoing |
|---|---|---|
| AI automation with a chatbot built on your own content | from €1,500 | Maintenance from €99/month |
| Full AI agent (chatbot plus further tasks, e.g. email triage or appointment preparation) | from €3,000 | Maintenance from €99/month |
| AI provider usage fees (token and API costs) | , | assumed €20 to 150/month |
That last line matters most: the provider's price depends on actual usage, how many visitors ask how many questions. A chatbot on a site with little traffic may sit around €20/month, while a heavily used one will be well above that. As a rule of thumb, budget roughly €120 to €250 per month for maintenance plus provider fees combined if the chatbot is used regularly.
Data protection and transparency: what to keep in mind
A chatbot almost always processes personal data: at least the visitor's IP address, and their name or email address as soon as they type it in. Two points are decisive.
First: transparency for visitors. Visitors should be able to tell they are talking to an AI, for example, a clear "AI assistant" label in the chat window, and know where their data goes. That builds trust and avoids disputes later.
Second: GDPR and data processing agreements. When questions are sent to an AI provider outside your company (which is the case for most models), you need a data processing agreement (DPA). For the concrete setups, which providers offer which agreements, when a DPA is enough and when you need consent, we have written a dedicated post: ChatGPT and GDPR in your business.
There is also the question of the handover to humans: when the chatbot prepares a contact and passes it to your team, the data should be transferred and documented cleanly, otherwise you end up with a second, unclear data source.
Checklist: is a chatbot right for my website?
Before weighing costs against benefits, answer these questions honestly:
- Do many of my enquiries repeat weekly in similar form (pricing, times, processes, availability)?
- Do I already have my most common questions and answers documented in writing (FAQ, handbook, internal docs)?
- Is there a clearly defined handover point where a human takes over (quote, complaint, appointment)?
- Can I name one person who keeps the chatbot's knowledge base up to date?
- Do I have a plan for how the chatbot is labelled and which data it stores?
- Do I realistically accept that the chatbot may make mistakes and is that acceptable to my customer base?
If you can answer "yes" to at least four of these, a chatbot is very likely worthwhile for your website. If most answers are "no", a simple contact process may be the better first step before you invest in a chatbot.
Conclusion: a good chatbot is an employee, not a magic wand
An AI chatbot for your website is not a set-and-forget product, but for many small and mid-sized businesses it is a worthwhile tool, provided the enquiries are recurring, the knowledge base is clean, and the handover to humans is defined. The costs are manageable (from €1,500 one-time plus maintenance and provider fees), and the benefit is directly measurable in fewer routine enquiries.
If you are not sure whether it makes sense for your site: talk it through with us: we look at your enquiries, sketch out the knowledge base and give you an honest assessment. And if you want to see how AI can be used sensibly across your business, take a look at our post AI for small businesses.
