If you've been looking into AI chatbots for your business, you've probably noticed that nobody gives you a straight answer on pricing. You get vague "contact us for a quote" forms or pricing pages that bury the real number behind enterprise tiers. So let me be direct: the cost of an AI chatbot for a small business in 2026 typically ranges from £50/month for a basic off-the-shelf setup to £5,000–£15,000 for a fully custom build. Here's what actually drives that difference — and how to figure out what makes sense for your situation.
The Two Fundamentally Different Things You Can Buy
There's a meaningful distinction between renting a chatbot platform and having one built for you. Platforms like Tidio, Intercom, or Crisp give you a configurable chatbot that you set up yourself, usually starting at £50–£200/month depending on the plan. These work well if your use case is fairly standard — answering FAQs, collecting lead info, routing visitors to the right page.
A custom-built chatbot, on the other hand, is something an agency or developer builds specifically around your workflows. It might connect to your booking system, pull from your internal knowledge base, or handle multi-step processes like insurance quotes or service scheduling. That's where the £5,000–£15,000 upfront figure (plus monthly maintenance) comes in. Neither option is inherently better — it depends on what problem you're actually trying to solve.
What Drives the Cost Up
The biggest cost drivers are integrations and complexity. A chatbot that just answers "what are your opening hours?" costs almost nothing to set up. A chatbot that checks appointment availability in real time, books a slot, and sends a confirmation email requires connecting to at least three separate systems — and each integration takes time to build and maintain.
Volume also matters. A restaurant getting 200 website visitors a week has very different infrastructure needs than a law firm processing 2,000 enquiries a month. Most platforms charge per conversation or per seat, so your costs scale with usage. It's worth modelling that out before you commit to anything.
The third factor is AI quality. Basic rule-based chatbots — which follow decision trees rather than understanding language — are cheap to run but brittle. They break as soon as a customer asks something slightly off-script. Chatbots powered by large language models (like GPT-4 or Claude) are more flexible but cost more to run per query. For most small businesses, you don't need the most expensive model, but you do need something smarter than a flowchart.
Realistic Price Brackets by Business Type
A dental practice wanting to handle appointment requests and FAQ answers outside office hours can typically get this working for £80–£150/month using an existing platform with some light configuration. No custom development needed.
A property management company that wants a chatbot to qualify tenant enquiries, pull availability from a property database, and hand off to a human agent is looking at a custom build — probably £6,000–£10,000 to set up, with ongoing maintenance around £300–£600/month.
A consultancy that wants to capture leads, ask qualifying questions, and book discovery calls sits somewhere in the middle — £2,000–£4,000 to build something solid, or £150–£300/month if they use a platform like Calendly combined with an AI front-end.
Hidden Costs Nobody Mentions
The upfront build or subscription is rarely the full picture. Training and maintenance are real ongoing costs. If your services change, your chatbot needs updating — and that's either your time or someone else's. Most businesses underestimate how much time goes into keeping the chatbot accurate.
There's also the cost of getting it wrong. A chatbot that gives incorrect information about pricing, availability, or legal matters can damage customer trust fast. That's not an argument against using one — it's an argument for scoping it properly from the start and being honest about what it can and can't handle.
When an AI Chatbot Actually Pays for Itself
The clearest ROI cases are high-volume, repetitive enquiries outside business hours. If your team spends two hours a day answering the same ten questions, a chatbot handling those frees up real time. If 30% of your website visitors leave without contacting you because they can't get a quick answer at 9pm, a chatbot changes that.
The cases where it doesn't pay off: very low enquiry volume (fewer than 50 contacts/week), highly complex or sensitive conversations that always need a human, and businesses where customers actively prefer phone contact over chat.
What to Ask Before Spending Anything
Before you talk to any vendor, answer these three questions for yourself: What specific problem am I solving — lead capture, support volume, after-hours coverage? How many conversations per month do I realistically expect? And what systems does the chatbot need to connect to? Those answers will tell you whether you need a platform subscription or a custom build, and give you enough to get a realistic quote.
An AI chatbot isn't a magic traffic driver or a replacement for good service. But if you have a genuine volume problem and a clear use case, it's one of the more straightforward automation investments a small business can make right now.