The short answer, with real numbers: an off-the-shelf chatbot subscription typically runs $20–$500 a month. A custom chatbot—trained on your business, connected to your booking and lead capture—runs $3,000–$10,000 to build in our pricing, plus running costs of roughly $20–$150 a month that any honest provider discloses up front.
Which one you need depends entirely on what the chatbot is for—and "because everyone has one now" is not a job. Let's price the difference properly.
Off-the-shelf vs. custom: what the money buys
| Off-the-shelf | Custom | |
|---|---|---|
| Typical cost | $20–$500/mo subscription | $3,000–$10,000 one-time, plus running costs |
| Knows your business | Generic scripts you fill in | Trained on your services, prices, and policies |
| Connects to your systems | Only what it integrates with | Your booking, your lead capture, your calendar |
| When it's stumped | Loops or dead-ends | Hands off to a human, with the conversation attached |
| Who owns it | The platform | You |
The table understates the practical difference. An off-the-shelf widget can genuinely handle "what are your hours?" A custom one can handle "do you take my insurance, and can I get in Thursday afternoon?"—which is the message that was worth money.
Where the $3,000–$10,000 goes
A custom build has four real cost centers:
Training on your business. Turning your services, pricing, policies, and the questions customers actually ask into behavior the bot gets right—and just as important, defining what it should refuse to answer.
Connections. Wiring it to your booking system and lead capture so a conversation ends in an appointment or a captured contact, not a suggestion to call during business hours. Integration effort is usually what separates the low end of the range from the high end.
The handoff. The most underrated feature: recognizing the moment a human should take over, and passing along the whole conversation so the customer never repeats themselves.
Refinement against real conversations. The first weeks of transcripts are where a decent bot becomes a good one. A build that ships and is never tuned was priced honestly but finished dishonestly.
The running costs nobody quotes
Every AI chatbot has ongoing costs: hosting, and the per-conversation cost of the AI models underneath. For small-business volume this typically lands between $20 and $150 a month. It's not a large number, but it's a permanent one—and a quote that omits it is either careless or hoping you won't ask. We put it in writing up front.
When a chatbot is the wrong buy
An honest pricing guide should include the "don't buy this" cases, so: skip the chatbot if your website barely gets visitors (fix how you get found first—a bot can't talk to people who never arrive), if your customers are overwhelmingly walk-ins and referrals who never touch the website, or if you're hoping it will paper over answers you don't actually have—a bot trained on vague policies gives vague answers, just faster.
The businesses where it pays are the ones bleeding after-hours inquiries: the practice whose phone rings at 7 PM after close, the contractor who gets form fills answered two days late, the shop whose Instagram messages pile up. For a wider look at which AI actually pays for itself, our plain-English AI guide sorts the reliable from the hype.
How to judge the payback
Ignore the technology and run the arithmetic: estimate how many inquiries per month arrive when nobody can answer—after hours, mid-job, during rushes—and what a converted inquiry is worth to you. A bot that catches a handful of missed conversations a month covers its running costs immediately; whether it covers the build cost within a year is a number you can estimate on one sheet of paper before spending anything.
That arithmetic-first approach is how we price all our AI and automation work: the projected payback has to make sense, or we'll tell you not to buy it. The consultation where we run that math with you is free—and "keep your $3,000" is a real possible outcome.