Most vendors will not put a number on their pricing page. Instead you get “contact sales”, and the real figure only appears after two calls.
This guide gives you the actual ranges, explains the three pricing models, and shows you the costs that never make it into a quote. By the end you will know what to budget and which questions expose a bad deal.
What ai chatbot pricing actually includes
A quote usually bundles four things, and vendors weight them differently.
The platform itself. The software that runs the agent. This is the headline number.
Usage. Conversations, messages, or minutes. Because this scales with your traffic, it is where surprises live.
Integrations. Connecting your CRM, calendar or help desk. Some vendors include this, others charge per connector.
Support and setup. Onboarding, training, and whether a human helps you build it.
When you compare two quotes, split them into these four parts first. Otherwise you are comparing a platform fee against a bundle and the cheaper one will look better than it is.
The three pricing models explained
Per seat
You pay for each human agent using the tool. This model comes from the help desk world, where the bot assists people rather than replacing work.
It suits support teams with stable headcount. However, it punishes you for growing the team and it does not reward automation, since your bill does not fall when the bot resolves more.
Per conversation
You pay for each chat, usually in bundles. Simple to understand and easy to forecast at steady traffic.
The risk is a spike. A campaign, a product launch, or a bad news day multiplies your conversations and your bill at the same time. Therefore always ask what happens when you exceed the bundle.
Per resolution
You pay only when the agent completes the job without a human. This is the newest model and it is spreading fast.
It aligns the vendor with your outcome, which is genuinely better. Nevertheless, read the definition of “resolution” carefully. Some vendors count a conversation the visitor abandoned as resolved.
Real price ranges by company size
These are the ranges businesses actually pay, not list prices.
| Business size | Typical monthly range | What you get |
|---|---|---|
| Small (under 1,000 chats) | Low hundreds | One channel, standard integrations, self serve setup |
| Mid market | Several hundred to low thousands | Multiple channels, CRM integration, some support |
| Enterprise | Thousands and up | Custom integrations, SLAs, dedicated support, security review |
Two things move you up a tier faster than headcount. High conversation volume is the obvious one. Less obviously, requiring a security review or custom integration work usually pushes you into enterprise pricing regardless of size.
Why the same product quotes differently
Two companies of identical size often get very different numbers from the same vendor. That is not arbitrary.
Volume is the first factor. Channels are the second, because voice and WhatsApp typically cost more to run than web chat.
Contract length is the third and it moves the number more than most buyers expect. An annual commitment frequently cuts the monthly rate meaningfully.
Finally, timing matters. Quarter end quotes are softer than mid quarter ones, which is unglamorous but true across the software industry.
How ai chatbot pricing compares to the alternative
The useful comparison is not against zero. It is against what you spend now to answer the same questions.
A part time person answering chat and phone costs a salary plus the hours nobody covers. Meanwhile an outsourced answering service charges per minute and rarely books anything into your calendar.
Set the quote against those two numbers rather than against nothing. Most businesses find the software cheaper on volume and the human better on complex conversations.
Because of that, the pattern that usually wins is a mix. Automate the repetitive majority and keep people for the calls that need judgment.
Hidden costs nobody quotes you
These are the ones that turn a good ai chatbot pricing deal into an expensive one.
Setup and content work. The agent needs answers. Somebody has to write them, and that somebody is usually your team. Budget real hours, not a token afternoon.
Integration engineering. If the platform is not no code, every connection costs developer time. Consequently a cheap licence can carry an expensive implementation.
Overage rates. Ask what a conversation costs once you pass the bundle. The overage rate is frequently several times the bundled rate.
Channel add ons. Voice, WhatsApp and SMS are often priced separately from web chat. A quote covering only web chat is not covering your actual need.
Maintenance. Answers go stale when pricing or policy changes. Somebody owns keeping them current, and that is an ongoing cost rather than a one off.
Migration. If you switch vendors later, your prompts and flows rarely move across. Ask what you can export before you sign, not after.
A quick sanity check on any quote
Take the annual figure and divide it by twelve. Then ask whether you would hire someone at that monthly rate to do the same job.
If the answer is an easy yes, the ai chatbot pricing is fine. If it is close, the deciding factor is volume, because software absorbs a spike and a person does not.
That framing cuts through most negotiations faster than a feature comparison does.
The question that reveals the real number
Ask this: “What did a customer like us actually pay in year one, including setup?”
A vendor who answers plainly is worth talking to. One who redirects to the monthly licence fee is hiding the implementation cost, and that cost is usually the bigger half.
How to work out your own cost per conversation
Do this before you talk to anyone. It takes ten minutes and it changes how you negotiate.
Count your monthly inbound conversations across chat, phone and messaging. Include the ones you currently miss, because those are the ones the agent is meant to catch.
Then work out what one conversation is worth. Take your average deal value and multiply by your close rate from an inbound enquiry. That is your ceiling.
Finally, divide the annual quote by the annual conversations. Compare the two numbers.
Most businesses find the maths obvious in one direction or the other within a few minutes. If it is close, the deciding factor is usually how many conversations you are missing now, since those are pure upside.
How Dapta prices this
Dapta is no code, so the implementation cost that dominates most quotes largely disappears. You describe the agent in plain language rather than paying for engineering.
The same agent covers chat, voice and WhatsApp, which removes the per channel add ons that inflate other quotes. It also handles English and Spanish including regional accents, so a bilingual market does not mean a second build.
You can start free and see the real behaviour before committing, which is the fastest way past the quote problem entirely.
For the full picture, explore our related guides:
- What an AI chatbot for business does and how to set it up
- How much an AI receptionist costs in 2026
- What an AI answering service really costs
- The 12 best AI chatbots for business, compared
Frequently asked questions
How much does an AI chatbot cost per month?
For a small business, expect low hundreds monthly for a single channel with standard integrations. Mid market lands in the high hundreds to low thousands. The licence is rarely the largest cost in year one, though, since setup and content work usually exceed it.
Is per resolution pricing better than per conversation?
Usually yes, because you only pay when the agent actually does the job. However, the definition matters. Ask exactly what counts as a resolution and whether an abandoned chat is included.
Why will vendors not publish ai chatbot pricing?
Because usage varies enormously between customers, and because a published number invites comparison. Some publish anyway. Treat transparent pricing as a positive signal rather than assuming the hidden ones are cheaper.
What is the cheapest way to start?
A free tier or trial on a no code platform. You get real conversations with real visitors before spending anything, and that tells you more than any demo.
Do I pay extra for WhatsApp or voice?
Often yes, and this is the most common gap between a quote and reality. Confirm which channels are included before comparing two prices.
What ongoing costs should I budget for?
Content maintenance and someone’s time reviewing transcripts. Neither is large, but both are real. Teams that budget zero for upkeep are the ones whose chatbot quietly gets worse.
The honest summary on ai chatbot pricing is that the licence is the easy part. The implementation and the upkeep decide whether it was a good deal.
