How Much Does an AI Chatbot Cost a Small Business?

The message usually arrives with a screenshot attached. A business owner has been quoted one price for an artificial intelligence (AI) chatbot, the first invoice has landed at a very different number, and the question underneath is always the same: what did I actually sign up for?

I get asked about chatbot pricing more than almost any other tooling question, and the confusion is rarely about the headline price, the one part every vendor prints in large type. What decides the real bill is the meter running underneath it: the unit the platform counts before it charges you. Learn to read the meter, and every pricing page becomes easier to judge.

A website chatbot is the simplest member of a wider family of software that answers customers and follows up on its own. I have covered that family in my guide to AI agents for small business; this piece stays with the chat window itself, and with the four ways vendors charge for it.

The four ways AI chatbot pricing actually works

Every plan I come across reduces to one of four meters, and each behaves differently as your chat volume grows.

Per resolution: paying only when the AI finishes the job

Under this model you pay when the AI agent reaches an outcome: it answers the question fully, completes a set procedure such as checking an order, or hands the conversation over. Intercom’s pricing guide shows the purest version, because its Fin AI Agent charges per outcome on top of a per-seat plan. Zendesk’s pricing page instead includes a small resolution allowance with each seat and charges per resolution beyond it. You pay for finished work, and the bill grows with the AI agent’s success.

Per conversation: paying for every thread, whoever closes it

Here the unit is the conversation itself. Tidio’s pricing page defines a billable conversation as one in which a human agent sends a message, while its Lyro AI Agent is sold in packs of AI-handled conversations, each counted once however many messages the thread contains. Counting last month’s threads forecasts this month’s bill fairly. Its weakness hides in the definitions, because platforms define a conversation very differently, and the definition moves the bill.

Per seat: paying for the people, not the volume

The oldest model in support software charges for each team member who logs in. Intercom’s base plans and Zendesk’s suites both sit on seats, with the AI layered on top. A seat price is indifferent to volume, so a busy week costs the same as a quiet one. The bill moves when your team does: a part-time helper can cost the same seat as the person answering all day, and every human who takes a handoff needs a seat.

Flat pricing: one fee for the workspace, whatever the month brings

Flat pricing charges a single fee for the workspace and lets volume do what it wants. Crisp’s pricing page is the cleanest example: one monthly price per workspace, seats included, and its Hugo AI Agent drawing on included AI credits. When the credits run out, automated conversations are billed per conversation, so the model is flat with a metered edge rather than flat without limits.

AI chatbot prices compared: what the main platforms charge

The table below records each platform’s entry point and the event it actually bills for. Every price names its billing variant, because vendors show different figures for monthly and annual billing and sometimes test new prices, so the figure facing you may differ slightly.

The same month of customer questions, priced four ways

Entry prices hide the real story, so here is one ordinary month priced four ways. Picture two people answering customers, a chat window busy all month, and a healthy share of questions simple enough for the AI agent to close on its own. The exact assumptions sit inside the table, where the figures belong.

The business, the customers and the questions are identical in every row, and only the meter changed. Two owners can compare invoices, find completely different totals, and both be describing exactly what a pricing page promised.

Which AI chatbot pricing model fits which small business

Before you sign, count last month’s conversations, ask in writing what counts as a billable event, and price your busiest month rather than your average one.

• Per resolution suits a business drowning in simple, repetitive questions about deliveries, opening hours and returns. Its risk arrives with success and with the seasons, because a great month for the AI is an expensive month.
• Per conversation suits a shop or service business with a steady, countable flow of chats. Read the billable definition first, and check whether the plan raises your tier automatically when you cross your allowance.
• Per seat suits a team where people handle most conversations and the AI drafts, summarises and covers overflow. Budget for every person who might take a handoff, because the meter counts logins rather than effort.
• Flat pricing suits the owner who values a predictable bill over a theoretically cheaper one. Keep an eye on the included AI credits as volume grows, because heavy automation eats them first.

The chatbot costs that never appear on the pricing page

The meter is only the first layer. The moment the AI cannot answer, a person takes over, and on seat-based plans that person needs a paid seat, so the true price includes seats for everyone who might ever take a handoff. The tools that help your humans reply faster are separate per-person add-ons on the larger platforms, and one can rival the price of the seat itself. Removing the vendor’s name from your chat window is another paid extra on lower tiers.

Smaller print catches smaller budgets. Some free and entry AI allowances are granted once rather than renewed monthly, which makes the second month dearer than the first on identical usage. Some platforms also raise your plan automatically when you cross your allowance, billing the busy month before anyone approves the spend. The last hidden cost is your own time: the agent answers from your help pages and policies, and tidying that material is part of the real price.

Frequently asked questions

How much does an AI chatbot cost a small business each month?

The monthly cost depends on the meter underneath the plan. A quiet month on a flat or per-conversation plan can stay close to the entry prices in the table above, while a busy month on a per-resolution plan climbs quickly, because every question the AI closes adds to the bill.

What is the difference between per-resolution and per-conversation pricing?

Per-conversation pricing charges for each conversation the platform counts, whether the AI closes it or a person steps in. Per-resolution pricing charges only when the AI reaches an outcome, such as resolving the question or completing a handoff, so the two meters price the same month very differently.

Is a free AI chatbot plan enough for a small business?

A free plan is enough to learn how your customers actually talk to a chatbot, because free tiers include a small monthly allowance. It is rarely enough to run support on, since allowances run out quickly and, on some platforms, the free AI allowance is a one-time grant rather than a monthly one.

Why can an AI chatbot bill jump in a busy month?

The bill jumps when the meter is tied to volume, because metered plans charge more in your busiest months. Automatic tier upgrades add to the jump, and resolutions bought pay as you go cost more than resolutions committed in advance, so the busiest month is often billed at the least favourable rates.

Which AI chatbot pricing model suits a business whose chat volume keeps growing?

A flat per-workspace plan usually keeps the bill steadiest as volume grows, because the fee stays fixed while the included AI allowance absorbs more conversations. Per-resolution pricing grows with every success the AI has, which makes it the model to watch most carefully at higher volumes.