All articles
Growth

What an AI Chatbot Actually Costs — And What Drives the Bill

Seat pricing, conversation pricing and per-resolution pricing behave very differently as you grow. Here is how to work out your real monthly cost before you commit.

Jennox Team9 min read
Business team planning a customer engagement project
Image from Freepik

The short answer

  • Most AI chatbot tools price on one of three models: per seat, per conversation, or per resolution. Which one is cheaper depends entirely on your traffic shape, not on the headline number.
  • Seat pricing punishes growing teams. Conversation pricing punishes high-traffic sites. Resolution pricing looks fairest and is the hardest to forecast.
  • The advertised price is rarely the bill. Onboarding fees, AI credit packs, integration tiers and annual-only discounts are where budgets actually break.
  • Work out cost per resolved conversation, not cost per month. It is the only number that stays comparable across vendors.

Why chatbot pricing is so hard to compare

Two vendors can quote the same monthly figure and cost you wildly different amounts a year later. One charges for the people who log in, another for the conversations visitors start, and a third only when the AI resolves something without a human. Those are not three prices for the same thing. They are three different bets on how your business will grow.

The confusion is not accidental. Pricing pages are written to make the first month look cheap, because the first month is what people compare. The bill that matters is month fourteen, after a marketing campaign tripled your traffic or you hired three more support agents.

So before comparing numbers, work out which model each vendor is actually using, and then run your own traffic through it.

The three pricing models, and who each one punishes

Almost every tool in this category uses one of three approaches. Each is reasonable from the vendor's side. Each has a customer profile it quietly penalises.

  • Per seat — you pay for each agent who logs in. Predictable while your team is stable, but it taxes the exact thing you want to do, which is add people as you grow. Teams respond by sharing logins, which breaks your audit trail and your routing.
  • Per conversation — you pay for each chat started. This scales with traffic rather than headcount, which suits small teams on busy sites, until a seasonal spike or a viral post arrives and the bill follows it.
  • Per resolution — you pay only when the AI closes something without human help. It sounds like the fairest model and it often is, but forecasting is genuinely hard, and the definition of a resolution is set by the vendor, not by you.
  • Flat plan with limits — a fixed monthly fee including a set number of conversations. Easiest to budget for, provided you understand what happens when you exceed the allowance.

Read what happens when you exceed the limit

This single clause separates a predictable bill from a nasty one, and it is usually three levels deep in the documentation rather than on the pricing page.

There are broadly three behaviours. The tool stops answering until the next billing cycle, which protects your budget and damages your customer experience at precisely the busiest moment. It auto-upgrades you to the next tier, which protects the experience and can double the bill without a conversation. Or it charges overage per unit, which is usually the most expensive per conversation but at least degrades gracefully.

None of these is wrong. But you should know which one you have bought before a Black Friday spike tells you.

Jennox uses fixed monthly plans with clear conversation allowances, so you know the ceiling before traffic tests it.

See what each plan includes

The costs that are not on the pricing page

The subscription is usually the smaller half of the first-year number. Budget for these before you sign anything.

  • Onboarding or implementation fees, common on enterprise tiers and often non-negotiable once you are in procurement.
  • AI credit packs sold separately from the plan, so your usage has two meters running rather than one.
  • Integrations gated behind a higher tier. A CRM sync you assumed was included frequently is not.
  • Annual-only discounting, where the advertised monthly price is not available monthly.
  • Removing vendor branding from the widget, which is a paid upgrade in a surprising number of tools.
  • The internal time cost of maintaining content, which is real whether or not anyone invoices you for it.

Work out your cost per resolved conversation

This is the only figure that stays comparable when one vendor charges per seat and another per conversation. It also reframes the decision usefully: you are not buying software, you are buying resolved customer questions at a unit price.

  1. 1Count the conversations you handled last month. If you have no chat tool yet, use inbound support emails plus contact form submissions as a rough proxy.
  2. 2Estimate the share the AI could plausibly resolve alone. Be conservative — 40 to 60 percent is a realistic starting band for a well-trained bot on ordinary support questions.
  3. 3Add the vendor's monthly subscription to any credit packs, integration tiers and amortised onboarding fee.
  4. 4Divide that total by the number of conversations you expect the AI to resolve. That is your true unit cost.
  5. 5Repeat at three times your current traffic. The ranking between vendors often flips, and that is the number you are really committing to.

What a realistic budget looks like

For a small business or a lean team, expect a serious tool to sit in the tens of dollars per month rather than single digits, and be sceptical of anything free at meaningful volume — the cost has simply moved somewhere less visible, usually your data or your branding.

Mid-market teams with several agents, CRM sync and booking generally land in the low hundreds per month. Enterprise deals with security review, SSO and custom integration are a different conversation entirely and are rarely priced publicly.

Prices in this category move quickly, so treat any figure you read anywhere, including here, as a starting point to verify on the vendor's own pricing page on the day you buy.

The question to ask before you buy

Ask the vendor to model your actual numbers, not their example customer's. Give them your monthly conversation volume, your team size, and your expected growth, and ask for the total twelve-month cost including every add-on you would need to run the setup you have described.

A vendor who answers that clearly is a vendor whose bill will not surprise you. A vendor who cannot, or who keeps returning to the headline monthly figure, has told you something useful too.

Frequently asked questions

How much does an AI chatbot cost per month?

For a small business, a capable AI chatbot typically costs between roughly $30 and $100 per month. Mid-market setups with CRM integration, multiple agents and booking usually run from $100 to $500 per month. Enterprise pricing with security review and custom integrations is normally quoted individually. The wide range reflects pricing model rather than quality — per-seat, per-conversation and per-resolution pricing produce very different bills for the same usage.

Are free AI chatbots worth using?

Free tiers are genuinely useful for testing whether a chatbot fits your workflow at all. They are rarely viable long term at real traffic, because the limits that matter — conversation volume, branding removal, integrations — are exactly what free tiers restrict. Treat a free plan as an evaluation period rather than a destination.

Is per-conversation or per-seat pricing better?

Per-conversation pricing usually favours small teams handling high website traffic, because your cost tracks demand rather than headcount. Per-seat pricing usually favours larger support teams with steady, predictable volume, because adding conversations costs nothing. Model both against your own numbers at current and triple traffic — the cheaper option frequently reverses as you grow.

What hidden costs should I expect?

The most common are onboarding or implementation fees, AI credit packs billed separately from the subscription, integrations locked behind higher tiers, annual-only discount pricing, and a charge to remove vendor branding from the widget. Ask for a total twelve-month cost including every add-on required for your intended setup.

How do I calculate the ROI of an AI chatbot?

Compare the fully loaded monthly cost against the support hours saved plus the additional revenue from leads captured outside business hours. If your chatbot resolves 300 conversations a month that would each have taken an agent ten minutes, that is roughly 50 hours returned to the team. Value those hours at your loaded staff cost and compare against the subscription.