Does Chatbot Cost Money? Free Plans, Prices, and Real Costs

Does chatbot cost money? Sometimes: business chatbots can be free, subscription-based, usage-priced, or custom-built. Published monthly prices range from $0 to more than $10,000, depending on the product, usage, and scope. The amount you actually pay depends on what the vendor bills for, what the plan includes, and the work needed to keep the bot useful.

A free plan can help you test common questions. A paid hosted plan usually has a recurring fee, while a custom build can add setup and maintenance costs. Before choosing, find out whether the price is per agent, conversation, resolution, reply, bot, or workspace. Those units behave differently when traffic or staffing changes.

Does a chatbot cost money? Typical monthly prices, from free to enterprise

Some business chatbots are free to start, while paid hosted plans can cost tens or hundreds of dollars a month. Custom work can add a substantial one-time bill and ongoing maintenance. There is no single typical price that applies to every chatbot: costs depend on scope and the unit used to calculate the bill.

One market estimate puts monthly chatbot costs from $0 to more than $10,000. Another estimates $29–$119 a month for many small businesses, $119–$400 for mid-size companies, and more than $500 for some enterprise setups. Treat those as broad reference points, not a quote for your workload. A small FAQ bot and a system with several integrations are different projects.

A hosted subscription is usually the easier cost to spot: it recurs monthly or yearly. But the subscription may include only a certain number of conversations, replies, queries, or seats. If the team exceeds that allowance, the bill may change. Some plans also charge for an AI feature on top of the helpdesk seats or platform fee.

A custom chatbot has a different cost shape. The build may be a one-time project, but the bot still needs care after launch. One estimate places a basic custom rule-based build at $5,000–$15,000 or more, and an advanced AI system at $20,000–$80,000 or higher. Another estimates basic maintenance at $1,000–$5,000 a year. These are broad estimates with different scopes, not a standard price list.

For a fair comparison, separate the costs into:

The recurring platform or subscription fee.
Any usage charges above what the plan includes.
Agent or team seats, if priced separately.
Setup, content preparation, or integration work.
Ongoing time spent checking answers, updating information, and fixing issues.
External services or infrastructure that the project requires.

If you are weighing a custom build against a hosted service, compare both the first bill and the ongoing work. A custom build can make sense when the business needs particular logic or integrations. A hosted tool may be simpler when the job is answering common questions and handing uncertain cases to a person. For more detail on build budgets, see this guide to chatbot build costs.

Which chatbot pricing model makes your bill rise?

A chatbot bill rises according to its billing unit. Per-seat plans grow with the team, per-conversation and per-reply plans grow with usage, and per-resolution plans grow when the bot succeeds under the vendor’s definition. A flat monthly bundle can make the fee more predictable, but only while actual usage stays within the included allowance.

Per-seat pricing charges for people who use the platform. The bill can rise when you add agents, administrators, or supervisors, even if customer conversation volume stays steady. This may be easier to forecast when staffing is stable, but it is worth checking how seats are counted and whether different roles have different prices.

Per-conversation pricing charges for conversations the bot handles, whether or not it solves the issue. If traffic increases during a sale, seasonal rush, or service disruption, the bill can rise with it. Ask how the provider treats an abandoned conversation, repeat visits, and a customer who starts a new chat about the same issue.

Per-resolution pricing charges when the bot resolves a case according to a defined rule. That can make the relationship between cost and successful automation easy to understand, but the definition matters. Ask whether a customer who asks several questions in one chat counts as one resolution, whether a handoff counts, and how the vendor decides a case is resolved.

Per-reply pricing ties the bill to the number of AI responses rather than the number of chats. A long conversation can use more replies than a simple FAQ exchange. A reply allowance is not the same as a conversation allowance, so estimate how many replies your typical conversation needs before comparing plans.

Per-bot pricing charges for each bot or agent configuration. Workspace pricing charges at the account or workspace level. Those models can make the bill less sensitive to the number of conversations, but usage limits, seats, and add-ons may still apply. Ask what the stated fee includes rather than assuming that a flat headline price means unlimited use.

For any usage-based model, ask these questions before committing:

What exactly counts as a billable conversation, reply, query, or resolution?
Are abandoned chats or conversations that end in a human handoff billed?
What happens when the included allowance is used up?
Is there an overage fee, or does the bot pause until the next billing period?
Can the team see usage before it reaches the limit?
Do you need a separate paid module for a feature the team depends on?

A vendor may define a resolution as the AI directly solving the customer’s question. In one example, multiple questions in the same chat still count as one resolution. Another provider may define the event differently. Do not compare a price per resolution with a price per conversation as if they were the same unit.

What drives chatbot cost beyond the plan price?

The plan fee is only one part of the cost. Estimate how many people will use the bot, how often they contact support, how long conversations run, and what portion of the work the bot is expected to handle. Then add setup and upkeep time, any usage charges, and the cost of extra services the implementation needs.

Start with actual support conversations rather than a vendor’s example. Look at your recent inbox and sort the questions into common, repeatable topics and cases that need a person. For an online store, common topics might include delivery timing, return steps, or product details. The bot needs reliable content for those answers; it should not be asked to invent policy or promise a refund.

Next, find the plan’s billable unit and allowance. If a plan includes replies, count replies. If it includes conversations, count conversations under the provider’s definition. A conversation may contain multiple AI replies, so a reply cap can cover fewer conversations than a conversation cap. The number of staff members who need access also matters if the plan charges per seat.

Then account for content preparation. Existing FAQs, policy pages, product information, and help articles may need review before a chatbot can answer from them reliably. Someone has to find conflicting details, remove outdated instructions, and make sure important exceptions are clear. That work can take staff time even when the platform itself is quick to configure.

Integration work can add cost too. A bot that only answers from approved support content has a narrower setup than a project that needs custom logic or data connections. Ask which integrations are included, whether setup is a separate project, and whether ongoing maintenance is billed. A connector or external service can have its own fee.

Finally, budget time for upkeep. A person should review cases where the bot was unsure, check whether answers still match current policies, and update knowledge when products or processes change. This is not a reason to avoid automation; it is part of running customer support. If no one owns the content, the bot can keep drawing on information that the team no longer wants customers to use.

A practical cost worksheet can include:

Cost item	What to check
Plan fee	Monthly or annual charge, currency, and billing period
Included usage	Conversations, replies, queries, or resolutions included
Seats	Which team roles count as paid users
Overage	Extra usage price, pause behavior, or upgrade requirement
Setup	Content review, configuration, training, and launch work
Integrations	Included connectors, custom work, and third-party charges
Maintenance	Staff time for content updates, review, and troubleshooting

For a written quote, ask the vendor to show the price at your current volume and at a higher-volume month. Include the assumptions behind each estimate. A quote that only shows the base plan fee will not tell you what happens when usage rises or the team adds seats.

Compare chatbot tiers by included usage, seats, and features

Compare plans by their billable unit, monthly allowance, seats, and included support features—not by price alone. The examples below show how different providers package usage. They are not a complete market survey, and one row’s unit should not be treated as equivalent to another row’s.

Plan example	Price	Included usage	Seats or team access	Billing detail
ChatBot.com Essential	$19 per user/month billed annually	AI chatbot; listed resolution allowance is 10	Per user	Annual price; monthly billing is listed at $25 per user
CustomGPT Standard	$99/month, or $89/month billed annually	1,000 queries/month	3 team members	10 agents
CustomGPT Premium	$499/month, or $449/month billed annually	5,000 queries/month	5 team members	25 agents
momo Free	EUR 0	30 AI conversations, one time	1 agent and 1 team member	No per-resolution fee
momo Starter	EUR 29/month or EUR 319/year	500 AI conversations/month	1 agent and 2 team members	Flat plan fee; no credit packs
momo Growth	EUR 129/month or EUR 1,419/year	2,000 AI conversations/month	3 agents and 5 team members	Flat plan fee; no credit packs
momo Scale	EUR 399/month or EUR 4,389/year	6,000 AI conversations/month	10 agents and 10 team members	Flat plan fee; no credit packs

The table illustrates why the unit matters. A plan billed per user can change when the support team grows. A plan billed per conversation can change with traffic. A query, reply, conversation, and resolution are different things, so compare the allowance to your own usage only after you understand the provider’s definitions.

These plans are priced in EUR and include a fixed number of AI conversations. Paid plans allow up to 25 AI replies in a conversation; Free allows up to 10. The Free allowance is one-time, not monthly. Every plan includes the helpdesk inbox, and the team can take over a conversation. This can suit a support team that wants an AI front line but needs a person to handle cases the bot cannot answer confidently.

The AI retrieves passages from the business’s own knowledge, drafts an answer, and checks that draft against those sources. If it is confident, it responds with citations; if not, it says it is not sure and opens a ticket. The team can save a human answer as approved knowledge. This does not remove the need to maintain the underlying content, but it gives the team a way to handle unanswered questions and improve the knowledge base.

Plan features differ as well as allowances. The Free plan includes a website widget, basic analytics, and knowledge from up to 10 sources, with a product badge on the widget. Starter adds email, a Help Center, and more knowledge capacity. Growth adds a Slack channel, API, Zapier, webhooks, and advanced analytics. Scale includes the same listed integration types as Growth, with higher knowledge and usage limits. Shopify is in preview; WhatsApp, Instagram, and Messenger are marked coming soon. Do not budget around a channel or connection until it is available for your use.

Yearly billing is 11 times the monthly price. If your team compares annual billing, compare the amount paid upfront and the included monthly usage rather than treating the annual figure as an extra allowance. To consider a different helpdesk bill separately, see this Zendesk cost guide.

Worked example: estimate a small business’s monthly chatbot bill

A useful estimate starts with the same usage unit across every option, then adds setup and maintenance. For a hypothetical online retailer with 1,200 conversations a month and two to three AI replies per conversation, usage would be 2,400–3,600 replies. A flat-bundle example puts that volume within a Growth plan at $99 a month.

That example shows why the unit changes the estimate. The retailer should not compare 1,200 conversations directly with a plan that includes replies without converting conversations into replies. If the average conversation is longer, the reply requirement rises. If many chats go straight to a person, actual AI usage may be lower, depending on how the provider counts the interaction.

Now compare the same workload with a hypothetical per-resolution fee. If a provider charged $1.00 for each resolution and the bot resolved 900 of the 1,200 conversations, the usage charge would be $900 before any platform fee. This is an illustration, not an actual vendor rate. It is not a like-for-like quote until you confirm the provider’s definition of resolution and what happens to handed-off or abandoned conversations.

To build a more realistic estimate for your own store:

Count recent support conversations. Use your actual inbox volume and identify recurring topics that are suitable for an answer from approved content.
Estimate how many chats the bot will handle. Do not assume every visitor starts a chat or that every chat should be automated.
Estimate replies per conversation. Separate short FAQ exchanges from longer cases that need clarification.
Translate usage into the plan’s unit. Convert to replies, queries, or resolutions only when the vendor’s definition is clear.
Add team access. Include the seats needed for the people who monitor and answer handoffs.
Add setup and upkeep. Estimate staff time for content review, configuration, testing, and ongoing updates.
Check higher-volume months. Ask the vendor to calculate the bill at your current volume, a month with fifty percent more usage, and a peak month.
Include add-ons and overages. Add any required modules, connectors, or usage charges to the base fee.

The estimate should include more than the software invoice. If the plan is $99 a month but the team spends time preparing content and checking answers, that time is part of the project’s cost. If a plan includes a set allowance and no overage is triggered, the plan fee may be predictable; if usage goes beyond the allowance, find out whether the bot stops, the plan changes, or extra charges apply.

Do not assume that a higher automation rate automatically makes one model cheaper. With per-resolution pricing, more successful resolutions can increase the usage bill. With a flat bundle, the fee stays within the plan’s terms until the allowance or another limit is reached. Compare each quote at your current volume and at a higher-volume month, using the same assumptions about conversations and outcomes.

When is a free chatbot plan enough—and what triggers an upgrade?

A free chatbot plan can be enough to test whether customers ask questions the bot can answer from your content. It is a practical starting point when the allowance, channels, team access, and knowledge limits fit the test. Upgrade when measured usage or a required capability exceeds those limits, not simply because a paid plan sounds more complete.

Before using a free plan in customer support, check whether its allowance renews monthly or is one-time. Then confirm whether the limit counts messages, conversations, replies, or queries. Those details determine how many real customer interactions the plan can cover. Also check what happens at the limit: the bot may stop answering, or the plan may require an upgrade.

The next check is whether the free plan supports the way your team works. A website widget may be enough for an FAQ test, but a team that needs email or another channel should verify that the plan includes it. Check the number of agents and team members, the available analytics, and the amount of knowledge you can add. Do not count an integration as available unless the plan includes it today.

A free plan is a sensible fit when the team wants to test content quality, see which questions recur, and observe which cases still need a person. It is less suitable when the allowance is too small to represent normal support, the team needs more seats, or the available channels do not match customer behavior. A short test that misses peak demand can make a plan look cheaper than it will be in regular use.

The free tier includes 30 AI conversations as a one-time allowance, one agent, one team member, a website widget, basic analytics, and a helpdesk inbox. It also includes up to 2 MB of knowledge in up to 10 sources, and the widget carries a product badge. That can support an initial evaluation, but it is not a recurring monthly conversation allowance.

Frequently asked questions

Is a free trial the same as a permanently free chatbot plan?

No. A free trial gives temporary access, while a permanently free plan remains available under its stated limits. For example, ChatBot.com offers a free 14-day trial and does not have a permanent free tier. Check what happens when a trial ends, whether the bot stops, and what the paid plan costs before you build a workflow around it.

Is per-resolution pricing cheaper than paying per conversation?

It depends on the price and the provider’s definitions. Per-resolution billing charges for outcomes the vendor counts as resolved; per-conversation billing charges for conversations the bot handles, even if they end in a handoff. Compare both models using your own volume, expected resolutions, and any platform fee. Make sure abandoned conversations and repeat questions are treated consistently in the estimate.

How do I estimate conversations from a chatbot’s reply allowance?

Use your support history to estimate how many AI replies a typical conversation needs, then divide the reply allowance by that estimate. One published example says 1,000 replies covers roughly 333–500 conversations. Treat that as an estimate, not a promise: short questions use fewer replies than conversations that need clarification or several answers.

What costs should I ask a chatbot vendor to include in a written quote?

Ask for the plan fee, included usage, seat charges, overage terms, required add-ons, setup, integration work, and any external service fees. Ask the vendor to define the billable unit and show the estimate at your current volume and a higher-volume month. Include the staff time your team expects to spend preparing and maintaining content.

How much can a custom chatbot cost to build and maintain?

Published estimates vary substantially because projects do not share a common scope. One estimate puts a basic rule-based build at $5,000–$15,000 or more and an advanced AI build at $20,000–$80,000 or higher. Another estimates basic maintenance at $1,000–$5,000 a year. Ask for a scoped proposal that separates the build, integrations, hosting, and ongoing maintenance.

Try a support workflow before committing

If a free test fits your needs, start with the questions your team answers repeatedly and check how uncertain cases reach a person. The Free plan includes a website widget and helpdesk inbox, so you can try momo free.

Does Chatbot Cost Money? Monthly Prices and Real Costs