Chatbot Pricing Model: Costs, Tiers, and How to Compare

The chatbot pricing model you choose determines whether your bill follows a plan tier, the number of agents, chatbot activity, or successful outcomes. Published plans range from free to over $1,000 per month, while some vendors charge by usage or offer custom pricing; the actual cost depends on what the plan includes and how it counts activity.

To compare options fairly, first translate each vendor’s billing unit into the support work you expect to handle. Then add required seats, channels, integrations, setup, and ongoing upkeep. A low entry price can still be a poor fit if the included allowance is small or the bill rises sharply when usage grows.

Chatbot pricing model: what you pay, and the broad price range

Chatbot plans range from free to over $1,000 per month, with subscription, usage-based, and per-agent models among the common ways vendors charge. These are broad reference points, not a quote for a particular tool: plan limits and the definition of a billable conversation vary.

A simple scripted chatbot and an AI support desk do different work, so compare them against the job you need done. The basic question is not only “What does the bot cost?” but also “What support work is included, and what remains for my team?”

For a store, that might mean answering recurring shipping, return, and product questions, then routing unclear cases to a person. For a support lead, it could mean handling common questions from company documentation while keeping a shared inbox for issues that need judgment. The right plan depends on those workflows and the amount of human follow-up they create.

Some published comparisons put AI-handled conversation rates at roughly $0.20 to $2.00, before any helpdesk seats. Treat that as an example range from published offers, not a market-wide rate: vendors define “AI-handled” differently, and an included platform subscription may also apply. See the guide to AI chatbot costs for a separate look at cost components.

A useful first pass is to collect the same details from each option:

• Monthly subscription and billing period.
• Included conversations, sessions, messages, resolutions, or credits.
• Number of support seats and bots or agents included.
• Channels and integrations needed for your workflow.
• What happens when you reach the included allowance.
• Any setup work and ongoing administration your team must do.

If a plan advertises an attractive base fee but does not match your normal traffic or team size, it is not yet a meaningful comparison. Use the billing unit and limits to estimate your own monthly workload.

Five chatbot pricing models—and what triggers the bill

The main pricing models charge for a fixed subscription, human-agent seats, chatbot conversations or sessions, completed resolutions, or a pool of credits. Hybrid plans combine a subscription with a variable usage charge, so the bill can have both a predictable base and a usage-sensitive part.

Flat subscription. You pay a recurring fee for a plan tier. The tier may include a set number of conversations, seats, features, or channels. The benefit is a more predictable base bill; the risk is outgrowing the included limits or needing a higher tier for a feature you rely on.

Per-agent or per-seat. The price increases with the number of people who need access. This can suit a team with stable headcount, but the bill may rise when the team expands even if the chatbot resolves more questions. Check whether a seat means every teammate, only agents who answer customers, or another defined role.

Per-conversation or per-session. The charge follows customer activity. This can be straightforward when traffic is steady, but a promotion, product issue, or seasonal rush can increase usage. Ask how a new conversation is counted, whether a returning customer starts another billable unit, and what happens after the allowance runs out.

A conversation and a session are not necessarily interchangeable. A session may be defined by a period of interaction, while a conversation could be counted as a thread or another unit. Do not use a vendor’s “conversation” estimate to predict a competitor’s session bill until you have checked the definitions.

Per-resolution. The charge is triggered when the vendor considers an issue resolved by AI. The key question is what qualifies as a resolution. Ask whether a customer leaving the chat, receiving an answer, or confirming that a problem is solved counts. Multiple questions in one chat may also be treated as one or several outcomes, depending on the vendor’s definition.

Credits. A plan gives you a pool that may be consumed by messages, AI replies, steps, or tokens. One customer conversation can use different amounts of credits depending on how the platform meters it. Ask for a worked estimate using your typical conversation, what model affects consumption, whether unused credits carry over, and what happens when the pool is exhausted.

Hybrid pricing adds a variable usage charge to a subscription. The subscription may cover the platform and an allowance, with additional usage billed separately. Read both parts together: a low base fee does not tell you what the plan costs at your expected volume.

For every model, write down the billable event, included amount, renewal or rollover rules, overage behavior, and billing period. If those details are unclear, you cannot reliably compare the offer with another plan. For an additional view of related usage models, see AI agent pricing models.

What each tier typically includes—and what published prices show

Entry tiers may limit conversations, customization, or integrations, while higher business and enterprise tiers can include more usage, analytics, and customization. Published examples vary by vendor, so use the following broad tiers as orientation and treat individual plan details as separate offers.

Tier	Broad published price indication	Typical inclusions or limits
Basic	Free to about $100 per month	Limited conversations, basic visual customization, and minimal integrations may be included.
Business	About $100 to $1,000 per month	May add AI capabilities, higher conversation limits, standard integrations, and analytics.
Enterprise	$1,000 per month or custom pricing	May be tailored around larger volumes, integrations, deployment needs, or support requirements.

These ranges do not mean every product follows the same ladder. A free tier can be a useful way to test a workflow, but it may not cover the volume, channels, or team access needed for everyday use. Likewise, the label “enterprise” does not tell you which features or services are included; compare the actual plan terms.

One published example from ChatBot.com lists Essential at $19 per user per month when billed annually, or $25 per user per month when billed monthly. Its Growth plan is listed at $79 per month and includes a stated allowance of AI Agent resolutions. Enterprise pricing is custom. These figures describe that vendor’s stated offer; they are not a general benchmark for other chatbots.

When reviewing any tier, check what “included” means in practice. A plan can include AI while still limiting the number of conversations, resolutions, agents, channels, or connected tools. A shared inbox may also be separate from the chatbot itself. Map every included item to an actual task your team handles.

For a small store, a constrained entry tier may be enough to answer common questions and see which topics still need a human. A larger support team may care more about team access, analytics, channels, or integrations than about the headline number of conversations. The plan should fit both the work and the people responsible for it.

The cost drivers that change your monthly bill

The largest cost drivers are expected usage, support seats, channels, integrations, and how the vendor meters AI activity. Estimate volume and successful resolution separately: a plan can charge for every interaction, while another may bill only for outcomes it defines as resolved.

Start with inbound customer conversations, not a guess at “AI volume.” Separate the total number of conversations from the portion you expect the AI to handle. Then separate the AI-handled portion from the share that reaches a satisfactory resolution without a person. Those are different measures and can affect usage-based and outcome-based plans differently.

Next, count the team members who need access. Include people who will answer escalations or review conversations, not only those who configure the bot. If seats are billed individually, more teammates can change the monthly cost. If seats are included in a tier, note the limit and the price or tier change needed when the team grows.

List the channels customers actually use and the integrations the workflow requires. A website widget alone may not cover a team that needs email or another channel. Connecting a chatbot to a helpdesk, CRM, or other system can add vendor charges or internal engineering work. Do not assume an integration is included just because a plan has a low starting price.

For credit-based plans, ask how the selected model and reply length affect consumption. A pool measured in tokens is not directly comparable with a pool measured in messages or actions. Ask the vendor to map your real conversation pattern to expected credits, including cases where the bot gives several replies before escalating.

Finally, consider customization. Extra languages, tailored analytics, or more complex integrations may require a higher tier or additional work. Decide which requirements are essential at launch and which can wait. Otherwise, it is easy to pay for capabilities before your team has a clear use for them.

Hidden costs: setup, overages, upkeep, and human time

The subscription is only one part of the operating cost. Setup, content preparation, integrations, staff training, ongoing knowledge updates, migration, and human handling of escalations can all take time or require additional spend.

Before launch, someone has to decide what the bot should answer and prepare the content it needs. If product details, shipping rules, and return policies are scattered or out of date, cleaning them up becomes part of the project. That work still matters even when the chatbot platform itself is quick to configure.

Integration engineering can also change the total. A simple website deployment may need little work, while connecting multiple systems or tailoring the workflow can take developer time. Ask what the vendor includes in setup, what your team must handle, and whether a connector or onboarding service has a separate charge.

Overages deserve particular attention. Once you use the included allowance, a vendor may charge for additional activity, require a plan change, or handle the limit in another way. Ask which applies before signing, whether overage billing is automatic, and whether you can set a spending limit or usage alert. Do not assume the bill stops at the plan fee.

Ongoing maintenance is a real operating task. Policies change, products change, and old answers can become misleading. Assign someone to review and update the material that informs the bot. Also decide who checks unanswered questions and who takes over conversations that the AI cannot confidently answer.

Include human time in the comparison. A system that handles common questions but creates difficult handoffs may shift work rather than remove it. Review a sample of escalations: does the person receive the customer’s details and context they need, or must they ask the same questions again? Poor handoffs can cost time even if they do not appear as a separate invoice line.

Migration matters too. If you change platforms later, content, workflows, and integrations may need to be rebuilt or adapted. Ask what can be exported and what would need to be recreated before the team depends on the system.

A useful total-cost formula is: monthly platform fee + usage + required integrations + allocated implementation + knowledge maintenance + administration and quality checks. Include staff time even when it is not a vendor charge. For another perspective on the non-subscription parts of a chatbot budget, see chatbot cost components.

Estimate your monthly bill with a workload-based example

Estimate a bill by recording monthly inbound conversations, expected AI-handled volume, resolution share, seats, and required channels, then adding the relevant fees and operating costs. A worked example is only useful when its assumptions are visible and its billing unit matches the vendor’s unit.

Use a worksheet like this before requesting quotes:

Input	Your estimate
Total inbound customer conversations	Use your own recent support volume.
Expected AI-handled conversations	Estimate the portion the chatbot can answer from approved content.
Expected AI resolutions	Estimate separately from AI-handled conversations.
Support seats	Count everyone who needs access to answer or review conversations.
Required channels	List only the channels your team needs to support.
Required integrations	Name the systems the chatbot must connect to.
Included usage and overage	Record the plan allowance and what happens when it is exceeded.
Setup and maintenance	Estimate the work your team or a provider will need to do.

For a hypothetical comparison, suppose your shop expects 1,000 chatbot conversations in a month. That figure is an example workload, not a claimed average. One vendor may charge for all 1,000 conversations; another may charge for sessions, credits, or only resolutions. You cannot calculate comparable totals until you know each vendor’s unit and the relevant prices.

For a per-resolution offer, you would need the number of billable resolutions and the stated price per resolution, plus any base subscription and seat charges. For a conversation plan, you would need to know whether the allowance covers the expected 1,000 conversations and what extra usage costs. For a seat-based plan, you would need the number of required seats and any separate usage fees.

For a flat subscription with a defined conversation allowance, check whether the example workload fits the plan. If it does not, calculate the next tier or usage charge rather than comparing the lower tier’s base fee. If the plan also includes an inbox and the team needs one, count that as part of the relevant offer, not as an assumed extra purchase.

Then add operating costs. If the team spends time preparing the content, testing answers, reviewing escalations, or maintaining integrations, include an estimated allocation for those tasks. Separate one-time setup from recurring work so a one-off implementation does not look like a permanent monthly charge—or disappear from the decision altogether.

The example should result in a total monthly cost and a cost per quality-approved resolution if the data allows it. To calculate the latter, divide total cost by the number of resolutions your team has reviewed and accepted as correct. Do not use a vendor’s resolution count as a proxy for quality unless your team has checked the definition and reviewed actual outcomes.

Compare chatbot tiers without confusing units

Normalize every vendor against the same workload, team seats, channels, billing period, and assumptions about AI handling. Then compare total monthly cost and the cost per quality-approved resolution, rather than choosing by headline price alone.

Build a side-by-side worksheet with a row for each tool and columns for:

• Base plan fee and whether billing is monthly or annual.
• Included usage and the exact unit it measures.
• Required seats and channels.
• Integrations needed and any associated costs.
• Overage price or what happens when usage is exhausted.
• Setup, maintenance, and administration.
• Total cost under your expected workload.
• Resolutions your team has reviewed and accepted as correct.

Do not mix annual and monthly prices without converting them to the same period. Do not compare a per-user fee directly with a per-resolution fee. And do not compare credits with conversations until you know how many credits your typical conversation consumes.

Ask each vendor how it handles an abandoned chat, a conversation that contains several questions, a handoff to a person, and a customer who returns later. These cases reveal whether the unit matches the work you want to measure. For a resolution-based model, ask how the vendor defines a resolution and how your team can verify it.

Test the workflow as well as the bill. Prepare questions that the bot should answer from your own content, plus questions it should not answer because the information is missing or a human decision is required. Check whether answers are grounded in the content you provide, whether citations or references are available if relevant, and what the visitor sees when the system cannot answer confidently.

For a support team, review the handoff end to end. Check which customer details are collected, what reaches the inbox, and whether a teammate can continue the conversation without making the customer start over. For a store, use examples involving returns, delivery estimates, and product fit. Pay particular attention to policy-sensitive answers: an incorrect refund promise is more expensive than a chatbot that admits it is unsure and routes the case to a person.

momo is one option for teams that prefer a flat plan fee with included AI conversation limits. It has no per-resolution fee or credit packs, and its helpdesk inbox is included on every plan. The plans are priced in euros: Free is EUR 0; Starter is EUR 29 monthly or EUR 319 yearly; Growth is EUR 129 monthly or EUR 1,419 yearly; Scale is EUR 399 monthly or EUR 4,389 yearly. Paid plans include 500, 2,000, and 6,000 AI conversations per month respectively; the Free allowance is 30 AI conversations one time. Each AI conversation can contain up to 25 AI replies on paid plans and up to 10 on Free.

That structure may be easier to budget for a team whose volume fits the included allowance, but it is not automatically the right fit for every workflow. Check the number of agents and teammates, available channels, knowledge limits, and the content you need to load. The AI answers from your business’s own content, checks its draft against retrieved sources, and cites sources when confident; when it is not confident, it says it is not sure and opens a ticket for the team.

If a flat plan is under consideration, estimate whether your expected traffic fits its monthly allowance and what your team needs to do when it does not. For other pricing approaches, use the same workload worksheet and compare the real operating cost rather than relying on a vendor’s starting price.

Frequently asked questions

Does AI chat cost money, or can a business start with a free plan?

Some chatbot products offer a free tier, while others offer a paid subscription or trial. Check whether a free option is ongoing or one-time, how much usage it includes, and whether it limits seats, channels, or knowledge sources. A free plan can help you test a workflow, but compare its limits with the support volume you actually need to handle.

How much do bots cost for 1,000 chatbot conversations each month?

There is no single price for 1,000 conversations because vendors may charge by subscription, conversation, seat, resolution, or credits. Published examples for AI-handled conversations range from about $0.20 to $2.00 per conversation before helpdesk seats, but that is not a universal rate. Calculate the bill using the vendor’s actual unit, included allowance, subscription, and required seats.

How do per-session and per-resolution chatbot charges differ?

A per-session charge is based on a defined interaction period or session. A per-resolution charge applies when the vendor counts an issue as solved by AI. Ask how each unit is defined, including what happens when a customer abandons a chat, returns later, asks several questions, or is handed to a human.

What should I ask about chatbot overage fees before choosing a plan?

Ask what happens when you use the included allowance: whether extra activity is billed automatically, triggers an upgrade, or stops being available. Request the overage rate and clarify how the vendor counts the activity. Also ask whether you can monitor usage and set an alert or spending limit.

How much do 1,000 tokens cost, and how do AI credits affect conversations?

There is no single cost for 1,000 tokens across chatbot plans. Credits may be consumed by messages, replies, steps, or tokens, so a credit allowance does not translate directly into a fixed number of conversations. Ask how a typical conversation uses credits with the model you plan to use, whether unused credits carry over, and what happens when the pool runs out.

Which costs should I include beyond the chatbot subscription?

Include required seats, usage, channels, integrations, setup, content preparation, staff training, ongoing knowledge maintenance, administration, and quality review. Also account for human time spent on escalations. Separate one-time work from recurring costs so you can compare the full operating cost.

Choose a model your team can explain

A workable chatbot pricing model makes the bill understandable at your expected volume and ties the charge to a unit your team can measure. Compare the same workload across options, check what happens at the plan limit, and test the handoff before relying on the bot for customer support. If you want to evaluate a flat-fee option, try momo free.