How much does a chatbot cost? For a public website chatbot, monthly chatbot cost often starts around $40 to $150 at low volume, then reaches hundreds or thousands of dollars as usage, integrations, and support needs grow. AI chatbot cost is not just the subscription: it includes messages, conversations, credits, paid outcomes, seats, add-ons, setup, and ongoing staff time. (Sources: Chatbase Pricing, SiteGPT Pricing, Botpress Pricing, Intercom Pricing)

The cheapest plan can cost more over time when it has a small quota, expensive overages, paid branding removal, or a usage unit that grows faster than your customer conversations. The table turns 100, 1,000, and 5,000 monthly conversations into planning figures using public list prices checked on September 21, 2026.

Key takeaways

  • A small website chatbot often starts around $40 to $150 per month in software, before staff time and add-ons. (Sources: Chatbase Pricing, SiteGPT Pricing)
  • Website chatbot pricing is only comparable after you define the unit. Four messages are not automatically one conversation, one credit, or one paid outcome.
  • At higher volume, a flat message plan can be cheaper than a low entry plan with costly overages.
  • Setup and maintenance belong in your budget even when the vendor advertises a self-serve chatbot.
  • AI chatbot pricing should be compared as an operating cost, not a sticker price.

Want to launch a branded, customer-facing agent without starting with a custom build? Explore SketricGen Brand Agents, then compare its current limits, integrations, handoff options, and maintenance needs against the same budget model.

The seven costs behind website chatbot pricing

A website chatbot budget has seven layers.

  1. Platform or workspace fee. This is the base subscription for the chatbot builder, widget, inbox, analytics, or helpdesk. Chatbase lists Hobby at $40 per month, Standard at $150, and Pro at $500. SiteGPT lists Starter at $59 per month, Growth at $129, and Scale at $429 on monthly billing. (Sources: Chatbase Pricing, SiteGPT Pricing)
  2. Customer usage. The plan may include messages, conversations, credits, or outcomes. SiteGPT includes up to 4,000 messages on Starter, while Botpress Plus includes 250 conversations and sells additional packs of 100. Those are different units and should remain separate in your worksheet. (Sources: SiteGPT Pricing, Botpress Pricing)
  3. Model or AI spend. Some platforms hide model cost inside the plan. Others meter it separately. Botpress describes AI Spend as the cost of calling an LLM and includes $0.10 of AI usage per conversation on its paid plans. A custom build may expose token charges directly, but the model bill is still only one part of operating the bot. (Source: Botpress Pricing)
  4. Knowledge and content upkeep. Website pages, catalogs, policies, PDFs, and FAQs need to stay current. Check crawling, file storage, page limits, automatic refresh, and correction tools. SiteGPT separates page capacity and refresh behavior by tier, so source volume can force a higher plan. (Source: SiteGPT Pricing)
  5. Integrations and tool calls. A bot that answers questions is cheaper to scope than one that checks orders, books appointments, creates CRM records, or calls an API. Ask whether each action consumes credits or AI spend. Chatbase exposes AI Actions by plan, while SiteGPT places API access on higher tiers. (Sources: Chatbase Pricing, SiteGPT Pricing)
  6. Branding and channels. Removing vendor branding, adding WhatsApp or SMS, enabling voice, or running multiple websites can change the bill. Chatbase charges $99 per month to remove its branding. Tidio shows no-branding as a $16.67 monthly add-on in its plan comparison, and its Lyro usage is priced separately from its customer-service platform. (Sources: Chatbase Pricing, Tidio Pricing)
  7. Setup, review, and maintenance time. Someone must define scope, clean source content, test answers, review gaps, update policies, and handle escalations. A self-serve interface reduces friction, but it does not eliminate the work. Treat staff time as a cost line even if nobody invoices it.

If your goal is a branded agent that serves website visitors rather than an internal copilot, see SketricGen Brand Agents. Compare its current billing rule, included usage, integrations, handoff, and content-maintenance needs before making a price claim.

How AI chatbot pricing units differ

“Chat” is not a billing standard. The definition determines how quickly your budget grows.

Conversations

A conversation is usually a thread or customer interaction. Botpress counts an exchange with at least two end-user messages, and a conversation spanning two months counts in both months. Tidio defines a Lyro conversation as an interaction with at least one AI reply. The same visitor behavior can therefore consume one unit on one platform and no unit on another. (Sources: Botpress Pricing, Tidio Pricing)

Messages

A message is one unit sent or received, depending on the vendor's rules. SiteGPT's quota is message-based and varies by model: its Starter plan allows up to 4,000 GPT-4.1-mini messages or 400 GPT-4.1 messages, with a mix calculated by its pricing tool. Four messages per conversation is a planning assumption, not a vendor definition. (Source: SiteGPT Pricing)

Credits

Credits are platform-defined units. Chatbase calls its allowance message credits, with 700 on Hobby, 4,000 on Standard, and 15,000 on Pro. Do not assume one customer conversation equals four Chatbase credits until the dashboard confirms how user turns, assistant replies, actions, and model choices are counted. (Source: Chatbase Pricing)

Tool calls

A tool call is an action taken outside the chat, such as a CRM lookup, calendar search, payment check, or ticket creation. It may be included, limited by tier, or metered separately from the conversation. Ask for the price and failure behavior of each action before promising automation.

Seats

A seat is a human teammate with access to the workspace or inbox. Intercom Essential starts at $29 per full seat per month, then adds $0.99 per Fin outcome. A bot may need one seat while a support team needs five. (Source: Intercom Pricing)

Add-ons

Add-ons cover the features buyers discover after the initial trial: extra messages, extra agents, branding removal, storage, analytics, channels, multilingual support, and premium onboarding. An independent comparison from Chatref also notes that per-seat, per-conversation, and per-message plans can stack with extra chatbot or channel fees. (Source: Chatref Pricing Comparison)

Chatbot pricing comparison at 100, 1,000, and 5,000 conversations

Use this table as planning math, not a ranking.

Assumptions: four billable messages per conversation; one website; one chatbot; one required full seat; monthly public list prices; no taxes, annual discount, extra channels, setup, staff time, or custom integration. Chatbase uses one credit per message as a planning proxy. Intercom uses one paid outcome per conversation. Botpress uses published annual-billing equivalents. Extra packs round up.

PlatformBilling unit kept separate100 conversations1,000 conversations5,000 conversationsMath note
SiteGPTmessages$59$59$247400, 4,000, and 20,000 messages; Growth plus two 5,000-message packs at the highest workload
Chatbasemessage credits$40$150$700400, 4,000, and 20,000 credits under the one-credit proxy; Pro plus 5,000 extra credits at the highest workload
Botpressconversations$150$670$2,500Plus covers 250; Team covers 1,500; extra packs are rounded to 100 conversations
Intercom Finoutcomes plus one seat$128$1,019$4,979$29 seat plus $0.99 multiplied by assumed paid outcomes

The Botpress 1,000 row is $150 plus eight packs at $65, or $670. The 5,000 row is $750 plus 35 packs at $50, or $2,500. Intercom is $29 plus 1,000 outcomes at $0.99, or $1,019. (Sources: Chatbase Pricing, SiteGPT Pricing, Botpress Pricing, Intercom Pricing)

Tidio is not assigned a 5,000-conversation number because its public page lists Lyro quotas up to 1,000 and points larger needs to custom pricing. (Source: Tidio Pricing)

How to calculate your own monthly chatbot budget

Worked 1,000-conversation chatbot cost example

This is a completed monthly chatbot budget for a realistic planning scenario. It uses 1,000 website conversations, four messages per conversation, a $129 SiteGPT Growth plan, six one-time setup hours, and two maintenance hours per month at a $40 loaded staff rate.

Cost lineAssumptionWorked value
Monthly website conversations1,0001,000
Messages per conversation44
Required messages1,000 conversations x 4 messages4,000 messages
Credits or paid outcomesNot used in this message-based example$0
Base platform feeSiteGPT Growth$129/month
Extra usage packs4,000 required messages fit inside the included 10,000-message allowance$0
SeatsOne administrator, included in the plan$0
Branding, channels, storage, and analytics add-onsNone assumed$0
Setup time, one time6 hours x $40/hour$240
Maintenance time, monthly2 hours x $40/hour$80/month
First-month total$129 platform + $240 setup + $80 maintenance$449
Ongoing monthly total$129 platform + $80 maintenance$209/month

The vendor comparison table excludes setup and staff time; this worked example adds both. It also keeps message-based billing separate from credits and outcomes instead of pretending those units convert cleanly. (Sources: SiteGPT Pricing, Intercom Pricing)

To calculate your own monthly chatbot budget, choose one vendor, copy its billing definition exactly, and replace the worked values with your traffic and staff assumptions. If the vendor bills message credits, use message credits. If it bills paid outcomes, count paid outcomes. Do not merge different units into one made-up rate.

Change the four-message assumption after reviewing real transcripts. A short FAQ bot may average two messages; a lead-qualification flow may use eight. If a vendor bills outcomes, use the share of conversations that meet its outcome definition. If it bills credits, measure the credits consumed by a representative test set. (Sources: Chatbase Pricing, Intercom Pricing)

Want to turn this budget into a live customer-facing agent? Explore SketricGen Brand Agents. Use the same cost lines to assess its current plan details before comparing it with any other website chatbot.

Where small businesses overspend

Small businesses usually overspend in the gap between AI chatbot pricing on the sales page and the workflow they actually need.

The first trap is buying for peak hopes instead of measured demand. A 5,000-conversation plan may be wasteful if 100 visitors use the bot. Another is buying the cheapest tier, then paying repeated overages or idle seats.

Other leaks include branding removal, a connector locked to a higher tier, duplicated knowledge bases, and staff time correcting stale answers. An independent cost guide also flags knowledge upkeep, extra channels, multilingual behavior, and compliance as recurring lines absent from many first quotes. (Source: Chatref Pricing Comparison)

Before you sign up, take one week of website questions and label each one: answer from content, needs a tool call, needs a human, or should be refused. That small sample gives you a better usage forecast than a generic “messages per month” guess. If you are still choosing the build approach, compare a public website agent with an internal workflow in our no-code AI agent builder guide.

When a higher-priced chatbot is cheaper in practice

A higher subscription can be cheaper when it includes the unit you would otherwise buy as an overage, or removes manual work.

For example, a $150 plan with 4,000 included units may beat a $40 plan with 700 units if your workload is 1,000 conversations and each conversation consumes four units. A plan with API access and a supported CRM connector may also cost less than a cheaper plan that forces a custom integration. The correct comparison is total cost for the same job, not the lowest first-month card charge.

Model choice matters when a platform passes AI spend through to you. Botpress says knowledge-base retrieval and other AI actions contribute to AI Spend, so a long prompt, repeated retrieval, or unnecessary AI step can consume budget without creating a better customer outcome. If you are evaluating raw model or infrastructure economics, see our GPU cost per million tokens comparison. (Sources: Botpress Pricing, SiteGPT Pricing)

The final test is operational: can the team monitor gaps, update source content, and handle escalation? Analytics, refresh controls, and a usable handoff can beat a low plan that leaves those tasks manual. Test that against your staff rate.

If you are ready to shortlist a customer-facing no-code agent, see SketricGen Brand Agents. Use the questions below to verify its current usage unit, integrations, branding, handoff, and support terms before you buy.

Website chatbot pricing buying checklist

Ask the vendor these questions before comparing chatbot pricing:

These questions mirror published limits and add-ons. (Sources: Chatbase Pricing, Botpress Pricing, Intercom Pricing, Tidio Pricing)

  • What exactly is one billable conversation, message, credit, outcome, or action?
  • Does a user message, AI reply, retrieval, tool call, or human handoff consume a unit?
  • What happens at 75%, 95%, and 100% of the allowance?
  • Are overage packs automatic, optional, expiring, or capped?
  • Are model usage and advanced models included or metered separately?
  • How many websites, chatbots, languages, source pages, files, and seats are included?
  • Are CRM, calendar, ecommerce, helpdesk, webhook, and API integrations included at this tier?
  • Is branding removal included, and are voice, SMS, WhatsApp, or email charged separately?
  • Can you export transcripts, usage data, and unresolved questions?
  • Who owns monthly content review, escalation coverage, and accuracy testing?

Save the answers beside your worksheet. A vendor that cannot define its usage unit clearly is difficult to budget, even if its starting price is attractive.

FAQ

What is the average cost of a chatbot?

There is no useful single average for a website chatbot. Public SaaS plans in this comparison run from $40 to $150 at lower workloads, while outcome-based pricing can reach thousands as usage grows. Custom development adds one-time implementation and maintenance. Your workload, billing unit, integrations, and staff time matter more than the category average. (Sources: Chatbase Pricing, Intercom Pricing)

How much does an AI chatbot cost per month?

The practical answer to “how much does a chatbot cost per month?” is about $40 to $150 for a small public website bot before staff time, add-ons, and extra usage. AI chatbot cost can reach hundreds or thousands of dollars monthly at higher volume. The billing unit and handoff workload determine the real total. (Sources: Chatbase Pricing, SiteGPT Pricing, Botpress Pricing, Intercom Pricing)

How much does a website chatbot cost?

A website chatbot can cost $0 to test, about $40 to $150 per month for an entry paid plan, or far more when volume, integrations, human seats, and outcome-based pricing increase. For an apples-to-apples website chatbot pricing comparison, price the same traffic, content, and handoff workflow across each vendor. (Sources: Chatbase Pricing, SiteGPT Pricing, Intercom Pricing)

Can I get a chatbot for free?

Yes. Chatbase lists a free plan with 50 message credits, and Botpress lists a free plan with 25 conversations. Free plans are useful for testing, but they may limit models, seats, branding, storage, inactivity, or overages. Budget for the first paid tier if the bot will serve real website visitors. (Sources: Chatbase Pricing, Botpress Pricing)

What are chatbots mainly used for?

Website chatbots commonly answer questions from company content, qualify leads, recommend products, collect intake details, route support requests, and hand conversations to people. The cost rises when the bot must retrieve live records, write to another system, or operate across more channels. Choose the workflow before choosing the plan.

Is there a better chatbot than ChatGPT for a website?

For a public website, a dedicated chatbot platform may be a better fit than a general AI subscription because it can add a website widget, source controls, analytics, routing, and usage limits. The right choice depends on your content, integrations, handoff needs, and budget. Compare the complete workflow, not model names alone.

References