Name the thing being metered
Credits, conversations, resolutions, seats, and actions are different units. Translate the plan into the job your site needs.
Pricing units
Before you pick a chatbot plan, translate the usage unit.
A low monthly price can mean very different things when one vendor meters AI replies, another meters support conversations, and another meters AI-handled support outcomes.
Short answer
Do not compare AI chatbot plans by monthly price until you know the unit that runs out first. A message credit is usually not a visitor, a conversation is usually not an AI reply, and an AI resolution is closer to a solved support case than a raw chat message.
FastBots and Chatbase are easier to model when the main job is trained website answers, because the first question is message credits and source capacity. Tidio and ChatBot.com can make more sense for support operations, but the buyer must model human replies, AI conversations or resolutions, seats, and workflow scope.
This guide explains the units. For the next layer, use the pricing-traps checklist, the scaling-cost guide, or the small-business plan picker.
Comparison method
Translate every plan into the same buyer question.
The practical question is not which plan has the lowest headline price. It is which unit grows first for your site.Estimate how fast it runs out
A visitor might ask one question, trigger several AI replies, hand off to a person, or require a resolved support case.
Find the first upgrade pressure
The first jump may be model cost, source size, live chat, AI actions, users, branding removal, or refill behavior.
Choose by workload, not sticker price
A lightweight FAQ bot and a customer-support workspace can both be good buys, but their cost math is not the same.
Unit map
Four pricing units buyers often mix together.
These are not perfect universal definitions. They are the practical buyer meanings from the official pricing surfaces checked for this article.
| Unit | Where it appears | What it means | Buyer check |
|---|---|---|---|
| Message credits | FastBots, Chatbase | Usually a budget for AI model replies or model requests inside the chatbot product. One visitor conversation can consume several replies, and some models or action workflows can spend more than one credit per answer. | Estimate monthly bot replies and preferred model choice before treating the credit allowance as traffic capacity. |
| Billable conversations | Tidio Customer Service | A support conversation can be billed when a human agent replies or starts the conversation. Human-handled support volume can grow separately from AI automation volume, so a busy inbox can hit the support meter first. | Separate unanswered visitor messages, human replies, Lyro AI replies, and automation flows before choosing a quota. |
| AI or Lyro conversations | Tidio Lyro AI Agent | A customer interaction that receives at least one reply from the AI agent. AI support can become its own quota even when live-chat conversations or Flows automation are priced separately. | Model AI-handled questions separately from live-chat seats, tickets, and rule-based Flows reach. |
| AI resolutions | ChatBot.com by Text | An AI-handled conversation that counts when it ends without the customer asking for more help after the AI's final reply. Resolution allowances, users, AI agents, API calls, workflows, and automatic refill packs can all affect the real monthly bill. | Ask what counts as a resolution, what is excluded, and what happens when the included monthly allowance is reached. |
Tool fit
Which pricing meter should you model first?
Use this as a unit-mapping aid, not as a final ranking or hands-on billing result.Model replies
FastBots
Message credits
Start here if
Small sites that want trained answers from website content, files, links, and simple lead-capture workflows.Before you choose
The official pricing page showed 50 free credits, then 2,000, 5,000, and 10,000 monthly credits on the self-serve paid plans. Standard models can use 1 credit per reply; advanced models can use more.Sources and actions
Chatbase
Message credits plus add-ons
Start here if
Source-heavy sites that need trained answers, AI Actions, agent limits, source limits, and workspace controls.Before you choose
The pricing page showed 50 free credits, then 500, 4,000, and 15,000 monthly credits on Hobby, Standard, and Pro. Actions can make multiple model requests, and auto-recharge is a separate credit behavior to check.Support workload
Tidio
Conversations, Lyro, and Flows
Start here if
Teams that want live chat, tickets, automation, Lyro AI, and human support in one customer-service workspace.Before you choose
The pricing page separates billable conversations from Lyro AI conversations and Flows visitors reached, so price human support, AI support, and automation reach separately.Resolved support cases
ChatBot.com
Users plus AI resolutions
Start here if
Support teams that want an AI agent inside the broader Text workspace with live chat, help desk, workflows, and API scope.Before you choose
The pricing page showed per-user plans with included AI resolutions, plus automatic 50-resolution refill packages when the limit is reached.Unit translation
Turn a month of chat into the right meter.
Start with the workload, then translate it into the pricing unit. This keeps a website-answer plan, a support-inbox plan, and an AI-resolution plan from being compared as if they sold the same thing.
Visitor volume
How many people are likely to open chat in a normal month and a busy month?
Visitor count is only the starting point. A small number of serious support or quote conversations can cost more than many short FAQ checks.
Reply depth
How many AI replies does a useful answer usually take: one answer, a clarifying question, or a multi-step flow?
Message-credit tools are closer to reply or model-request math than raw visitor math.
Handoff rate
Which chats need a person, ticket, callback, booking review, or quote approval?
Support platforms can move from AI usage into billable conversations, seats, inbox rules, and reporting.
Resolved work
Which questions should count as solved by AI, and which still need a person or tested workflow?
AI-resolution pricing is closer to support outcomes than message counts, so unresolved or routed chats need separate handling.
Workload first
Choose the meter that matches the job.
A chatbot that answers public website questions and a chatbot that resolves support cases can both be useful. The mistake is forcing both into the same cost model.
Message credits and source size
Website FAQ bot
The buyer usually needs enough replies, crawl pages, files, and training data before support-workflow features matter.
Check: Start with FastBots or Chatbase-style credit math, then check handoff only if the bot must route leads.
Billable conversations and seats
Live chat with human replies
The cost can grow when a person replies, when more staff need access, or when inbox/ticket features become required.
Check: Use Tidio-style conversation math and separate human replies from AI conversations and Flows reach.
AI conversations or AI resolutions
AI support deflection
The buyer is paying for solved or AI-handled support workload, not just for a fixed number of chat messages.
Check: Ask how the vendor counts solved issues, what is excluded, and what happens after the included quota.
Form fields, handoff, actions, and CRM routing
Sales or quote intake
The chatbot may answer a few questions but still need safe handoff, lead details, scheduling links, or action gates.
Check: Use the quote-request and appointment-booking guides before assuming a low-price plan covers the full workflow.
Sources checked
Official pages checked on June 1, 2026
This guide uses current public pricing pages to identify usage units, included allowances, and overage/refill behavior. Use it as a pricing-unit guide, then run your own trial or account check before relying on invoice behavior, refill settings, or live support volume.
Are message credits the same as conversations?
No. Message credits usually meter AI replies or model requests. A visitor conversation can use several replies, especially when the chatbot asks clarifying questions or uses a higher-cost model.
Are AI resolutions the same as AI replies?
No. A resolution is closer to an AI-handled support outcome than a reply count. On the checked ChatBot.com pricing page, a conversation counts when it ends without the customer asking for more help after the AI's final reply.
What should I compare first?
Start with the job. For website answers, compare message credits and source limits. For support, compare billable conversations, AI conversations or resolutions, users, seats, and handoff.
Next checks
Compare the unit before the plan.
Once the pricing unit is clear, check scaling costs, plan gates, source limits, handoff, and quote or booking workflows before you commit.
FAQ
Pricing-unit questions.
Are AI chatbot message credits the same as conversations?
No. Message credits usually meter AI replies or model requests, while conversations usually describe a support thread. A single visitor conversation can use several messages, and some platforms also meter AI conversations or AI resolutions separately.
Reviewed
What is an AI resolution?
An AI resolution is closer to an AI-handled support outcome than a raw message count. For ChatBot.com by Text, the pricing page says a conversation counts as a resolution and is billed if it ends without the customer asking for more help after the AI's final reply.
Reviewed
Which chatbot pricing unit should a small business model first?
Model the unit that matches the job. A public website FAQ bot usually starts with message credits and source limits. A support workspace should model billable conversations, AI conversations or resolutions, seats, handoff, and workflow limits.
Reviewed
Does the AI model change how fast message credits run out?
Yes, model choice usually changes the credit burn per reply. On the checked FastBots pricing page, standard models such as Gemini Flash and GPT mini variants ran at one credit per reply, while higher-tier models from OpenAI, Anthropic, and Google ran at five or even ten credits per reply. We said this means a credit allowance is closer to a reply budget at a chosen model tier than a flat message count. Pick the model the chatbot will actually use before treating the monthly credit number as traffic capacity. Recheck the current pricing page because model lists and per-reply costs can change.
Reviewed · Sourced from FastBots pricing
When does Tidio count a billable conversation?
On the checked Tidio pricing page, a billable conversation is any live chat, ticket, email, or social channel conversation that includes a message from a human agent, and unanswered visitor messages are not counted. Replies from Flows automation and from the Lyro AI Agent are not counted as agent replies for this meter. Our view is that this makes the billable-conversation meter closer to human-support workload than to AI traffic, so visitor count alone is the wrong way to size a plan. Confirm the definition on the current pricing page before treating chat volume as the billing driver.
Reviewed · Sourced from Tidio pricing
How do I translate my chatbot traffic into the right pricing meter?
Start with the workload, not visitor count. Estimate how many AI replies a useful answer takes, how many chats need a person, and how many should count as solved by AI. Match each to the meter the vendor actually bills against: replies map to message credits, human-handled threads map to billable conversations, AI-handled outcomes map to AI conversations or resolutions, and proactive popups map to visitors reached. Our view is that this prevents comparing a website-answer plan to a support-inbox plan as if they sold the same unit. Run a one-month trial before locking in a quota.
Reviewed
Decision recap
If this, then that.
- Pick FastBots — if message credits and source size are the meter that matches the website-answer job.
- Pick Chatbase — if message credits plus AI actions and source control match the real workload.
- Use Tidio — if billable conversations, Lyro AI, and Flows visitors are the right meter for a support workspace.
- Use ChatBot.com — if per-user seats plus AI resolutions match an outcome-based support workflow.
- Translate the workload first — into the unit that will run out; do not compare plans by monthly price alone.