Usage
FluentChat measures usage in two ways. Allowance usage is what counts against your team’s limits: chatbots, documents, and messages. Token usage measures how much AI processing each reply took. They answer different questions and don’t add up to each other.
| Where | What it shows |
|---|---|
| Settings > Billing | Allowance usage for the whole team. See Billing. |
| Bot > Analytics | Token usage and requests for one bot. See Analytics. |
What counts as a message
A message is one AI reply admitted against your team’s allowance. When a reply starts, FluentChat reserves one message from the allowance. If the reply finishes, the message is used. If it fails, the message is given back.
| Activity | Uses a message? |
|---|---|
| A bot reply on any channel, including the API and Fluent Support | Yes, one per reply |
| A reply in the Playground or a channel preview | Yes |
| A reply served from FluentChat’s answer cache | Yes, but it records no tokens |
| Suggest replies in the live chat inbox | Yes, one per request, which drafts three replies |
| A reply where the bot calls actions or searches knowledge | One for the whole reply, however many steps it takes |
| A reply that fails before it finishes | No, the reserved message is given back |
| A message a human agent sends in the inbox | No |
| A visitor’s message to your team during live chat | No |
Deleting a conversation doesn’t give back the messages it used. The Messages meter counts replies, not saved transcript rows.
What counts as a document or chatbot
| Allowance | What it counts |
|---|---|
| Chatbots | Every bot in the team. |
| Documents | Every document in the team’s Knowledge Base. A document counts once, even if several bots use it. |
When the team is at its document allowance, you can’t add a source, and a website crawl stops adding pages.
Token usage
A token is a small unit of text the AI model reads or writes. Input Tokens are what the bot sends to the model, such as the question, instructions, and knowledge. Output Tokens are what the model writes back.
Tokens help you understand cost and answer length. They don’t count against your allowance. Find them in each bot’s Analytics page, in the summary cards and the Token Usages table, where you can filter by Source and export the rows.
When you reach a limit
| Limit | What happens |
|---|---|
| Messages used up | New AI replies are refused with “You have crossed your message limit. Please upgrade your plan.” |
| No message allowance, or the balance expired | New AI replies are refused with “Please buy a subscription or your trial has expired.” |
| Chatbots reached | Creating a bot fails with “Chatbot limit reached. Please upgrade your plan.” |
| Documents reached | Adding a source fails with “Document limit reached. Contact an administrator for more credits.” |
Live chat with your team keeps working when messages run out. Visitor messages to your team and agent replies don’t use the allowance.
Limitations
- Analytics and Billing won’t match. Analytics counts requests and tokens per bot. Billing counts admitted AI replies for the whole team.
- Usage is per team. Each team has its own allowance.
- Message usage isn’t split by bot on Billing. Use each bot’s Analytics to compare activity.