Conversation-based pricing charges for customer threads rather than individual messages or user seats. One customer going back and forth twenty times is one conversation, not twenty. It has become the standard for AI chat tools because it roughly tracks the value delivered rather than the effort spent.
What counts as one conversation
This is the detail that decides your bill, and vendors define it differently. Check three things:
- The window. Is a conversation a thread with no fixed end, or does it reset after 24 hours? A 24-hour window means the same customer over three days is billed three times.
- Who starts it. Are you charged for conversations you initiate as well as ones customers start?
- Whether unanswered threads count. A message that arrives while the AI is switched off, or that gets no reply, should not be billed. Not every vendor agrees.
How it compares to the alternatives
Per-message billing punishes exactly the conversations you want, since a customer asking six follow-up questions is a customer close to buying. Per-seat billing has nothing to do with customer volume, so it suits large teams and penalises small ones. Per-resolution billing charges you more the better the tool works. Conversation pricing avoids all three problems, at the cost of needing a clear definition of the unit.
The question everyone forgets
Ask what happens when you exceed the allowance. There are two answers and they are very different. Either you are billed for the overage, which is flexible and can produce a large surprise invoice, or replies pause until the next cycle, which caps the bill but means real customers get silence. Neither is wrong, but you should know which one you bought.
Our definition, and what happens at the limit, is written out at hustle-flow.com/pricing.