From February 1, 2022, the WhatsApp Business API started to adopt a billing model based on Talks, counting at 24-hour windows from a session start trigger. The change impacts service operations, notifications, and automations, because the cost now depends on How many sessions are open by end user number over time.
In this content, you will understand which conversations are charged, when the 24-hour window begins to count, and how to classify the most common everyday interactions.
What WhatsApp considers' conversation 'in the billing model
Uma Talk It is a service session or message exchange that remains active for 24 Hours after the start of the window. Within this period, the company can exchange multiple messages with the user without opening a new session, provided that the sending eligibility rules are respected.
When the window expires, any resumption that opens a new session is counted again as New Conversation.
Types of conversations in the WhatsApp Business API
The model separates conversations into two groups, defined by the trigger that opens the session.
1) User-initiated conversations
The session opens when The user sends a message And the company Respond, starting the 24-hour window from the response.
What goes into the charge in this case:
- Reactive service (questions, support, tracking, exchange, cancellation).
- Chatbot interactions when the first contact came from the user and the bot responds.
- Continued support while the window is still active.
2) Conversations initiated by the company
The session opens when The company triggers a proactive message, normally associated with Automatic Messages and Notifications (very common in communications that use templates/templated messages).
What goes into the charge in this case:
- Order confirmation and update.
- Expiration notices, reminders, and announcements.
- Authorization/validation messages that require a response from the customer to unlock next steps.
When charging “becomes” a new conversation
The operating rule that changes the account the most is the expiration of the window. If the conversation started and the service continues after 24 hours, a new session may be counted, depending on the type of resumption.
Points that usually generate new charges:
- Response from the company after the window expires.
- Proactive trigger to resume a topic with no recent message from the user.
- Bot flows that attempt to re-engage the customer when the time limit has passed.
Practical examples of charging for user-initiated conversations
Situation 1: Concerning a plan upgrade
- The customer sends a message asking when the plan update will take effect.
- The company responds and opens the 24-hour session.
- The support message exchanges follow within the same window.
What does the charge look like: A conversation initiated by the user, provided that the resolution occurs within 24 hours.
Situation 2: delivery problem that exceeds 24 hours
- The customer asks about tracking the order.
- The store responds and logs in.
- The window expires without a complete solution.
- The following day, the store responds to the customer again after the 24-hour period.
What does the charge look like: 2 conversations initiated by the user, because the resumption occurred after the expiration of the first session.
Situation 3: customer activates the company and the chatbot responds
- The customer requests order status.
- The chatbot responds and opens the session initiated by the user.
- The conversation expires.
- The bot sends a message later, without a new recent message from the user.
What does the charge look like: a conversation initiated by the user in the first window and, upon proactive resumption after expiration, the new session tends to enter as a conversation initiated by the company.
Practical implications for automation: re-engagement flows need to align with the use of standard messages/templates when the reactive window has already closed.
Practical examples of charging for conversations initiated by the company
Situation 1: Sequence of notifications within 24 hours
- The company sends an order confirmation notification.
- A few hours later, it sends payment confirmation.
- Both messages are delivered within the same 24-hour window.
What does the charge look like: A conversation initiated by the company, because the submissions occurred within the same session.
Situation 2: message that requires a response to release next steps
- The company sends a message asking the customer for authorization to continue.
- The customer responds after the window expires.
- The company responds to the customer's new message.
What does the charge look like: A conversation initiated by the company in the first window; the subsequent response from the customer can open a conversation initiated by the user when the company replies within the new session.
How charges are calculated
The charge is based on End user number (recipient) and varies by Country/region, with different values depending on the type of conversation. To estimate monthly cost more accurately, the most useful metric in planning is:
- Amount of New Sessions Open every day
- Ratio between user-initiated and company-initiated sessions
- Rate of cases that burst 24 hours before resolution
Best practices to reduce unnecessary sessions and maintain quality of care
- Draw Flows to Solve Inside the Window when the case is typically simple (screening, duplicate, status, frequently asked questions). This reduces reopenings that saw new conversations.
- Centralize context collection at the start of the service, because “back and forth” messages consume window time and increase the chance of expiration.
- Create bot fallback trails To forward to the human before it expires when the customer signals urgency or complexity.
- Standardize resumption templates for cases that require proactivity after 24 hours (delivery update, procedure confirmation, review return).
- Monitor Expiration Reasons (queue, SLA, third-party dependency, lack of data) and attack the causes with process improvements.
The chatbot's role in charged conversations
Chatbots help maintain consistent service when:
- the bot resolves repetitive demands quickly, reducing service time within the window;
- the flow collects minimum data for humans to act without rework;
- The rules for resuming after 24 hours are mapped to the use of standard messages, avoiding uncompliant shooting.
Frequently Asked Questions (FAQ)
How many messages can I send within 24 hours?
The volume of messages exchanged within the session tends to be treated within the same conversation, as long as the window remains active and the sending is within the channel's rules.
What happens when 24 hours pass?
A resumption after expiration may open a new conversation, and this will count as a new charge depending on the opening trigger (user or company).
Does a chatbot open a conversation in the same way as a human agent?
The classification depends on the session start trigger. If the user started and the bot responded, it enters as initiated by the user; if the automation reengages without a recent message from the user, it tends to enter as initiated by the company.
