The 24-hour window has been the cheap part of WhatsApp since November 2024. From October, replying inside it costs money.
What is true today, and verifiable
Meta moved from conversation-based to per-message billing on 1 July 2025. You are charged when a template message is delivered, and the old arithmetic of "one 24-hour conversation, one charge" no longer applies.
Meta’s pricing documentation currently states that all non-template messages are free, that they can only be sent inside an open customer service window, and that utility templates delivered inside that window are also free. Lower utility and authentication rates unlock as monthly volume rises; marketing templates are excluded from those volume tiers.
- Charged per template message, only when it is delivered
- Free-form replies inside the open 24-hour window: free
- Utility templates inside that window: free
- Volume tiers reduce utility and authentication rates, not marketing
What changes on 1 October 2026
Meta announced on 1 July 2026, alongside the Meta Business Agent Platform, that service messages stop being free. From 1 October 2026 every free-form reply your business sends inside the 24-hour window is billed at the same per-message rate Meta already charges for utility and authentication templates, set per market. Utility templates sent inside the window are reported to become billable on the same date.
The window itself is unchanged. It still opens each time a customer messages you, still resets on each new inbound message, and you still need an approved template to re-engage anyone outside it. What changes is that being inside the window no longer means free.
A caveat worth stating plainly: at the time of writing, Meta’s public pricing documentation still describes the service window as free and lists only market-specific rate adjustments for 1 October. The change is reported consistently by solution providers citing Meta’s partner announcement rather than a public page. Confirm against your own billing before you budget on it.
Who this actually hurts
If you mostly broadcast, very little changes: marketing templates were already the expensive part and they are billed the same way. If you run support on WhatsApp, this is a real new line item, because the messages that were free are exactly the ones you send most.
The businesses most exposed are the ones that answer a lot and sell a little per conversation: support desks, order-status enquiries, appointment rescheduling, anything where a customer asks three questions and gets three answers.
- High-volume support teams: every agent reply becomes billable
- Businesses running third-party AI agents: those replies count as service messages
- Ecommerce sending in-window utility updates: previously free, now charged
- Pure broadcast senders: largely unaffected
Estimating your exposure before October
- Export a month of conversations and count outbound messages sent inside an open window.
- Separate agent replies from in-window utility templates; both become billable, but the volumes usually differ a lot.
- Multiply by Meta’s utility rate for your market, since that is the rate reported to apply.
- Compare that against your current monthly Meta spend to see the percentage increase.
- Do it now rather than in October, while the messages are still free and the sample is honest.
What to change, and what not to
The obvious reaction is to reply less. That is the wrong lever: fewer replies means worse service and more blocks, and blocks cost you the quality rating that governs your daily messaging limit.
The useful levers are about resolving in fewer messages rather than fewer conversations. A bot that answers the common question completely in one message costs one message. An agent who sends five short lines costs five.
- Answer completely in one message rather than in five short ones
- Move genuinely repetitive questions to a bot flow that resolves them in one reply
- Use in-chat forms instead of asking six questions one at a time
- Keep order updates as utility templates, which still get volume-tier rates
- Do not cut response quality to save messages; blocks cost more than messages do
What is still unclear
Two things are worth watching rather than planning around. Whether replies sent by Meta’s own Business Agent are treated differently from third-party AI replies is reported but not documented publicly. And whether the date holds: Meta has moved pricing dates before, and its public documentation has not yet caught up with this one.
We will update this page when Meta’s pricing documentation reflects the change, and note the date it did.