How to reduce your WhatsApp message costs
Meta bills per delivered template message, by category and by the recipient's country. Four of the six things that move that bill are decisions about how you work, not features you buy.
Before you start
- Visibility of what you currently send
- Your Meta invoice
The steps
- 1
Reply inside the 24-hour window
Free-form replies within 24 hours of the customer's last message cost nothing. A team answering within the hour pays for almost nothing; a team answering in two days pays for a template every time.
- 2
Get your categories right
Utility is roughly 7.5x cheaper than marketing in India. Write utility templates that are genuinely about a transaction the customer initiated, and you pay utility rates honestly rather than discovering a re-categorisation on the invoice.
- 3
Send utility templates while the window is open
A utility template sent inside an open service window is not billed at all. Sequencing an order update to land while the conversation is still live is free money.
- 4
Consolidate messages
Under the old conversation model, splitting a message in two for readability was free. Under per-message billing it is not. Two template messages are two charges.
- 5
Clean the list
You pay for delivered messages. Sending to 50,000 contacts of whom 15,000 are dead numbers wastes money, and the poor engagement drags your quality rating down, which caps how much you can send at all.
- 6
Send fewer marketing messages to better segments
The expensive category is also the one that generates blocks. A relevant message to 5,000 people who asked for it beats the same message to 50,000 who did not, on revenue as well as cost.
Where this usually goes wrong
-
Filing marketing as utility to save money
Meta re-categorises, and you find out through the invoice or a sudden rejection rather than a warning.
-
Optimising the platform fee instead of the Meta bill
The Meta bill is usually the larger of the two. Check whether your provider also marks up the per-message rate.
-
Using a cost calculator built on conversation pricing
That model was retired in July 2025. Any estimate based on it will be wrong.
Read next