WhatsApp template vs service messages: which one you just sent, and what it cost
Last updated 23 October 2026
Short version. Meta sorts every WhatsApp message a business sends into one of five categories and charges each differently. Four of them are templates you wrote in advance and submitted for approval. The fifth, service, is everything else: the ordinary replies your team or your assistant types inside a conversation a customer started. Since 1 October 2026 service messages are billed, after 1,000 free per phone number per month, at your market's utility rate and with no volume discount.
If you only remember one thing: a template is a message you wrote before the customer existed. A service message is a reply to the customer in front of you.
What is a service message?
Meta's own definition is the useful one: a service message is "any non-template message not powered by the Meta Business Agent Platform".
Unpack that and it means three things at once. It is free-form, so you wrote it just now rather than submitting it for approval weeks ago. It can only be sent while the 24-hour customer service window is open, which is to say only in reply to a customer who messaged you. And it is not one of Meta's own AI replies, which are billed separately per token.
Every reply a support agent types is a service message. Every reply an assistant like Jabee sends is a service message. "Yes, we deliver to Al Barsha, it's 20 dirhams" is a service message.
What is a template message?
A template is a message you wrote in advance and submitted to Meta for approval, with placeholders for the bits that change. Meta splits templates into three categories, and the category decides the price:
- Marketing — promotional. An offer, a new collection, a
reminder that you exist. The most expensive category in every market.
- Utility — transactional or informational. Your order has
shipped. Your appointment is tomorrow at ten.
- Authentication — identity verification. One-time
passcodes.
Templates are the only thing you can send when the 24-hour window is shut. That is their real purpose: they are how you start a conversation, or restart one that went quiet.
How do I tell which one I just sent?
Two questions, in order.
Did you send a pre-approved template? If yes, it is billed as marketing, utility or authentication, whichever category that template was approved under. If no, keep going.
Was the customer's last message less than 24 hours ago? If yes, you sent a service message. If no, the send failed, because free-form messages outside the window are rejected rather than delivered.
In the Cloud API the distinction is visible in the request body: a template send carries "type": "template" and a template name, and a service message carries "type": "text" and the words themselves. If you are building against the API, that field is your billing category.
What does each one cost?
Everything is priced per delivered message, and every rate varies by the customer's market rather than yours. Service messages are charged at the same unit value as utility and authentication messages in that market.
| Category | What it is | Volume tiers? |
|---|---|---|
| Marketing | Promotional template | Varies by market |
| Utility | Transactional template | Yes |
| Authentication | One-time passcodes | Yes |
| Service | Your free-form replies | No |
| Meta Business Agent | Meta's own AI replies | Per token, $2.00 per million |
For the actual numbers in the Gulf and a worked monthly total, see what you actually pay after 1 October. In the UAE a service message is $0.0157 delivered.
Why does service pricing have no volume tiers?
Because Meta said so, explicitly: "Meta will not offer volume tiers for service messages." Utility and authentication templates get cheaper as you send more of them in a month. Service messages do not.
This matters more than it sounds. A business that scales support conversations — which is exactly what happens when you put an assistant on the number — pays the same per reply at 50,000 a month as at 500. Your support cost per conversation does not improve with scale on this line item, so the thing worth optimising is the number of replies a conversation needs, not the rate.
In practice that means an assistant that answers in one message is cheaper than one that answers in four, and an assistant that hands over to a human after two useless replies has already cost you three messages before the human starts. We built Jabee's retrieval around that: search before answering, answer in two or three sentences, and never send a holding message that says somebody will get back to you.
What changed on 1 October 2026?
Two things, both of which make replying cost money where it previously did not:
- Service messages became billable. Each business phone
number gets 1,000 free per month. They do not roll over and they reset monthly. After that, your market's utility rate per delivered message.
- **Utility templates sent inside an open 24-hour window became
billable too.** Those had been free since 1 July 2025, which is a detail that catches people out: a flow that was free a year ago is not free now.
Separately, from 1 August 2026, Meta Business Agent replies are billed per token at $2.00 per million tokens. Meta puts typical consumption at 20,000 to 25,000 tokens per message, which it says works out at roughly 4 to 5 US cents a reply.
What this means if you run an AI assistant
Every reply your assistant sends is a service message, so the arithmetic is simple and worth doing before you turn one on.
Take a business answering 3,000 customer messages a month. The first 1,000 replies are free. The remaining 2,000 at the UAE rate of $0.0157 come to about $31 paid to Meta. That is Meta's bill, and it arrives separately from whatever your software costs.
Compare it with Meta Business Agent at 4 to 5 cents a reply: the same 3,000 replies would be roughly $120 to $150 in token charges, on top of the delivery. The cheaper line is not always the right choice — but you should know which one you are on.
One more consequence that is easy to miss. Because the free allowance is per phone number, a business running three numbers gets 3,000 free service messages a month, not 1,000. That is an argument for splitting by branch or by brand if you were going to anyway. It is not an argument for splitting a single support queue across numbers, which costs you the shared conversation history that makes an assistant useful.
Frequently asked questions
What is the difference between a template and a service message? A template is a message you wrote in advance and submitted to Meta for approval, and it is the only thing you can send when the 24-hour window is closed. A service message is a free-form reply you write in the moment, and it can only be sent while a customer's 24-hour window is open.
Are service messages free? The first 1,000 per business phone number per calendar month are free. They do not roll over. After that each delivered service message is charged at your market's utility rate.
Do service messages get volume discounts? No. Meta has said explicitly that it will not offer volume tiers for service messages, unlike utility and authentication templates, which do get cheaper at higher volumes.
Which rate applies, my country's or the customer's? The customer's. A business in Dubai pays the UAE rate for UAE customers and a different rate for customers in another market, on the same number in the same month.
Is an AI assistant's reply a service message? Yes, if it is your assistant. Meta defines a service message as any non-template message not powered by the Meta Business Agent Platform, so replies from your own software are service messages. Replies from Meta's own Business Agent are billed per token instead.
Does the 1,000 free allowance apply per business or per number? Per business phone number. A business running three numbers has three separate allowances of 1,000 service messages a month.
Sources
- Meta, WhatsApp Business Platform pricing: message categories, the 1,000 free service messages per phone number per month, and per-delivered-message charging. Read 23 October 2026.
- Meta, pricing updates for Meta Business Agent, service and utility messages: the 1 October 2026 service-message change, the end of free in-window utility messages, and the $2.00 per million token rate effective 1 August 2026. Read 23 October 2026.
Written by Mu'men Tayyem, founder of M4B1 in Dubai. These are notes from building Jabee, a WhatsApp AI assistant that answers customers from a business's own website and calls the business's own APIs as agent tools.