WhatsApp Business API pricing explained
Almost every guide to WhatsApp pricing on the web, including the earlier version of this page, describes a model Meta retired in July 2025. Here is what the WhatsApp Business Platform actually charges for in 2026, which messages are free, and the numbers we measured on our own account.
Meta stopped charging per conversation in July 2025
The model most articles still describe works like this: a customer messages you, that opens a 24-hour conversation window, and Meta charges once for the window no matter how many messages travel through it. It was a tidy model and it is gone. Since July 2025 the WhatsApp Business Platform bills per delivered message.
This matters more than a billing detail usually would, because it inverts the advice that follows from it. Under conversation pricing, the cheapest thing you could do was cram as much as possible into one open window, and length was free. Under per-message pricing, every message you send is a line item, and a chatty automation that used to be free now has a running meter on it. If you built your WhatsApp flows before mid-2025 and never revisited them, they were optimised for a price list that no longer exists.
So when you read a WhatsApp pricing guide, check its date before you check its numbers. A page that talks about conversation categories and 24-hour windows is describing 2024.
Your bill has three axes, and Meta moves all three
The price of a billable message is decided by three things at once:
- Template category. Authentication, marketing, utility or service. These are priced very differently from one another, and the gap is not small.
- The recipient's country. Not your country. A business in Doha messaging customers in Lagos is charged at Nigeria's rates, not Qatar's.
- Your own volume tier. Send more and the per-message rate moves.
Meta reprices all three on its own schedule and does not ask anyone's permission. That is the real reason this page does not contain a rate table. Any table a software vendor publishes is a snapshot of a moment, it starts drifting the day it goes up, and the people quoting it usually have no way to notice it has drifted. We learned this the expensive way on our own product: we used to estimate WhatsApp cost from a local rate table, and the estimate slowly walked away from the actual invoice until it was worth nothing. We deleted the table and now read the money from Meta's own reporting instead.
If you want a rate, get it from Meta's current published pricing for the countries you actually message, on the day you need it. If you want a forecast, the rate is the least important input, for the reason in the next section.
The part nobody explains: most of your messages are probably free
Every delivered message carries a pricing type, and only one of the three costs anything.
- Regular. Billable. This is a business-initiated message, or a message outside any free window.
- Free customer service. Not billable. When a customer messages you first, replies you send inside the following 24 hours are free, however many you send.
- Free entry point. Not billable. A conversation that starts from a click-to-WhatsApp ad opens a 72-hour free window, not 24.
This is the single most useful fact about WhatsApp pricing and it is almost never the headline of an article about it, because it is not a number and cannot be put in a comparison table. A support account, where customers write in first and you answer, lives mostly in the free customer service window. A broadcast-heavy marketing account lives mostly in regular. Two businesses with identical message volumes can be two orders of magnitude apart on cost, and the thing that separates them is not the rate they negotiated. It is who started the conversation.
What we actually pay, on our own number
Rather than repeat someone else's table, here is our own bill. This is Replio's own WhatsApp number over one 30-day window ending 4 September 2026, on a traffic mix that is mostly Qatar and Nigeria. It is a small account and it is not a benchmark for yours, but it is real and it is measured rather than modelled.
| Pricing category | Messages delivered | What Meta charged |
|---|---|---|
| Service (replies inside the free window) | 9,062 | $0.00 |
| Utility | 95 | $0.59 |
| Authentication (OTP codes) | 55 | $0.52 |
| Marketing | 1 | $0.03 |
| Total | 9,213 | $1.14 |
Read the first row again. 9,062 of them, 98 per cent of everything the account sent, cost nothing at all, because they were replies to customers who had written in first. The entire month's WhatsApp bill was $1.14, and effectively all of it came from the 151 messages that were business-initiated. That is what a support-shaped account looks like on per-message pricing.
The billable rates behind those totals, measured per billable message rather than per delivered message:
- Authentication: $0.0094
- Utility: $0.0077
- Marketing: $0.0341
Treat those as one account's observation in one window, not as a rate card. The point is not the digits, it is the shape: marketing costs about four and a half times what utility costs, and service costs nothing.
The arithmetic mistake that makes every forecast about 20% light
Here is a trap worth more than any rate table, and we walked straight into it while building our own cost reporting.
Free messages are still delivered messages. They appear in your volume. So the obvious way to work out your cost per message, total spend divided by total messages, quietly blends thousands of free messages into the denominator and hands you a number that is far too low. Then you forecast a campaign with it, because a campaign is business-initiated and every message in it is billable, and your forecast is wrong in the expensive direction.
When we first built this reporting, dividing by total volume gave a utility rate of $0.0062 a message. Dividing by billable volume gave $0.0077. That is a 20 per cent error, baked into every projection downstream of it, and nothing about the low number looks suspicious on a dashboard.
So: always divide cost by billable messages, never by delivered messages. If your provider's analytics does not separate the two, you cannot forecast with it, and you should ask them why not.
Marketing templates are where bills quietly multiply
Of the four categories, marketing is the one to watch, and not only because it is the most expensive. It is also the easiest to end up in by accident. A message that tells someone their order shipped is utility. The same message with a discount code attached to the bottom is marketing, and on our numbers that addition cost roughly four and a half times more to send.
Meta decides the category, not you, so the practical discipline is simple: if a message genuinely confirms, updates or follows up on something the customer already set in motion, keep it clean and send it as utility. Put the promotion in a separate, deliberate marketing send that you have actually budgeted for, rather than stapling it to every transactional notification you have.
Click-to-WhatsApp ads buy you a free 72-hour window
The free entry point type deserves its own paragraph because it is the one lever that makes outbound acquisition cheap on WhatsApp. When someone taps a click-to-WhatsApp ad and lands in your chat, that conversation is free to run for 72 hours rather than the usual 24, and every message in it is free.
We have watched an account on our own platform deliver several thousand WhatsApp messages in a month and be billed nothing, because almost all of that traffic arrived through click-to-WhatsApp ads. Their Meta ad spend was real, of course, but their WhatsApp messaging cost was zero. If you are already buying Meta ads, pointing some of them at WhatsApp instead of a landing page changes the cost structure of the conversation that follows, not just the click.
The other half of the bill is the software, and that is where markups live
Everything above is Meta's charge. Separately, you pay whatever platform gives you an inbox, automation, AI, broadcasts and reporting. Providers differ enormously here, and not only on the sticker price:
- Some resell Meta's messages. You buy message credits from the vendor rather than being billed by Meta, which means the vendor chooses the rate and you cannot see Meta's.
- Some add a per-message markup on top of Meta's rate, sometimes disclosed as a percentage, sometimes just baked in.
- Some meter the software itself by monthly active contacts, so a successful broadcast pushes you into a higher bracket even though your question volume never changed.
- Some pass Meta through at cost and charge only for their own software.
Three questions settle it quickly with any vendor. Does Meta bill me directly, or do I buy messages from you? If I buy from you, show me your rate next to Meta's for my top three countries. And does your reporting separate billable from free messages? A provider that cannot answer the third question cannot tell you what your WhatsApp actually costs, whatever their pricing page says.
How Replio prices this
Replio is a flat monthly subscription and nothing else: $49, $149 or $349 a month, with unlimited messages on every paid plan and a free plan at $0 to start on. Starter covers one of WhatsApp, Instagram or Messenger with Telegram and the website widget included free on top; Scale runs WhatsApp, Instagram and Messenger together in one shared inbox; Pro adds every channel at full scale, and there is an Enterprise tier for high volume, multi-brand and agencies.
On WhatsApp, you connect your own number and Meta bills you directly. We do not resell messages, we do not mark them up, and we never sit between you and Meta's invoice, which also means we cannot quietly change the rate. On Telegram, Instagram, Messenger and the website widget there is no per-message charge from anyone at all, which is why routing routine traffic to those channels is a genuine cost lever rather than a talking point.
So the whole bill is: one flat subscription, plus Meta's own per-message charges if and only if you use WhatsApp, with nothing added in between.
Five things that actually lower a WhatsApp bill
- Answer inside the 24-hour service window. Every reply there is free, so the faster your first response, the more of the conversation lands in free territory. This is the main reason an always-on AI agent pays for itself on WhatsApp before you count the labour.
- Resolve rather than escalate. A question closed today does not become a business-initiated follow-up tomorrow, and the follow-up is the billable one.
- Send transactional messages as utility, not marketing. Roughly four and a half times cheaper on our own numbers.
- Use click-to-WhatsApp ads as the entry point where you are already advertising, for the free 72-hour window.
- Move routine traffic to channels with no message fee. Telegram and a website widget cost nothing per message, and plenty of the questions arriving on WhatsApp do not need to be there.
Try Replio free for 7 days
Connect WhatsApp, Instagram, Messenger or Telegram and have an AI support agent answering in minutes, inside the free window where it costs you nothing. No card required.
Start free trialFrequently asked questions
Does the WhatsApp Business API cost money?
Is WhatsApp still priced per 24-hour conversation?
How much does one WhatsApp message cost?
Why is my WhatsApp cost per message lower than the rate I was quoted?
Are WhatsApp message fees included in Replio's subscription?
Which is cheaper, a marketing template or a utility template?
Related
Last updated 2026-09-12. Meta's rates and categories change on Meta's schedule; the figures above are our own measurements over one 30-day window ending 4 September 2026 on a mostly Qatar and Nigeria traffic mix, shown to illustrate the shape of a bill rather than to predict yours. Verify current rates with Meta for the countries you message.