Since 1 July 2025 the WhatsApp Business Platform bills each delivered template message individually, instead of charging for a 24-hour conversation window. Under the old conversation-based model a single template opened a billable session and everything sent inside it rode along on that one charge. Under per-message pricing, five templates delivered to the same person on the same day are five billable events. Only two variables set the price of each one: the recipient's country and the template's category. Meta presented the change as aligning WhatsApp with how SMS and comparable channels bill, and the practical effect is that your invoice now tracks message volume rather than session timing.
This guide covers what actually changed at the billing layer, the message categories that set your rate, what is still free today, how volume tiers lower the price, a worked example of a monthly bill you can rebuild with your own numbers, and the pricing change Meta has already queued for service messages.
What per-message pricing replaced
Conversation-based pricing charged for a 24-hour session opened by a business-initiated template. Once that window was open, additional messages in the same category cost nothing extra. That created a real optimization: batching sends, timing campaigns to reuse an already-open window, and consolidating notifications all lowered the bill.
Per-message pricing removes that lever entirely. Every delivered template is a line item. Meta did not re-base the rates at the switch — the per-message rate started as the rate that had previously applied per conversation — but it did adjust utility and authentication rates in several markets on the same date. So the model changed, and separately the numbers moved.
The important consequence is structural: cost is now linear in message count. A campaign that sends three follow-ups instead of one costs three times as much, regardless of whether the recipient was already in an open conversation with you.
The categories that set your rate
Marketing. Promotions, launches, re-engagement, cart recovery, anything inviting a new purchase. The most expensive category in essentially every market, and the one with the tightest quality scrutiny.
Utility. Messages tied to a specific transaction or account event the customer already triggered: order confirmations, shipping updates, appointment reminders, payment receipts. Priced well below marketing.
Authentication. One-time passcodes and login verification, on its own pricing track and its own template rules.
Service. Free-form, non-template replies you send while a customer-initiated conversation is open. These are not template sends and, today, are not charged.
The category is assigned to the template when Meta approves it, not chosen at send time. That single fact is why template design is a pricing decision: the words you submit for approval determine the rate of every message that template will ever deliver.
What you are not billed for today
Utility templates and service replies inside the open customer service window. When a customer messages you, a 24-hour window opens from their last message. Utility templates and free-form service replies sent inside it are currently not charged.
The free entry point window. A conversation that starts from a Click-to-WhatsApp ad or a Facebook or Instagram page call-to-action opens a free entry point window. If you respond within 24 hours, messages are free for 72 hours from your response — including template messages.
Both of these reward the same behaviour: getting the customer to open the conversation, then doing the work inside their window rather than pushing everything through paid templates.
Volume tiers
Since the same July 2025 date, utility and authentication templates unlock lower per-message rates as monthly volume rises. Volume is aggregated at the business portfolio level across every WhatsApp Business Account the portfolio owns, and tiers are evaluated per market-and-category pair. If you run several brands or several WABAs under one portfolio, their volume counts together — which can put you in a better tier than any one of them would reach alone. Check that your accounts actually sit under the same portfolio, because this is a common and expensive misconfiguration.
A worked example
The rates below are placeholders chosen for clean arithmetic, not Meta's published prices. Rates vary by country and change; substitute your own market's numbers from Meta's current rate card and the structure still holds.
Assume a mid-size ecommerce sending, in one month: 40,000 marketing templates (two campaigns to 20,000 opted-in contacts), 60,000 utility templates (three per order across 20,000 orders), of which 15,000 land inside an already-open service window, and 8,000 authentication messages. Call the rates M = 0.0500, U = 0.0100, A = 0.0080 per message.
Marketing: 40,000 x 0.0500 = 2,000. Utility: 15,000 are free inside the window, so 45,000 x 0.0100 = 450. Authentication: 8,000 x 0.0080 = 64. Total: 2,514, before any BSP fee.
Two things fall out of that arithmetic immediately. First, marketing is about 80% of the bill on 33% of the volume — the mix, not the total message count, is what you manage. Second, the free-window effect saved 150, roughly a third of the entire utility line, purely because those messages followed a customer's own reply. Every improvement to inbound response rate compounds directly into the bill.
The number that never appears on the invoice is opt-in quality. Marketing templates delivered to unengaged contacts cost the full rate, return nothing, and drag down the quality rating that governs your messaging limits. A cheaper rate on a worse list is not a saving.
The change Meta has queued
Meta has published an upcoming pricing update covering non-template messages. Providers across the ecosystem report it taking effect on 1 October 2026: service messages become chargeable per message at the utility and authentication rate, and utility templates sent inside the customer service window lose their free treatment. Free entry point windows are reported to remain free, and service messages are reported not to receive volume-tier discounts. Separately, a Meta Business Agent category priced on token consumption has been reported as launching in mid-2026.
Treat the date and the scope as the single assumption most likely to break your model, and confirm both on Meta's own pricing pages before committing a budget. If those changes land as described, the economics shift from "get the customer to reply and everything is free" to "get the customer to reply and everything is merely cheaper" — the direction of the advice is unchanged, the magnitude is not.
When per-message optimization is the wrong focus
If you send low volume, the engineering work to split templates, re-categorize, and re-time sends will cost more than it saves; the rate card is not your bottleneck, conversion is. If a message's real job is to drive a purchase, it is marketing — rewriting it to look transactional invites reclassification, template pausing, and quality-rating damage that dwarfs the per-message difference. And cutting genuinely useful utility messages to shave the bill trades a small, visible saving against a larger, invisible cost in support load and retention. Optimize the mix and the opt-in list first; optimize the rate last.
Building on a model that keeps moving
Because the billing model, the rates, and the free categories have all changed inside eighteen months, the durable advantage is operational: correct template categorization, clean opt-in, and enough inbound conversation that most of your volume happens inside the customer's own window. TikJoy runs the official WhatsApp Business API with a 24/7 AI concierge built to absorb inbound questions in exactly that window. Whichever provider you use, ask them to show your bill broken down by category and by inside-versus-outside the window — if they cannot, you cannot manage the cost.
A note on rates. WhatsApp pricing is country-specific and revised periodically, and the changes described above are scheduled rather than settled. Verify current rates, categories and effective dates on Meta's official pricing documentation before modelling costs (guidance current as of July 2026).