WhatsApp is Charging for Service Messages — Big Pricing Changes Coming H2 2026

Meta is rolling out the biggest WhatsApp Business Platform pricing shift in years. Starting in H2 2026, messages that used to be free inside the 24-hour customer service window will start incurring charges. Here's what's changing, when, and what to do about it.

User-Initiated vs Business-Initiated — Today

Today the model is simple:

• Business-initiated — you message the customer first (or after 24 hours of silence). You must use a pre-approved template. Paid per conversation, priced by template category (marketing / utility / authentication).
• User-initiated — the customer messages you first. This opens a 24-hour service window where any reply (text, image, template) is free.

The 24-hour free window is what most businesses actually rely on for customer support. That's the piece Meta is now changing.

What Changes in H2 2026

Meta is splitting non-template replies into two new billed categories:

1. Meta Business Agent Messages — Aug 1, 2026

Replies generated by Meta's own Business Agent Platform (their new first-party AI, announced at Conversations 2026). Billed on token consumption, not per message.

• Rate: $2.00 USD per 1 million tokens
• Typical cost: ~4-5 cents per message
• Scales with complexity — longer conversations cost more
• Bundled: the price includes AI processing and message delivery in one charge

2. Service Messages — Oct 1, 2026

Any reply your team (or a third-party AI like Teambox's built-in bot) sends inside the 24-hour service window. Billed per message.

• Rate: matches the utility/authentication rate for the recipient's country
• No volume tiers — no bulk discount
• Updated quarterly — Meta will publish the first Oct 2026 rates by September 1, 2026

Same Conversation, Very Different Bill

Meta's own example shows how a typical support flow goes from 1 charge today to 5 charges by October 2026:

Meta Business Agent vs Third-Party AI

Meta's own comparison, for 10,000 AI-powered messages in Brazil:

• Meta Business Agent messages: $400–$500 (token-based, includes delivery)
• Service messages sent by third-party AI: $268–$968 (per-message rate; varies by country)

Meta is positioning their own Business Agent as a bundled deal — one price for AI + delivery. Third-party AI users pay the per-message rate, which can be cheaper or more expensive depending on the destination.

What This Means for Your Business

1. Your WhatsApp bill is going up

If you lean on the 24-hour free window for support, expect a jump. Businesses with heavy support volume could see hundreds or thousands of dollars in new charges each month.

2. Chat efficiency matters more than ever

The old game was: keep the customer engaged so you stay in the free window. The new game is: resolve conversations in fewer messages. Every back-and-forth costs money.

3. Templates get relatively cheaper

A single utility template with a button that resolves the issue is often cheaper than a five-message back-and-forth service conversation.

4. You need message-level cost visibility

The free window hid your true cost. Starting October, every reply has a price tag — you need to know which conversations are profitable and which are eating margin.

How to Prepare — Concrete Steps

1. Enable Pricing Analytics API — use PRICING_CATEGORY: SERVICE to track service-message spend separately.
2. Update webhooks — Meta is adding a category: service field so you can attribute costs correctly.
3. Reduce reply-count per resolution — audit your templates and canned responses so you answer more with fewer messages.
4. Batch responses — one well-structured message beats three short ones from a cost perspective.
5. Use buttons and quick replies in templates so customers self-serve instead of triggering long chains.
6. Reconsider chatbot design — high-turn bots that ask many questions become expensive fast. Fewer, smarter prompts win.
7. Watch quarterly rate updates — Meta will publish new rates every quarter starting September 2026.

Teambox's Take

We're updating Teambox to give you per-message cost visibility ahead of October 1, 2026, so you can see which conversations are eating margin. Our AI agent will also be tuned to resolve inquiries in fewer turns, cutting service-message volume automatically.

For high-volume workflows we're exploring keeping the AI cost outside the WhatsApp bill — our Claude Connector already works that way: the AI runs on the user's Claude subscription, not on your metered WhatsApp API.

Source: Meta for Developers — Non-Template Message Pricing