The same customer enquiry can now cost more to handle through the WhatsApp Business API. Since October 1, 2026, Meta has charged for service messages and utility templates delivered inside the customer service window, subject to free allowances and other exemptions. The change is documented in Meta’s current pricing guide.

The Leadsales article behind this topic highlights the pressure this creates for sales and support teams. This guide turns that concern into an operating plan, including the monthly free tier and the distinction between messaging permission and free delivery. Sources were checked on October 26, 2026.

The 24-hour window still controls when you can reply

The WhatsApp API 24-hour window is a customer service rule. Meta explains that a customer message or call opens the window, and another such interaction resets the timer. You can send non-template replies while it remains open. Once it closes, reaching out requires a template. Appropriate opt-in is still required.

For a sales manager, this creates two separate decisions: whether the team may send a free-form reply and whether delivery is billable. An open window answers the first question. It no longer guarantees free delivery. Sending another business message does not replace the customer interaction needed to reset the timer.

Make the last customer interaction and window status visible to everyone handling the thread. Before scheduling a follow-up, the owner should know which format will be available when it goes out. A useful workflow joins the response deadline, the approved template options and the cost record.

What became billable in October 2026

Meta distinguishes ordinary service replies from messages generated by its own Business Agent platform. A person or a third-party AI can produce a service message. Using AI therefore does not automatically place a reply in the Meta Business Agent category, as the official update makes clear.

• Service messages: per-delivered-message charging began on October 1, 2026, after the free tier is used and outside applicable exemptions.
• In-window utility templates: charging also began on October 1, 2026, subject to applicable exemptions.
• Meta Business Agent: token-based charging began on August 1, 2026. The published rate is USD 2 per million input and output tokens, with cached tokens excluded. Its charge covers processing and delivery.

These dates and billing rules come from Meta’s pricing update. Service rates match the market’s utility and authentication rates, but service messages have no volume tiers. Each message receives one Meta category and one corresponding charge. Do not add both a service fee and a Business Agent fee to the same reply. Your software supplier or external AI provider may charge separately.

The free allowances you should include in your plan

Every business phone number receives 1,000 free service messages each month. Unused messages do not roll over, and billing begins with service message 1,001. Meta’s monthly free tier policy describes the reset and eligibility rules. This is an allowance for messages, not customer conversations, and it does not cover utility templates simply because they are sent during an open service window.

A different exemption applies to eligible conversations started through ads that click to WhatsApp. The current developer documentation says this free entry point window may stay open for up to seven days. The customer must message from the ad using the mobile app, and the business must reply within the customer service window with an eligible message to activate it. Marketing, utility, authentication and service delivery are free while that window is open; Meta Business Agent remains chargeable. See Free entry point window for the conditions.

Keep the free delivery window separate from the customer service window. A pricing exemption does not itself permit non-template replies after the service window closes. Ad spend also remains part of the acquisition budget, so assess the full customer journey before changing where enquiries come from.

Build a forecast from delivered messages

Start with delivered message volume rather than inbox thread count. Meta charges by delivery, message category and recipient market, as explained in its pricing documentation. A business serving customers across several countries should separate those markets before estimating costs. A blended rate can conceal changes in the mix.

• Break down delivered service messages by business number and recipient market, then reconcile the free tier and exemptions actually applied.
• Keep marketing, utility and authentication templates separate, with their applicable rates and tiers.
• Add CRM or provider subscriptions, external AI usage and the team’s operating costs.
• Compare the forecast with actual charges and investigate differences in category, delivery status or billing period.

The working calculation is billable delivered messages by category and market × the applicable rate, plus the other costs of providing service. Ask the supplier to identify the billable volume rather than subtracting allowances without understanding how they were allocated. Meta also exposes pricing analytics and billing information in delivery notifications for teams with technical support.

Make each reply useful without cutting service quality

Our recommendation is to inspect repetitive journeys first. A greeting, a separate acknowledgement and another message asking for the order reference may be easier to handle as one clear reply. Group information that belongs together, keep it readable on a phone and make the next step explicit.

Give each enquiry an owner and preserve context at handoff. Customers should not have to repeat an order issue because the bot and the human agent use different records. An automated assistant should gather useful details, answer within its knowledge and escalate when it cannot move the case forward. Extra conversation is not a useful outcome by itself.

Judge the change through resolution, reopened cases, satisfaction and progress towards the customer’s goal. A shorter exchange can reflect better information, but it can also mean someone gave up. Review actual threads alongside cost data before deciding that a new automation is improving the experience.

What to check with your team and provider now

First, confirm that the account can continue delivering replies after its free tier runs out. Meta states that service delivery stops beyond the allowance if the messaging account has no payment method. Then ask for a breakdown showing categories, recipient markets, business numbers and the exemptions applied.

Clarify every additional fee and who bills it. If your workflow uses an AI assistant, establish which platform powers it and how its usage appears on the invoice. Compare options using the same service scope, including supervision, escalation and human handling. A cheaper delivery line alone does not establish a cheaper support operation.

Higher WhatsApp API costs are a reason to review the service process. Document how enquiries arrive, what customers need and what a successful interaction should accomplish. That gives marketing, sales and support a shared basis for adjusting budgets and automation while protecting the brand experience.