--- title: "WhatsApp Service Messages Now Cost Money: Rebudget" author: "Nate Laquis" author_role: "Founder & CEO" date: "2026-10-25" category: "Cost & Planning" tags: - WhatsApp API pricing - WhatsApp Business Platform - WhatsApp chatbot cost - Meta Business Agent - messaging costs excerpt: "Since October 1, 2026, Meta bills service messages past the first 1,000 a month per phone number, and utility messages inside the 24-hour window are billable too. Here is how to find your exposure and cut it." reading_time: "10 min read" canonical_url: "https://kanopylabs.com/blog/whatsapp-service-messages-billing-october-2026-startup-cost-guide" --- # WhatsApp Service Messages Now Cost Money: Rebudget ## What Changed on October 1 For almost two years, replying to a customer on WhatsApp inside the 24-hour window cost you nothing. Meta made service messages free on November 1, 2024, and a lot of support bots, order-status flows and booking assistants were built on that assumption. As of October 1, 2026, the free ride is over. Here is what Meta's developer documentation says now. Service messages, meaning the free-form replies you send inside the 24-hour window after a customer writes to you, are billable per delivered message. Each business phone number gets 1,000 free service messages a month. The 1,001st is charged. Unused free messages do not roll over, and the quota resets monthly. It is per number, so two numbers do not share a pool. Utility messages sent inside an open customer service window were free from July 1, 2025. They are charged at published rates again as of October 1. A utility message sent during a free entry point window stays free. There is a billing trap in the rules too. If your account has no payment method, service messages beyond the free tier will not be delivered. Meta asked for payment methods to be added by September 30. If you run a bot for a client or for your own product and nobody checked, you may already have silent failures. Open your WhatsApp Manager billing settings today. Two exceptions are worth knowing. Eligible government and non-profit organizations keep service messages free beyond the free tier through December 31, 2027. And AI providers are excluded from the service message pricing update. If you are a regular commercial business, neither applies to you. ![Smartphone showing a messaging app conversation on a desk](https://images.unsplash.com/photo-1611746869696-d09bce200020?w=800&q=80) ## Why Founders Get Surprised by This Most teams budget WhatsApp by the template. Marketing messages cost the most, utility and authentication messages cost less, and the support conversation that follows was a rounding error. That mental model is now wrong in a specific way: the cheap part of your volume is the part that just got a price. Think about what a healthy WhatsApp support flow looks like. A customer sends one message. Your bot or your agent replies, the customer replies again, and the thread goes six to ten messages deep before it resolves. Every one of your replies is a service message. A single conversation can be eight billable sends where it used to be zero. The 1,000 free messages sound generous until you do the math. A small clinic confirming appointments might never cross it. A direct-to-consumer brand with a busy order-status bot crosses it in the first week of the month. An AI assistant that answers in several short bubbles instead of one long message burns through it fast, because the quota counts delivered messages, not conversations. Notice what is not changing. Pricing still depends on the recipient's country, and rates differ a lot between markets. I will not quote one rate as if it applied to you. The published rate card in Meta's documentation is the source, and you should read the row for each country your customers live in. If you built your bot on a third-party provider such as Twilio, 360dialog, Gupshup or Infobip, their invoice adds a platform fee on top of Meta's charge. Ask your provider whether they pass through Meta's service message fee at cost or mark it up. Markups on pass-through fees are common, and nobody mentions them until you ask. ## A Worked Example With Your Own Numbers I am going to use an assumed rate of one cent per billable message, because it makes the arithmetic easy to follow. Replace it with the actual rate for your customers' countries before you present anything to a finance person. Say you run a customer support bot on one WhatsApp number. Customers start 2,000 conversations a month. Your bot sends an average of six replies per conversation, so 12,000 service messages a month. - **Before:** 12,000 service messages at zero is $0. - **After:** the first 1,000 are free, the other 11,000 are billable. At the assumed one cent, that is about $110 a month. That is not a catastrophe. Now scale it. At 20,000 conversations a month you are at 120,000 messages and about $1,190 a month, still with only one number. Splitting the same traffic across five numbers for separate brands would give you 5,000 free messages instead of 1,000, which trims the bill by about $40 at the assumed rate. The point is that this is now a variable cost that grows with conversations, and it should sit in your unit economics next to LLM tokens and support headcount. If your gross margin per customer was thin before, check it again. Utility messages inside the window work the same way. If your flow sends an order confirmation, a shipping update and a delivery notice as replies inside an open window, each of those now counts. Count them separately from the conversational replies so you can see which flow drives the bill. Do this in an hour. Export one month of message logs from your provider or from the Cloud API webhooks. Group by phone number and by pricing type. Meta changed the status webhook so charged service messages now report a pricing type of **regular** instead of **free_customer_service**. That field is your cheapest way to see what you are actually being charged. ## Seven Ways to Cut the Bill Without Hurting Customers Do not start by rationing replies. A support experience that feels stingy loses customers, and the savings here are small compared with a lost account. Start with waste. ### 1. Send fewer, fuller messages If your bot answers in three short bubbles, merge them into one well-formatted message. Each delivered message is a unit of billing. A well-written single reply is also nicer to read on a phone. ### 2. Stop acknowledging everything "Got it, one moment" before every lookup is a billable message that tells the customer nothing. Use a typing indicator where the API supports it and send the answer when you have it. ### 3. Kill the pointless closers "Anything else I can help with?" after every resolved question adds a message per conversation. Keep it only where it measurably prevents a repeat contact. ### 4. Use the free entry point window on purpose Conversations that start from a Click-to-WhatsApp ad or a Facebook Page call-to-action open a 72-hour free entry point window. Message delivery fees are waived inside it. If you spend on those ads anyway, route support-heavy flows through them. ### 5. Move the easy stuff out of WhatsApp Order tracking and receipts do not need a chat thread. Link to a page in your product, or send the update by email. Keep WhatsApp for the conversations where customers want to talk. ### 6. Split numbers deliberately Because the free quota is per number, a company with separate brands or regional teams gets more free messages by using separate numbers. Do not invent numbers just to farm quota. Every extra number adds a verification, a display name and an operational burden, and Meta can change the rules again. ### 7. Price it into the product If you resell a WhatsApp feature, your pricing needs a messaging line. A per-seat plan with unlimited WhatsApp conversations is now a margin leak. Include a message allowance and charge for overage, or build the fee into a usage tier. ![Founder reviewing messaging cost reports with a teammate](https://images.unsplash.com/photo-1556742049-0cfed4f6a45d?w=800&q=80) ## Meta Business Agent: Token Billing, Not Message Billing There is a second pricing model hiding in the same documentation, and it matters if you are thinking about letting Meta run the conversation for you. Meta Business Agent, introduced July 1, 2026, is billed per token at $2.00 per million tokens, input and output alike. Cache tokens are not charged. Meta's own figure is that a typical message uses 20,000 to 25,000 tokens, which comes to roughly 4 to 5 cents per message. Volume tiers and free messages do not apply, and these messages are charged even inside free entry point windows. Compare that with the service message fee. A service message is a fraction of a cent to a couple of cents depending on the country. A Business Agent message is several times that, because you are paying for the model, not just the delivery. For a bot answering a high volume of simple questions, that is expensive. For a bot closing a high-value sale or handling a complicated booking, 5 cents is nothing. My view: Business Agent is worth a trial if you have no engineering capacity and your average conversation is worth more than a few dollars. It is a poor fit if you have a high-volume, low-value flow, or if you need the bot to read your own database, apply your business rules and hand off cleanly to a human. For those cases a custom assistant on your own model, with careful context management, is cheaper per message and far more controllable. Our guide to [managing LLM API costs](/blog/how-to-manage-llm-api-costs) covers the levers: model routing, prompt caching and shorter context. Also note that 20,000 to 25,000 tokens per message is a lot. It tells you Business Agent is carrying a large amount of context on every turn. If you build your own, you can usually keep the average well below that by retrieving only what the question needs. ## Country Rates Moved Too, So Check Your Markets The same October 1 update changed rates in several markets. A summary published by a messaging provider, which I am reading as a secondary source, lists marketing rate increases in Kuwait, Morocco, Mexico, Saudi Arabia, the UAE and parts of Asia Pacific. It also lists steep jumps in utility and authentication rates in markets such as Morocco, Kuwait and Oman, and decreases in Sri Lanka, Nepal and Bangladesh. I am not going to reproduce those tables, because rate cards change and you should read the current one from Meta. What matters is the pattern. Rates moved in both directions, by very different amounts, and a company with customers in the Middle East or South Asia can see a very different bill from a company that only serves North America. There is also a new standalone authentication-international rate for a set of markets that used to sit in a regional "rest of" group. Those markets now count toward their own volume tiers. If you send one-time passcodes through WhatsApp to users abroad, this is the line item to check first, because authentication messages are the ones that tend to run at high volume without anyone noticing. Practical advice: build a small spreadsheet with one row per country where you have more than a few hundred active users. Columns for message type, monthly volume, current rate and monthly cost. It takes an afternoon and it will tell you where the money actually goes. In most cases it is one or two countries and one or two message types. If one-time passcodes are your biggest cost, consider whether WhatsApp is the right channel. SMS has its own costs and its own fraud problems, and passkeys or app-based authentication avoid per-message fees entirely. That is a product decision with a security angle, so involve whoever owns login. ## What to Do This Week You do not need a project for this. You need a short list and one owner. 1. **Add a payment method** to every WhatsApp Business Account that sends service messages, and confirm messages beyond the free tier are delivering. 2. **Pull last month's logs** and count delivered service and utility messages by phone number and country. 3. **Price the exposure** at the current rate card for each country, not at an average. 4. **Ask your provider** whether Meta's fees pass through at cost and how they show on the invoice. 5. **Trim the waste** from the seven items above. Merge bubbles, drop acknowledgements and closers. 6. **Put a line in your pricing model** if you resell or bundle WhatsApp conversations. 7. **Set an alert** on spend so a viral campaign does not produce a surprise invoice. The deeper lesson is that a platform you do not control can reprice a feature you built a business case on. It happened with WhatsApp service messages. It has happened with API tiers at social networks and with model pricing at AI vendors. When you scope a messaging feature, assume the free tier is temporary and keep the channel behind an interface you can swap. That costs a few days of engineering up front and saves a rebuild later. If you are planning a WhatsApp assistant, start with our walkthrough on [building a WhatsApp Business chatbot](/blog/how-to-build-a-whatsapp-business-chatbot), and read the cost breakdown in [what an AI chatbot costs to build](/blog/how-much-does-it-cost-to-build-an-ai-chatbot). Both assume you are paying per message, which is now the honest assumption. ## When WhatsApp Is Still Worth It A price on service messages does not make WhatsApp a bad channel. In much of the world it is where your customers already are, and a reply inside a thread they already use is a better experience than a ticket portal. A few cents a conversation is still cheaper than a human agent minute. What changes is discipline. Messaging used to be a cost you could ignore. Now it is a line in the model, like cloud hosting or LLM tokens, and the teams that measure it will make better choices about which conversations deserve a thread and which deserve a link. ![Dashboard with charts showing monthly cost trends on a laptop screen](https://images.unsplash.com/photo-1551288049-bebda4e38f71?w=800&q=80) If you are not sure whether your current WhatsApp setup is exposed, or you are scoping a new assistant and want the cost model done properly before you build, we can walk through it with your real volumes. [Book a free strategy call](/get-started). --- *Originally published on [Kanopy Labs](https://kanopylabs.com/blog/whatsapp-service-messages-billing-october-2026-startup-cost-guide)*