How much WhatsApp Business API costs in 2026: the napkin math

USD 0.05 for the message you start. USD 0 for everything the customer writes to you and everything you reply within the next 24 hours, whether that's three messages or fifty. That pair of numbers is the entire pricing model of WhatsApp Business API since July 2025, when Meta stopped charging per open conversation and moved to charging per template sent.

The practical consequence is counterintuitive, and it orders the whole decision: your bill doesn't depend on how many people write to you, it depends on how many people you write to first. A business that receives 800 inquiries a month and never starts a conversation pays less than one that receives 40 and sends three notifications to each customer.

That's where the estimate comes from, and it's two numbers you already have: how many distinct customers you serve per month and how many times per month you start the conversation with each one. Multiply one by the other and by 0.05 and you have your monthly bill in dollars. That 0.05 is set high on purpose — it's the marketing rate, and the typical mix of an SMB averages quite a bit lower — so the result comes with a cushion. For most, it lands between USD 15 and USD 150 a month.

Most of the budgets still circulating assume the old model, where every open conversation had a price and the fear of the bill was reasonable. That model is over, and so is the intuition that came with it.

Whoever opens the conversation decides who pays

There are exactly two ways an exchange can start on WhatsApp Business, and the price changes depending on which one it is.

If the customer writes first — because they tapped a button on your site, scanned a QR code at the counter, clicked on your Instagram profile or replied to an ad — a 24-hour service window opens. Inside that window you write whatever you want, in free text, with no templates and no cost per message. Ten messages or fifty, it costs the same: zero.

If you write first, or if the window has already closed, you can't send free text. You have to use a template pre-approved by Meta, and that template is billed one by one according to its category. Two categories matter for an SMB: utility, the operational notifications about something the person already bought or booked, and marketing, everything that offers something they haven't bought yet.

The orders of magnitude, rounded off: a utility template in LATAM and Spain is around USD 0.02 to 0.03; a marketing one, between USD 0.05 and 0.09. These are typical ranges, not fixed rates: they vary quite a bit by country and Meta updates them, so you check the precise number once against the current rate card for your market. You don't need it for the math that follows.

The napkin math: two numbers and one multiplication

Number A: how many distinct customers you serve per month. Don't estimate it, count it where it already is: this month's orders in your store, appointments on the calendar, new rows in your spreadsheet, new contacts in the business WhatsApp. It's a number any owner has at hand in two minutes.

Number B: how many times per month you write first to each of those customers. This is the one nobody has counted and the one that decides the entire bill. Don't count conversations: count the moments when you are the one starting. The notice that what they ordered has arrived. The reminder. The confirmation. The due date. The monthly promo.

Monthly Meta bill ≈ A × B × 0.05

An optical shop with 350 customers a month that sends two notifications — "your glasses are in, come pick them up" and "your obra social (the Argentine health plan) approved the coverage" — comes to 350 × 2 × 0.05 = USD 35. Since both are utility and that rate runs around 0.025, what it will actually pay is closer to USD 17. The 350 conversations that come in when someone asks whether they take their plan, sends a photo of the prescription or asks whether the order is ready: USD 0.

That 0.05 is set high on purpose. In a normal mix — almost everything operational, the occasional campaign — the real average per message falls between 0.025 and 0.04, so the napkin gives you a number with a cushion. The cushion runs out when marketing goes past 55% of volume: that's where the real average crosses 0.05 and the math starts falling short. If you send more promotional than operational, redo the multiplication with 0.09 instead of 0.05.

The same business, three notification policies, three different bills

The table below doesn't compare industries: it compares decisions. It's always the same business, with the same 500 customers a month, changing only how many times it starts the conversation. These are examples built with the formula, not measured clients.

Three things that read better in a column than in prose.

The first row is the business many owners believe is expensive, and it comes to USD 5. If your operation lives off inbound inquiries and you almost never start one, the Meta bill is accounting noise: there's no decision to make there.

Between the first and the second row the bill multiplies by seven and the number of customers didn't move at all. The only thing that changed is number B. Every new notification you add to your flow is a permanent line of cost, multiplied by your entire base, every month. It's usually worth it; but it's a budget decision, not an implementation detail.

And in the third row is the number that doesn't add up for anyone at first glance: adding a single message a month adds USD 35, almost the same as what all 1,500 operational notifications cost together (USD 38). One message against fifteen hundred, and they weigh the same. It's not an error in the math: it's that the promo goes to the whole base at the marketing rate, and the marketing rate is almost triple. Before debating whether to send the campaign once or twice a month, note that this single decision moves half your bill.

What changed in July 2025 isn't the price: it's the price of following up

Under the old model, everything that went out inside the same 24-hour window was charged once. The reminder, the follow-up to whoever didn't reply and the second nudge all fell into the same conversation and the same charge. Not anymore: every template is a line. The cost of a sequence stopped scaling with the number of people and started scaling with people × steps.

That produces two copywriting rules that sound contradictory and aren't.

A longer message is free. One more message is another message. You aren't charged by character. Splitting a notification into two templates — one with the information and another with the clarification — exactly doubles that line of the bill without adding anything. If the content fits in a single template, let it fit in a single template: the greeting, the information, the instruction and the sign-off.

And the follow-up has to be thought about separately, because now it's paid for separately. A second message to 500 people who didn't reply costs on the order of USD 12 at the utility rate. It pays for itself by recovering a single mid-ticket customer, so by arithmetic the follow-up always says yes. And when the math always says yes, the math has stopped being the criterion.

The ceiling on follow-ups isn't the price: it's how many people block you. Meta watches that — blocks, reports, complaints about templates — and what it cuts when things go badly isn't your budget but your daily recipient limit, which is the one thing money can't buy. How that thermostat works is in why Meta blocks WhatsApp numbers with a bot. For budgeting, it's enough to know this: Meta's rate is so low that it stopped being the brake, and the business is left without the brake that price used to provide.

One extra line of text changes the category and triples the price of the notification

This is the expensive mistake, and it's about copy, not configuration.

Hi Ana, your glasses are in and you can pick them up
at Sucre 2340, Monday to Saturday from 9 a.m. to 7 p.m.
Bring the ID of the person named on the prescription.

That message is utility. It talks about something Ana already bought. It costs around USD 0.025, and it could be twice as long without costing a cent more.

Hi Ana, your glasses are in and you can pick them up
at Sucre 2340, Monday to Saturday from 9 a.m. to 7 p.m.
Bring the ID of the person named on the prescription.
Take advantage: this month we have 30% off sunglasses.

That message is marketing. It costs close to USD 0.07, almost triple, because of one line you added thinking it was free since it was going inside a message you were going to send anyway. And you don't pay the difference once: you pay it across the 700 notifications that go out that month. At the optical shop in the example, that turns a USD 17 bill into a USD 49 one, without having sent a single campaign.

The decision rule, to apply today: if the message mentions something the person hasn't bought yet, the whole thing becomes marketing. There's no such thing as a cheap mixed message. So separate them: the operational one goes out clean, with no offer attached, and can be as detailed as you like; the promotional one goes out on its own, with its own frequency, to the part of the base it belongs to. Sticking the offer onto the operational notification means paying the expensive rate on exactly the messages with the highest volume, which are the only ones the person was going to read anyway.

The biggest lever isn't the rate: it's getting the customer to write first

Since inbound messages cost zero, every mechanism that gets the person to write to you instead of waiting for your message pushes the bill down and improves the conversation along the way. A direct WhatsApp link in your email signature and in the site footer, a QR code at the counter and on the receipt, the button on the Instagram profile, the notice that instead of saying "we'll let you know when it's ready" says "write to us and we'll confirm it for you". Everything that gets resolved inside that 24-hour window — quoting, sending photos, closing the sale — doesn't add a cent.

And that produces the least obvious consequence of the whole model: if your team replies the next day, the window has already expired and to resume contact you have to pay for a template. Replying late stopped being just one less sale: now it's a line on your Meta bill. It's the only service metric that shows up explicitly in the cost.

Meta's bill is almost never the biggest line in your budget

If your napkin came to between USD 15 and USD 150, you already know something important: Meta's rate isn't the variable that decides whether this is worth it for you. This post covers a single line of your budget, the one you pay Meta per message. What it costs to build and sustain the whole operation — the platform, the BSP, the AI model, the setup, the support — is another calculation and it's in how much a WhatsApp chatbot costs, which also compares doing it yourself, with a freelancer or with a partner.

The order of magnitude, so it doesn't take you by surprise: in a WhatsApp operation that works for real, what Meta charges is usually a minor fraction of the total. The rest is the platform and, above all, the hours of the person who replies, which don't change price because WhatsApp changed its billing model.

That's why the useful question isn't how much a message costs, but how many messages you'll be able to answer inside the window with the people you already have.

Where the napkin stops being enough

The math works as long as number B is your decision: messages that go out because there's a schedule or a calendar behind them. It stops working at three specific moments.

When notifications are triggered by an event and not by a date. The message that goes out because a shipment hasn't moved in 48 hours, or because a payment came due yesterday, can't be counted by eye: it doesn't depend on you, it depends on data someone has to be watching. There B stops being a stable number and becomes a function of your operation; how that flow is built is in automating shipment tracking on WhatsApp.

When inbound volume exceeds what one person can answer within 24 hours. As a rough reference, someone who handles WhatsApp on top of another job starts leaving windows expired somewhere around 20 to 30 new conversations a day; the real number depends on your team and on how similar the inquiries are to each other. What breaks first isn't the budget: it's the schedule. And every expired window is paid for twice, in a template and in a sale. The volume threshold above which automating is worth it is worked through in how many inquiries justify a chatbot.

When you start segmenting for real. Sending the whole base the same monthly promo is what makes a single message account for half the bill, as in the third row of the table. Separating by behavior — who bought 60 days ago, who abandoned the cart, who hasn't come back — lowers spend and raises response, but it no longer holds up with a spreadsheet and an afternoon.

All three have the same shape: someone has to be watching a piece of data — a stalled shipment, a 24-hour clock running, a customer who hasn't come back — and deciding whether a message goes out. A napkin no longer holds that up.

Redo the napkin when the first bill arrives

The first time, the math is done by eye, because you have nothing else. The second time it's done differently, and that's the one that matters: when Meta's first statement arrives, open it next to your napkin and compare a single column, the one with the categories.

There you'll see, itemized, how many of your sends were billed as marketing and how many as utility. There's almost always a surprise and it almost always goes in the same direction: notifications you thought of as operational — the reminder that took the chance to mention a promotion, the back-in-stock notice that closed by inviting people to see the catalog — were charged at the high rate. That single line usually explains the gap between what you calculated and what you paid, and it's fixed by editing the template copy, not by changing providers.

Repeat the comparison a quarter later with the real B instead of the estimated one. If the gap between napkin and bill narrowed, your notification policy is under control and you can stop looking at this budget line for a good while. If it widened, your rate didn't go up: your operation changed, and it's probably one of the three things above.

If you'd like us to look at that comparison with you, send both things — your math and your bill — to info@striqtech.com.