WhatsApp Business app vs WhatsApp Business Platform (API)

The app is underrated by people who sell the platform. For a business where a handful of people talk to customers, it does a great deal: a business profile with hours, address and website; a catalogue customers can browse without leaving the chat; quick replies for the answers you type every day; labels to sort chats by stage; greeting and away messages; and short links and QR codes that open a chat with you. It also does everything ordinary WhatsApp does, voice notes, calls, photos, documents, groups and Status. Customers get a conversation that feels personal because it is.

It stops at the phone. Linked devices let a few colleagues use the same number from WhatsApp Web, the desktop app or another device, but there is no assignment, no queue, no per-person reporting and no reliable way to know who is dealing with which customer beyond labels and habit. Two people can reply to the same customer without knowing. History sits on devices and in personal cloud backups, not in a system the business controls, and when the person who holds the main phone leaves, a good deal of operational knowledge leaves with them.

The app’s automation is a set of shortcuts for people, not automation in the software sense. There is no official API for the app: nothing your order system can call to send an update, and no notification to your CRM when a customer writes. Tools exist that drive WhatsApp Web or a reverse-engineered client to fill that gap, and they break WhatsApp’s terms. Numbers using them can be banned, often without warning, and a business that loses its number loses the channel its customers know.

Broadcast lists deserve a precise description, because they are often mistaken for a marketing tool. The app sends the same message to many contacts individually, but only people who have saved your number in their address book receive it, and the size of each list is capped. That is a deliberate spam control. It makes broadcast lists useful for customers who already know you and have chosen to keep your number, and useless for reaching anyone else.

The platform’s central rule is about who spoke first and when. When a customer messages your business, a customer-service window opens for 24 hours from their latest message, and inside it you can reply with anything: free text, images, documents, buttons, a conversation. Outside it, you can send only a message template, a pre-written message with placeholders for details such as a name or an order number, which has been submitted to Meta and approved before use. Every design decision about a WhatsApp integration starts from this rule.

Templates fall into categories: marketing, utility for updates about a transaction or account the customer already has, and authentication for one-time codes. Meta reviews each template, can reject it, and can place it in a different category from the one you chose, which matters because category decides price. A message that is promotional in substance is a marketing template, whatever its label. Templates can carry media, quick-reply buttons and links, and their quality is monitored: a template that recipients block or report can be paused or disabled.

Before you message someone who has not just written to you, Meta’s policy requires that they have agreed to receive messages from your business on WhatsApp. Opt-in can be collected on a website, at checkout, in an app, by text message or inside WhatsApp itself, and it should name your business and make clear what kind of messages to expect. Meta’s rules are a floor, not a ceiling: the privacy and marketing laws where your customers live apply as well, and for marketing they may ask more of you. Record when and how each person opted in, make opting out easy, and make sure every system that sends honours it.

Three consequences follow. Service conversations happen inside the window, so a chatbot answering a customer who has just written to you can be entirely free-form. Anything you start, a reminder, an update, a re-engagement message, needs a template, and an unexpected template is the quickest way to collect blocks and damage your quality rating. And marketing reach is not guaranteed: Meta also limits how many marketing messages an individual receives from businesses, so a marketing template can be accepted by the API and still not be delivered.

We will not quote rates. They vary by country, they are revised, and Meta has changed the model itself more than once. What is worth understanding is the structure, because it decides which of your messages cost money. Check Meta’s current rate card, and your provider’s, before you build a forecast.

Meta moved from charging per conversation to charging per message in 2025. It now charges for each delivered template message, and the price depends on two things: the template’s category and the country code of the recipient’s number. Marketing templates generally cost the most, and utility and authentication templates less. Meta has also introduced lower rates at higher volumes for some categories.

Several things are not charged per message. Free-form replies inside an open customer-service window are not, which means a support operation where customers write first can run largely without per-message charges from Meta. Utility templates sent while a window is open are not charged either. Conversations that begin from certain Meta ad formats that click through to WhatsApp have their own free period. The exact rules on each of these are the kind of detail Meta revises, so confirm them against current documentation.

Who bills you depends on the route. Connected directly to the Cloud API, Meta bills the payment method attached to your WhatsApp Business Account. Through a provider, Meta’s charges may be passed through at cost, marked up per message or bundled into a package, and the provider adds its own fees, typically some mix of a monthly platform fee, per-agent seats for the inbox, set-up and charges for chatbot features. Ask every provider to show Meta’s charges and their own as separate lines, because that is the only way to compare them.

To model it, count the messages your systems will start each month by category and by recipient country, price them on the current rate cards, and add the provider’s fees. Then add what never appears on an invoice: agent time, and the engineering to build and maintain the integration. The app costs nothing per message, so the platform has to earn its cost through agent efficiency, automation, or revenue from messages you could not otherwise send.

There are two ways onto the platform. The direct route is to connect to the Cloud API yourself: create a Meta developer app, attach a WhatsApp Business Account in your business portfolio, register a phone number, subscribe to webhooks, and write software that sends messages through the API and handles inbound messages and status updates. There is no intermediary and no provider fee. There is also nothing to use until you build it: no agent inbox, no bot and no reporting.

The other route is a Business Solution Provider, a company Meta has approved to onboard businesses and connect them to the platform. Providers range from developer platforms that expose their own API over the Cloud API, often alongside SMS, voice and email, to complete inbox products for non-technical teams with routing, chatbot builders and ready-made CRM connectors. Most offer an onboarding flow that sets up the WhatsApp Business Account and number with you, plus support when something goes wrong.

The choice usually makes itself. With no engineers, use a provider with a good inbox, because the direct route would leave you with an API and nobody to use it. If you are building WhatsApp into your own product or CRM, compare the direct Cloud API with a developer-focused provider. A provider earns its fee through one API across several channels, fallback to SMS when WhatsApp fails, consolidated billing and support. Going direct removes a layer of cost and a dependency, and puts the full integration and its maintenance on your team.

Whichever provider you consider, ask the same questions. Is the WhatsApp Business Account held in your own business portfolio or the provider’s, since that decides how easily you can leave? How are Meta’s charges billed, and at what markup? Can you export conversation history in a usable format? Does the provider support the features you need, such as templates, interactive messages, in-chat forms and connecting a number already used on the app? And what exactly happens to your number if you move to another provider?

The platform is event-driven. Every inbound message, delivery receipt and read receipt arrives at your webhook as it happens, and your software can send a message whenever it decides to. That is what makes automation possible in both directions: a customer’s message can trigger a lookup, a ticket, a CRM update or a bot reply, and an event elsewhere, an order shipped or an appointment tomorrow, can trigger a template. It is the same pattern as any other business automation, with WhatsApp as one channel among several.

Chatbots on the platform range from simple button menus, built with interactive reply buttons, list messages and in-chat forms, to language-model assistants that answer from your own content through retrieval-augmented generation. The good ones share one property: they know when to stop. A bot should hand over to a person cleanly, with the conversation so far attached, the moment a customer is frustrated, the question is outside its scope or money is involved. Replies a bot sends inside the customer-service window carry no per-message charge from Meta, but model and hosting costs still apply.

Meta’s business terms also draw a line on AI worth knowing about before you build. A business using AI to answer its own customers about its own products and services is the ordinary case. Offering a general-purpose AI assistant over WhatsApp as the product itself is something Meta’s terms now restrict. If your plan sits closer to the second than the first, read the current terms before you start.

CRM integration is the reason many businesses make the move at all. On the platform, every conversation can be logged against the contact in HubSpot, Salesforce or a CRM you built, with the agent, the timestamps, the templates sent and the opt-in status. Many CRMs and inbox providers offer WhatsApp connectors, and where they do not fit your process, a custom integration built through CRM development connects the webhook to your own records. Keep opt-in status in the CRM, so that every system that sends can check it before it does.

On the app, none of this is officially possible. If you have been relying on a tool that syncs the app with a CRM by automating it, understand what it is doing and the risk it puts on the number, and plan the move to a supported route.

On the app, customer conversations live on the phones that run it and in whatever backups their users have switched on. The business has little say over who can read them, how long they are kept or what happens to them when somebody leaves, and a lost or replaced phone is a data question as well as an inconvenience.

On the platform, messages pass through Meta’s Cloud API and, if you use one, your provider’s systems before they reach your own. Each of those is in the data path. Read the data-processing terms for each, check where processing happens and how long each party keeps message content, and make sure your privacy notice tells customers that you use WhatsApp, and any provider, to talk to them.

Once messages reach your software the decisions are yours, which is much of the point. Decide who can read which conversations, set retention periods that match your obligations, delete what you no longer need, and keep the opt-in record with the customer. In sectors with rules about recording customer communications, a platform integration that stores every message centrally is far easier to evidence than a collection of staff phones.

Be deliberate about what customers send. People share photos of identity documents, account numbers and medical details on WhatsApp because it feels private. On the platform that content lands in your systems and possibly in a chatbot’s context, so decide in advance how sensitive attachments are stored, who can open them, and whether the bot is allowed to see them at all.

Verification means several different things, and confusing them causes delays. Meta business verification confirms your legal business details in your Meta business portfolio, and on the platform it is what lets you scale beyond the starting limits on how many people you can message first. The display name customers see is reviewed against Meta’s rules and must reasonably match your business. The verified badge, the check mark beside a name, is a separate matter with its own eligibility criteria, and on the app it comes as part of a paid Meta subscription where that is offered.

Messaging limits control how many different people a business can contact first, with templates, in a rolling 24-hour period. New senders start low, and the limit rises as you send at volume with good quality. Replies to customers who wrote to you are not restricted in the same way. If you plan a launch or a campaign that needs to reach a large opted-in audience on day one, find out where your limit stands well before the day.

The quality rating reflects how recipients respond to what you send, chiefly whether they block or report you. Low quality can stop your limits rising, reduce them, or pause the templates involved. The surest ways to damage it are sending marketing templates to people who did not expect them and importing an old list of numbers that never opted in to WhatsApp messages. Treat quality as the account’s credit score: slow to build, quick to lose and expensive to repair.

Two further checks belong at the start of a project, not the end. Meta’s commerce and business policies restrict or prohibit some categories of goods and services, so confirm that yours is allowed before you build anything. And consider who will own the Meta business portfolio, its administrators and its verification documents, because a portfolio administered by a former employee or an old agency is a common and avoidable obstacle.

The number is the asset your customers know, so moving it deserves more care than any other step. For most of the platform’s history a number could be on the app or on the platform but not both, and moving it meant deleting the WhatsApp account on the app before registering the number on the platform.

That classic move has real costs. Chat history held in the app does not transfer to the platform. The catalogue, labels, quick replies and away messages stay behind and must be rebuilt in whatever software you use. The number then works only through the platform, so nobody can open WhatsApp on the phone and use it, and groups and Status do not come with it. What customers keep is the thing that matters most: the same number to message. Export any conversations you need from the app before you begin.

Meta now also offers a route, usually called coexistence, that connects a number already in use on the Business app to the Cloud API through a provider’s onboarding flow, so the app keeps working and messages appear in both places. For a small team that wants automation or a CRM connection without giving up the phone, that changes the calculation considerably. It has limits: not every app feature is supported while the number is connected to the platform, not every provider offers it, and the details are still changing. Confirm with your provider that it is available for your number, and what works under it, before planning around it.

Moving a number already on the platform between providers, or from a provider to direct Cloud API access, is supported, and you can keep the same number. You will need the number’s two-step verification PIN or the outgoing provider’s cooperation in turning it off, and you should ask in advance what carries over, such as templates and quality history, and what has to be recreated. This is where it pays to have kept the WhatsApp Business Account in your own business portfolio from the start.

Moving back from the platform to the app is possible after the number is removed from the platform, and it is the same kind of break in reverse: the history stays in your software, not in the app. Whatever the direction, do it at a quiet time, check that verification, display name and payment details are in order beforehand, and have the new inbox or bot tested and ready before the number moves, so customers who write in during the change get an answer.