The useful thing first: WhatsApp pays you back for messages people want and charges you — in money and in reach — for messages they do not. Every rule Meta has added since 2024 pushes in that direction. Design around it and WhatsApp is the most reliable channel an Indian business has; treat it as a cheaper SMS blast and you will spend six months building something that gets rate-limited. One deadline before you budget anything: on 1 October 2026, days after this is published, two categories of message that have been free since last year start costing money.
Which WhatsApp are you actually talking about?
This is where conversations go wrong first, because "WhatsApp Business" names two different products.
| WhatsApp Business App | WhatsApp Business Platform | |
|---|---|---|
| Who it is for | One number, a handful of people answering | Programmatic sending, systems integration |
| Cost | Free to download and use | Per delivered message, no per-seat fee |
| Automation | Quick replies, greeting and away messages, labels | Full API: webhooks, templates, routing, CRM writes |
| Setup | Minutes, on a phone | A business portfolio, a Messaging account, an app, tokens — or a partner who has done it |
| Regulated verticals | Not permitted | Listed countries only |
Meta's own guidance is that the app is for small businesses working from a single device, and that the Platform needs someone who can call APIs and set up webhooks — or one of its partners to do it for you. Both can run on the same phone number.
Most businesses that ask us for "WhatsApp automation" need the app configured properly and one integration, not a platform build. An off-the-shelf tool on the Platform covers a surprising amount; a build earns its keep when messages must be driven by state in your own systems rather than by a marketer clicking send.
What does Meta actually let you send?
Two rules do most of the work.
The 24-hour customer service window. When a user messages or calls you, a 24-hour timer starts. While it is open you can send free-form messages — ordinary text, whatever your agent or bot wants to say — and a further user message resets it to 24 hours. Once closed, you can only send pre-approved templates. (Pricing documentation, where the window is also defined.)
Templates and their categories. Anything sent outside the window is a template Meta has reviewed. There are three, and the category decides both the price and how much rope you get:
- Utility — a follow-up to something the user did: order confirmations, shipping notices with tracking, a survey about a specific order, an outage alert for affected postcodes. It must be non-promotional and tied to the user's order, account or transaction.
- Authentication — one-time passcodes, nothing else. The most restrictive: templates come from the Template Library, carry a copy-code or one-tap button, and cannot contain URLs, media or emojis.
- Marketing — everything else, including anything ambiguous and any template that mixes an order update with an offer.
That last clause is the one people get wrong. Meta's categorisation rules treat mixed content as marketing, and unclear content — a template that is just a variable, or a bare greeting — as marketing too. Review takes up to 24 hours.
What does it cost now?
Conversation-based pricing is gone. Meta replaced it with per-message pricing on 1 July 2025: you are charged per delivered message, by category and recipient market, and never for messages users send you.
Then the free tiers began closing. Service conversations became free in November 2024, utility messages inside an open service window in July 2025. Both reverse on 1 October 2026:
| Message type | Until 30 Sep 2026 | From 1 Oct 2026 |
|---|---|---|
| User messages to you | Free | Free |
| Service (free-form reply in the window) | Free | First 1,000 per number per month free, then charged |
| Utility, sent inside an open window | Free | Charged |
| Utility / marketing / authentication, outside the window | Charged | Charged |
The 1,000-message allowance is per business phone number per month, does not roll over, and resets at midnight in your Messaging account's timezone. With no payment method on file, Meta delivers service messages up to the free tier and then stops — hence its instruction to add one by 30 September 2026. Eligible government and non-profit organisations keep free service messages until the end of 2027.
On India rates, I am going to be unhelpfully honest. Meta publishes India's per-message amounts nowhere public: both the developer pricing page and the commercial pricing page send you to rate cards behind a login. Get the current INR rate card from WhatsApp Manager or your provider, in writing, before you model anything. Anyone quoting India rates from memory is quoting last quarter.
Two India facts I can verify and that matter. Meta raised India's marketing rate on 1 January 2026 and its authentication-international rate on 1 April 2026. And INR billing is no longer optional: every Messaging account in an eligible portfolio must migrate to INR by 31 December 2026, or Meta stops delivering non-INR messages from 1 January 2027. If you onboarded through an international partner, check your billing currency now, not in December.
One useful lever: click-to-WhatsApp ads. When someone messages you from such an ad on the Android or iOS app, your reply is free and opens a free entry point window — desktop and web do not count. That makes paid media and WhatsApp cheaper together than either alone, with a caveat on the window's length below.
What does opt-in actually require?
Less than people fear, and more than people do.
Meta requires the person's number and their permission, and that your opt-in makes two things unmistakable: that they are agreeing to hear from a business, and which business. You must also comply with applicable law. Since November 2024 the opt-in can be general rather than WhatsApp-specific. You choose the method — Meta lists a website, SMS, an IVR flow and a signed paper form — and you are solely responsible for it being lawful. (Opt-in guidance.)
Meta's recommendation, which is also ours, is to collect opt-in per category: order updates are not the same promise as offers. It costs one extra checkbox and it is the cheapest insurance available against the quality-rating spiral below.
The Indian context that genuinely bears on this: the Government notified the Digital Personal Data Protection Rules, 2025 on 14 November 2025, with an eighteen-month phased compliance period and a requirement for a separate consent notice in plain language stating the specific purpose. Enforcement is not imminent; the design decision is. A per-category consent record is far easier to build into the form today than to retrofit into three years of contacts. This is a summary, not legal advice.
Which automations actually earn their keep?
In our experience four of them, and they are all boring.
Order and status updates. The highest-read, lowest-risk thing on the platform, because the user is waiting for it. Utility category, triggered by state in your own system rather than by someone remembering.
Appointment and payment reminders. Clinics, salons, anyone with a no-show rate. The value is not the message but the reply handling — a reply opens the service window, so confirmations and reschedules happen in free-form text with no template approval at all.
Document chasing. The one that pays for the build. Any process waiting on people to send something — KYC, admission documents, GST invoices, insurance paperwork — carries a hidden cost measured in weeks of staff time following up. Escalating automated chasing works far better here than on email, because the open rates are not comparable.
Lead qualification and routing. Capture the question, ask two or three qualifying questions, then hand a human a conversation that already has context. The win is the routing, not the chatbot: a bot that answers everything is a worse investment than one that answers nothing and routes correctly in thirty seconds.
Each is triggered by a state change somewhere else — an order, a booking, a missing document, a form submission. That is why these are mostly integration projects. We publish current ranges for marketing automation work separately; if the messages must read and write records, the honest comparison is against what a custom CRM costs.
What gets accounts restricted?
Four things, in rough order of how often we see them.
Miscategorised templates. Labelling marketing as utility to get the cheaper rate is the most common deliberate mistake, and Meta has a ladder for it: a written warning, then a rolling 24-hour cap on utility messages for at least seven days, then every approved utility template converted to marketing with no new ones for 7 days — 30 for repeat offences — then the same across every account in the portfolio. Templates that drift out of category are recategorised automatically, normally with a day's notice; after a misuse warning, immediately and with none.
Blasting a cold or purchased list. People block and report, the quality rating falls, templates get paused, and the account enters the enforcement ladder: warning, a 1- or 3-day template block, a 5-, 7- or 30-day full messaging block, an indefinite account lock that only a successful appeal lifts, then removal. Appeals are decided in roughly 24 to 48 hours, and some spam violations cannot be appealed at all — you wait it out.
Ignoring the per-user marketing cap. Meta caps how many marketing templates any one person receives from all businesses, and the cap adapts to their recent read rate and how full their inbox is. There is no published number. Blocked sends fail with error 131049, and retrying inside 24 hours can lock you out of that user for another 24 while quietly corrupting your delivery reporting. Messages sent inside an open service window do not count against it — another reason the window is the most valuable thing on the platform.
Prohibited and restricted verticals. The Business Messaging Policy rules out a long list at Meta's discretion, including alcohol and tobacco, drugs, medical and healthcare products, real-money gambling, dating, multi-level marketing, payday loans and debt collection. Online gambling, over-the-counter drugs and alcohol are restricted rather than banned — Platform only, listed countries, nobody under 18, never on the Business App. Read the list before you scope.
There is also a ceiling that is not a punishment: the messaging limit on unique recipients outside service windows in a rolling 24 hours. New portfolios start at 250, shared across every number, so one number can consume the lot. Verifying your business, or delivering 2,000 high-quality template messages in 30 days, moves you to 2,000; it then scales to 10,000, 100,000 and unlimited while quality holds. A campaign designed for 50,000 recipients on a two-week-old portfolio simply does not run.
What is genuinely ambiguous
Two things, which I would rather flag than paper over.
The free entry point window's length. Meta's developer documentation describes a free entry point window of up to 7 days after a click-to-WhatsApp ad conversation; its commercial pricing page says 72 hours. Both are Meta's pages and they disagree. Treat the developer docs as authoritative and reconcile against your invoice before building a sequence that depends on day five being free.
The residue of conversation pricing. Meta's own marketing pages still carry conversation-era language while the headline says per-message. If a vendor quotes you "per conversation", they are either reselling an abstraction of their own or working from old material. Ask which.
When not to do this
If your volume is low and two people answer messages on one phone, use the Business App and spend the money elsewhere. The Platform's economics reward volume and integration; below that threshold you are buying complexity.
If your list is not opted in, do not start here. Build the consent first, with a method you can evidence. The alternative is a restricted account and a phone number you cannot reuse.
If what you actually want is to push offers at people, WhatsApp is the most expensive and most fragile place to do it, and getting more so — marketing is the priciest category, India's rate rose in January, and the per-user cap exists specifically to stop you. WhatsApp is where you tell people things they are waiting to hear. And if nobody internally owns the message content, do not start: templates age and categories drift, and an estate with no owner becomes a slow quality decline nobody notices until sending stops.
Five questions settle most of this: app or Platform; does your opt-in name your business and say what people will receive, per category; which currency do you bill in; which state changes should send a message; and who watches the quality rating. Answer those and the build is mostly integration work. Fail to, and it is premature.
Tell us what you are trying to automate and we will tell you which it is — including the common answer that you need your existing WhatsApp configured and one integration, which is not a project worth our quoting for.
Bharat Vasireddy
Founder & Digital Marketing Strategist8+ years of experience in performance marketing and web development. I help businesses scale through data-driven SEO, Google Ads, and high-performance websites.