WhatsApp for B2B sales — opt-in, templates, and what converts

Meta prices the send and gives away the conversation. A template sent to a cold contact costs money and can be refused; a message sent inside an open 24-hour window costs nothing and needs no approval. That single asymmetry decides everything else about WhatsApp in a B2B motion: where your opt-in comes from, which template category you register, what the first message should ask for, and why treating the channel as a cheaper broadcast list is the fastest route to a paused template and a downgraded number.

Opt-in is the entry condition, and B2B has four honest sources

Meta's opt-in guidance is short and specific. A business "must clearly state that a person is opting in to receive communication from the business" and "must clearly state the business's name that a person is opting in to receive messages from", and the method is left to you provided it complies with applicable law. The methods Meta itself lists — a website, an SMS reply, an interactive voice response flow, a paper form — are all outside WhatsApp. You cannot collect the opt-in inside the channel you are trying to open.

In a B2B motion that leaves four sources that hold up when somebody asks you to produce the evidence:

• A Click to WhatsApp ad or a Page call-to-action. The prospect starts the conversation, which both establishes the opt-in and opens a free entry point window of 72 hours during which every message is free. This is the cleanest acquisition path Meta offers and the one most B2B teams ignore because it looks like a consumer tactic.
• A checkbox on a form the prospect already fills in. A demo request, a pricing enquiry, a webinar registration. The checkbox has to name your business and say what you will send, and it must not be pre-ticked if the recipient's jurisdiction says so.
• An existing customer or contract relationship, where the messages are genuinely operational — an implementation update, a renewal notice, a support handover. This is the strongest ground for utility-category messaging.
• A reply on another channel where you asked. "Would it be easier to continue this on WhatsApp?" in an email, answered yes, with the answer stored. Cheap, honest, and the one most sales teams find works.

What does not hold up: a phone number appended by a data vendor, a number scraped from a website footer, a number given for a different purpose entirely, and a list bought from someone who says their contacts opted in. The opt-in names your business or it is not an opt-in for your business.

Marketing, utility and authentication: the category decides both delivery and price

WhatsApp templates carry a category, and the category is the most consequential field on the object. Meta defines three: marketing, utility and authentication.

The category decides price. Since 1 July 2025 Meta charges on a per-message basis rather than per conversation. Marketing and authentication templates are charged per message. Utility templates delivered inside an open customer service window are free, and every non-template message inside an open window is free.

The category also decides whether the message is delivered at all. Meta's documentation on per-user marketing template message limits states that WhatsApp does not currently deliver marketing template messages to WhatsApp users with United States phone numbers, and that WhatsApp "may limit the number of marketing template messages a WhatsApp user receives from any business in a given period of time when they are less likely to be receptive". No number is published for that per-user limit. When it bites, Meta returns error code 131049.

Two consequences for a B2B plan, both of which get missed:

1. A US-facing WhatsApp programme cannot be built on marketing templates. It has to be built on utility messaging tied to a real transaction, and on conversations the buyer starts.
2. Category is not a labelling choice. Writing a promotional message and registering it as utility is how a template ends up rejected, or approved and then paused once recipients react to it as promotion. Meta pauses a template when its quality rating hits the lowest band — 3 hours on the first instance, 6 hours on the second, and permanently disabled on the third.

The reply is the economic event, not the send

Say this plainly because it inverts how most teams write the first message. The template's job is not to convert. It is to earn a reply, because the reply is what makes everything after it free-form and free.

Meta's own wording is that "when a WhatsApp user messages you or calls you, a 24-hour timer called a customer service window starts", during which you can send free-form service messages with no pre-approval, and "when the window closes, you can only send pre-approved template messages". Every subsequent message from the customer restarts the timer. An active conversation therefore stays open, free and unconstrained for as long as it stays active.

That reframes the metric. A template with a high delivery rate and no replies has bought you nothing except cost and quality-rating risk — and low read rates are one of the two things Meta names as a pause trigger. A template with a lower delivery rate and a strong reply rate has bought you a live channel.

So write the first message as a question with a low answering cost. Not "here is what we do", which asks for attention and returns none. Something closer to a single closed question about the thing you already know they are dealing with, answerable in one word, with your business name visible in the first line so the recipient can tell instantly who this is. The character budget is generous compared with SMS; the attention budget is not.

What a first template can and cannot say

Templates are submitted, reviewed and approved before use, and a body carries positional variables that Meta only sees as {{1}}, {{2}} and so on. That has two practical implications.

The variable has no meaning to Meta, only a sample. When you submit a template you declare an example value for each placeholder. Meta approves the shape, not the mapping. This is the most dangerous property of an imported template: a template synced out of WhatsApp Manager carries text and sample values but no statement of what each placeholder stands for, so a system that fills in the samples will send "Hi Ada" to every lead on your list and report it as delivered. Autocloz refuses to dispatch a synced template whose placeholders have not been mapped to a lead field, and names the unmapped count in the refusal, precisely because a wrong-but-delivered message is worse than a refusal — nobody re-reads a message that says it was sent.

The parameter count is a property of the text, not of your variable list. Meta renumbers gaps at submission time, so a body written with {{1}} and {{3}} becomes two positions. Supplying three parameters against a two-parameter template fails for every recipient in the campaign, because it is a property of the template rather than of the person.

What a first template should carry: your business name in plain words, one specific reference the recipient can verify, one question, and nothing else. What it should not carry: a link in the first message to a contact who has never messaged you, a pitch longer than three lines, or a variable you cannot populate for every row on the list.

Where WhatsApp fits in a B2B sequence, and where it does not

The honest placement is second or later, never first, and only against a number you obtained with consent.

A workable shape, and one Autocloz runs as steps in a single sequence: email or a call opens, because neither requires a phone number or a documented opt-in. Somewhere in that exchange you ask whether WhatsApp is easier and record the answer. From then on, WhatsApp carries the coordination — the confirmation, the document pointer, the "are we still on for Thursday" — while email carries anything that needs length or an attachment. The channel is at its best continuing a relationship and at its worst starting one.

Where it does not fit: cold first touch, purchased numbers, US-facing marketing sends, and any message you would be uncomfortable defending if the recipient asked where you got their number. That last test is not sentimental. Meta measures blocks and reports per number and per template, and a B2B contact who did not expect you has a lower threshold for blocking than a consumer does, because their phone is a work tool.

Autocloz's free plan includes WhatsApp alongside email, calling, LinkedIn and SMS on 5 users and 10 mailboxes, with a shared suppression list and a 250-message daily cap on the WhatsApp channel — start free and wire the opt-in capture before you register a single template.

Measuring a channel that reports delivery but not intent

WhatsApp gives you more telemetry than SMS and less trustworthy telemetry than it looks.

You get sent, delivered, read and failed statuses per message, plus a quality rating on the number and on each template. Read is a genuine read receipt where the recipient has them enabled, which is more than SMS offers and less than universal. What you do not get is any signal between "read" and "replied", so a read with no reply is genuinely ambiguous.

Track four things and ignore the rest:

• Reply rate per template, because the reply is the thing you are buying, and it is the number a single inbox spanning every channel exists to make countable.
• Template quality rating, checked before you scale a send rather than after. A template sitting at medium quality is on the pause ladder already.
• The number's messaging tier and its direction of travel, since a tier that stops climbing while volume rises is a quality signal.
• Block and report signals, which surface through the quality rating rather than as a count you can see directly — which is exactly why the rating is worth watching weekly.

The architecture, tiers and error codes behind those numbers go a level deeper than a go-to-market view needs, and are worth reading once before you scale.

India, and why the same playbook reads differently there

WhatsApp is a primary business channel across much of India rather than a supplementary one, which changes the sequencing but not the consent requirement — and adds a second regulator on top of Meta.

Commercial messaging sits under TRAI's Telecom Commercial Communications Customer Preference Regulations, 2018, notified on 19 July 2018 and in force from 28 February 2019, which mandate distributed ledger technology for the registration of senders, headers and content templates. Personal data handling sits under the Digital Personal Data Protection Act, 2023 (No. 22 of 2023, dated 11 August 2023), whose implementing Digital Personal Data Protection Rules, 2025 were notified on 14 November 2025 with data-processing obligations for data fiduciaries taking effect 18 months after publication.

The practical difference in a B2B motion is that a WhatsApp-first opening is culturally normal in India and still requires the same documented opt-in, so the winning pattern is to make the opt-in capture visible and early — on the enquiry form, on the pricing page, in the IVR — rather than to skip it because the channel feels informal. The wider India outbound playbook covers how that interacts with calling and SMS.

What WhatsApp will not do for you, and what Autocloz does not do

WhatsApp will not make a cold list warm. It is a permissioned channel by design, and the permission is the expensive part — the messaging is comparatively easy once you have it. Any plan that treats acquiring the opt-in as an afterthought has skipped the only difficult step.

On the product, five boundaries worth stating. Opt-in enforcement in Autocloz is off by default and switched on per workspace. When it is off, a WhatsApp step will dispatch to a contact with no recorded consent — the do-not-contact list is always checked, the consent ledger is not, unless you turn it on. Consent evidence is recorded either way, so switching enforcement on later does not lose the history, but until you do the enforcement is yours rather than the software's.

Autocloz refuses to send free-form text outside the 24-hour window and refuses any template whose status is not approved, which are Meta's rules rather than choices. It does not run Click to WhatsApp ads or build the opt-in capture surface for you; those live in Meta's ad tools and on your own site.

Number formatting is enforced rather than trusted, and for a specific reason. Meta's phone-number reference documents that when the recipient value omits the leading + and country code, the Cloud API does not reject it — it prepends the business number's own country code and delivers to whatever that produces. The request returns success with a real message id, so the CRM reports "sent" and nobody ever arrives. Autocloz normalises the number and refuses the send when it cannot, naming the lead's number as the thing to fix. That refusal is deliberate and it will cost you sends on a badly formatted list.

Template approval is Meta's decision and takes as long as it takes; nothing in the product accelerates it. And if you are comparing dedicated WhatsApp platforms against a multi-channel CRM that also carries email, calling, LinkedIn and SMS, the trade-offs are set out in the comparison with Wati, while the sender, template and window handling itself is the surface to look at first. None of this is legal advice — the rules that bind you are the recipient's, and a lawyer should confirm how they read on your facts.