WhatsApp Marketing Strategy: What Actually Works

TL;DR. WhatsApp is an owned channel that is easy to ruin. Anyone who receives an unsolicited message there can report or block the sender, and we see people use those buttons more readily than an email subscriber clicks unsubscribe. The strategy that works is to earn the right to send marketing by proving usefulness on transactional messages first, and to accept that the channel needs staffing.

Before any strategy, one honest qualification question: is your audience on WhatsApp as a commercial channel, or only as a private messenger? The answer changes the entire setup.

The economics: paying per delivered message

Since 1 July 2025, Meta bills the WhatsApp Business API per delivered template message rather than per conversation (Meta pricing). Every template belongs to a category, marketing, utility or authentication (template docs), and the rate depends on the template category, the recipient number's country calling code and, for utility and authentication templates, the monthly volume tier reached. Meta publishes a separate rate card per market, so check yours on the day you scope the programme.

Two mechanics shape everything else:

The customer service window. When a user messages or calls the brand, a 24-hour counter starts, and it resets with each new message or call from them. Inside that window, the brand can reply with free-form messages, no pre-approved template needed. Outside it, only approved templates get through (send messages docs). Direction matters: the window opens on the customer's initiative, not on the brand's.

Utility templates sent as replies. A utility template delivered inside an open customer service window is not billed (Meta pricing). A brand that answers quickly and well therefore pays structurally less than one pushing cold templates.

The strategic consequence: every outbound send has a unit cost, and every inbound conversation is a chance to serve the customer more cheaply. Build the programme in that direction.

Where the channel earns its place

Adoption of WhatsApp as a commercial channel varies by market, and the same programme does not carry the same weight everywhere. That is the first thing to settle, before choosing any sequence. We tie this observation to no public measurement: it comes from our own client work, and it should be validated against your own audience data before you commit anything.

In markets where instant messaging is a normal commercial contact channel, opt-in is easier to collect and a promotional message causes less surprise.

In France, we see two opposing effects on the programmes we open: messages read quickly, and a markedly lower tolerance for intrusion, with usage of the app still perceived as private.

On the legal side, the CNIL guidance on electronic direct marketing (the CNIL is France's data protection authority) requires, for a private individual, consent that is prior, freely given, specific, informed and unambiguous, collected through a positive action, with pre-ticked boxes ruled out. It provides an exception where the person contacted is already a customer and the marketing concerns similar products or services supplied by that same company. That exception does not apply where no sale or service has taken place, including where the customer has merely created an online account.

The guidance cites email, SMS-MMS and automated calling systems as examples, with the list ending in "etc.". It names neither instant messaging nor WhatsApp, and it says nothing about whether consent given for one channel extends to another. We draw an agency precaution from that: collect a WhatsApp-specific opt-in rather than reusing the email one. The regulatory base common to electronic channels is covered in GDPR and email marketing.

The practical consequence is that a French programme has to start with transactional and service messages. The brand earns the right to send marketing after proving its usefulness on order tracking, not before.

The channel also earns its place outside those markets when the purchase journey involves advice: bespoke cosmetics, nutrition, technical equipment, furniture, or when average order value justifies a human exchange. On a low basket with no advisory dimension, we still favour email and SMS: they tie up no advisor time, and WhatsApp adds nothing those two are not already doing.

The four sequences, in order

1. Order confirmation and tracking

This first scenario goes out as a utility template, the category Meta separates from marketing. It is the message customers most expect, and the one that builds your messaging quality, which Meta assesses across all your business phone numbers and templates. That quality is one of two cumulative criteria for automatic increases to your business portfolio's messaging limits, alongside how much of your current limit you actually use (Meta messaging limits).

Its economic value is twofold. Sent as a reply to a customer message, inside an open customer service window, it is not billed. And it is exactly the kind of message that gives customers a reason to write to the brand, which opens the windows where the exchange becomes free-form.

We let it run on its own for four to six weeks before switching anything else on.

2. Cart abandonment with a fast reply

A cart reminder works here as it does elsewhere, with one difference you can use: customers reply often on WhatsApp. They ask their question, about sizing, delivery time or compatibility, instead of walking away in silence. And their reply opens a 24-hour customer service window in which the conversation continues in free-form messages.

The reminder itself goes out as a marketing template, billed per delivered message. It becomes profitable when it triggers the reply, so the message should invite an answer, not just a click. The logic of the equivalent email scenario sits in Klaviyo abandonment flows.

3. Back in stock and pre-launch

This scenario rests on a dedicated opt-in, requested at the exact moment the customer hits the out-of-stock page. It matters most on limited-availability products, for the same reason it does in SMS: the message is explicitly expected, on a date the customer triggered themselves.

A billed marketing template is easier to justify when purchase intent has already been expressed. Two guardrails: a list per product rather than a global list, so only the relevant subscribers are notified, and a single send per restock, with no reminder follow-up.

4. Advised reactivation

A human takes over on high-value segments, with a message that asks a question instead of pushing an offer. Reserve it for VIPs identified through RFM segmentation: the combined cost of templates and human time makes the approach unworkable at scale.

It is the only one of the four sequences run by hand, in short batches, and judged on the margin it produces rather than the volume it processes.

The phased rollout

Phase 1, four to six weeks: transactional only. No promotional messages. Judge this phase on your messaging quality as Meta assesses it, not on revenue generated. It also reveals the real volume of inbound questions, which is usually underestimated at scoping.

Phase 2: opening up the conversation. Cart abandonment and back in stock, with human replies during published hours. Team sizing gets decided here, on real volumes.

Phase 3: targeted marketing. Approved marketing templates, reserved for high-value segments and low-frequency moments: launches, private sales, awaited restocks. Avoid duplicating an email campaign sent the same day. We find that a recipient who gets the same offer twice reports it more readily.

Skipping the first two phases buys a few weeks of good numbers, then a long recovery.

What the channel demands of you

An owned channel where customers reply often needs staffing. Three points have to be settled before rollout.

The response time commitment. The customer service window lasts 24 hours from the customer's last message or call. Replying inside the window allows free-form messages; replying after it closes forces a billed template. Answering fast therefore costs less and converts better.

Coverage hours, published on the channel and honoured. An unanswered WhatsApp message is expensive, because the medium implies presence.

The escalation path: what automation handles, what goes to customer service, what goes to an advisor. That boundary decides whether the channel scales or quietly consumes a full-time role.

Those three points carry a budget consequence that projections tend to miss: the channel's cost is the price of delivered templates plus the human time the exchanges generate. A scenario that produces a lot of replies can show an excellent conversion rate and destroy margin, once each exchange takes an advisor several minutes.

So the calculation to run before launch is margin per engaged customer, not conversion rate. It often lands the other way round from what was expected: quiet transactional sequences come out ahead of chatty promotional ones.

The rule that follows: open the channel on scenarios where the expected reply is short or non-existent, and keep the conversational work for segments whose margin supports it.

How to measure

Revenue per delivered marketing template, the channel's real economic unit since pricing moved to a per-message basis, compared against your cost per send on email and SMS.

Your messaging quality. Meta assesses it across all your business phone numbers and templates, and calculates messaging limits at the business portfolio level, shared by every number it contains (Meta messaging limits). It is the most direct health indicator available to you on this channel, and it should be watched as a trend rather than as a point value.

It does not replace email's complaint rate, which is quantified and tied to a threshold: Google asks senders to stay under 0.10% complaints and to avoid reaching 0.30% (Google requirements). None of the Meta pages cited here publishes a comparable threshold for WhatsApp. So you steer without an external numeric benchmark, on your own series.

The share of conversations resolved without human intervention, which decides whether the channel scales.

Two reading traps come up repeatedly.

The first is comparing WhatsApp's conversion rate to email's. The channel requires an opt-in, and the programme only has an audience among customers who gave it: the population joining it is already among the most engaged in your database. A higher rate is therefore expected, and proves nothing about the channel itself. As with SMS, the honest measurement is a holdout on the eligible segment.

The second is counting messages instead of customers. The same customer contacted three times in two weeks produces three billed templates and one order. The messages-per-customer ratio tells you whether the channel is absorbing questions the site should be answering, in which case the work sits on the product page, not on the channel.

FAQ

Does WhatsApp marketing work in France?

Yes, provided you do not start with promotional messages. We find usage there is still perceived as private, so tolerance for intrusion is low. A French programme starts with transactional and service messages, and only opens up marketing once its usefulness is proven.

How is WhatsApp billed?

Since 1 July 2025, Meta bills per delivered template message, at a rate that depends on the template category (marketing, utility, authentication), the recipient number's country calling code and, for utility and authentication templates, the monthly volume tier reached, with a separate rate card per market. Meta also states that utility templates delivered inside an open customer service window are not billed.

Which sequences should you set up first?

Order tracking as a utility template, on its own, for four to six weeks. It builds your messaging quality and gives customers a reason to write to the brand, which opens customer service windows where the exchange is free-form.

Does WhatsApp replace email or SMS?

Neither. The three split by message type: email carries long content and the catalogue, SMS carries short time-bound information, WhatsApp carries the exchange that calls for a reply. Opening WhatsApp does not excuse you from running the other two well. The trade-off between the first two is covered in SMS or email.

Do you need a dedicated team?

Not necessarily dedicated, but identified. The channel generates inbound replies, and three things have to be settled before opening: who answers, on what published hours, and at what point the exchange goes to an advisor.

Going further

The technical setup, Meta's template approval and the detail of messaging limits are a separate workstream, to open once the strategy is settled. Scoping a programme is covered on our WhatsApp marketing agency page. If you want the channel scoped against your existing database rather than bolted on beside it, book a Klaviyo and CRM diagnostic.