What a WhatsApp Automation Project Actually Involves in Malaysia

Every WhatsApp automation pitch you'll receive skips the same three things: what actually gets built, what it costs in ringgit, and which constraints will bite you in month two. This is that walkthrough — a full project from first conversation to launch, written for the person signing off on it rather than the developer building it.

The project below is representative rather than one named client — the scope, architecture and pricing are drawn from engagements we've quoted and run. Client-specific results aren't ours to publish, so this covers the mechanics and the money, which is what you're evaluating anyway.

The situation this solves

A consumer brand runs a six-month customer engagement program. New contacts join continuously — 500 to 750 a month, building toward 3,000–4,500 people. Every contact enters at a different point in their journey, and what's useful to someone in week one is useless to someone in month four.

The brand has the content. What they lack is delivery: the right message, to the right segment, at the right point in each person's own timeline, on the channel their audience actually opens. In Malaysia that channel is WhatsApp — it is the country's default messaging app, tracked year on year in MCMC's Internet Users Survey, and it reaches an audience email and mobile apps simply don't here. In our own projects it is where enquiries arrive, where customers reply, and where they expect the answer to come back.

Right now the work is manual. Someone keeps a spreadsheet, works out who should get what, sends in batches. That works at 200 contacts. It breaks at 4,000, and it breaks badly when every contact sits on a different personal timeline.

If that sounds like your operation, the rest of this is a fairly accurate preview of what you're buying.

What clients ask for versus what they need

The opening request is almost always some version of: "a system where we can set up and customise our own campaigns." It's the most common thing we hear, and it's usually the wrong thing to build — not because the client is naive, but because a flexible campaign-authoring platform costs roughly five times what solving the actual problem costs.

So we ask one question back: who operates this, and for how long?

If the answer is "our team, daily, for the next three years," then yes — a self-service console earns its price. If the answer is "you run it for us over six months," then a configurable platform is an expensive way to buy something a scheduled job and a well-structured contact database already do. Existing tools plus a few deliberate manual steps beat a custom console that gets used twice.

That single question removes something like two-thirds of the scope before anyone writes code. It's the highest-leverage conversation in the whole project, and it either happens in the first meeting or it doesn't happen at all — by month two the spec is set and you're paying to build the console.

The useful instinct: be suspicious of any quote that grows the moment you ask "who's actually going to use this?"

The five parts that actually get built

Strip away the brochure language and a program like this has five moving parts.

• A contact database that knows where each person is. Every contact carries a marker for their lifecycle position and entry date. This is the spine — segmentation is just a query against it, and everything else depends on it being clean. Importing and reconciling the client's existing contacts is the least glamorous and most time-consuming part of the build, every single time.
• Message templates, pre-approved by Meta. The constraint that surprises everyone. More on this below, because it reshapes how you plan.
• A scheduler with two triggers. Send when a contact hits a milestone on their own timeline, or send to a whole segment on a fixed date. Most programs want both, and the interaction is where edge cases live — what happens when someone hits a milestone two days before the monthly bulk send? Decide that before launch, not after a customer gets both.
• Opt-out handling. Non-negotiable. Legally under the Personal Data Protection Act 2010, and practically because WhatsApp restricts numbers that attract spam reports. Opt-outs must be honoured immediately, permanently, across every trigger path — including the one you added last week.
• Delivery reporting. What sent, what landed, what failed. Without it you're broadcasting blind and you'll discover a broken template when a customer complains, not before.

Underneath sits a WhatsApp Business API account via a provider such as Twilio, a rented number, and the integration between scheduler and API. Not the WhatsApp Business app on someone's phone — that route has no API, no scheduling, no reporting, and stops being viable a few hundred contacts in.

The Meta rule that reshapes your plan

This deserves its own section because it catches out nearly every first-time buyer.

WhatsApp only lets you message someone freely within 24 hours of their last message to you. That's the service window. Outside it — which is where a scheduled outbound program lives, almost by definition — you can only send a template that Meta has reviewed and approved in advance.

Three consequences that change how you plan:

• You can't write copy on the day you send it. Templates go through review. They can be rejected, and you rewrite and resubmit. Campaign spontaneity is not a thing WhatsApp offers.
• Every template is a small commitment. Variables are allowed, but the structure is fixed. Sixty templates across four languages isn't a content task, it's a project workstream.
• Cost follows category. Since 1 July 2025 Meta charges per message rather than per 24-hour conversation, and the rate depends on the template's category and the recipient's country. Marketing templates are charged on delivery; utility templates are free when they land inside an open customer service window; ordinary non-template replies inside that window are free. That pricing shifts, so check Meta's current WhatsApp pricing for Malaysia rather than trusting a figure in any proposal, ours included.

Anyone who tells you WhatsApp broadcast is "just like email but better opened" hasn't shipped one.

What it costs

A program of this shape prices at RM 12,000 one-time for a single-phase build covering a six-month engagement. That includes everything on our side: build, integration, server costs, platform fees, third-party subscriptions for the period.

We quote all-in deliberately. The alternative — low headline number, plus setup, plus monthly, plus support — is how projects end up over budget and vendors end up distrusted.

Two things sit outside, and here's the honest reasoning:

• The rented phone number — a small recurring fee to the API provider.
• Per-message charges — Meta bills per message delivered, and your API provider adds its own per-message fee on top. We don't mark this up and we don't absorb it, because it scales with your sending volume rather than our work. A fixed all-in price covering unlimited messaging means someone is padding heavily or planning to renegotiate.

We also don't fix message volume in the contract. Programs grow and segments split; a client who must file a change request to send one extra broadcast quietly stops using the system.

For where this sits against other automation work, the AI cost calculator breaks down the running costs, and each service page carries its own tier pricing.

What moves the price

Budgeting something similar? These are the real variables, in order of impact.

• Self-service versus done-for-you. The biggest lever by far, as above.
• Contact data quality. A clean structured list is a day's work. Scattered spreadsheets with inconsistent phone formats, duplicates and no lifecycle marker is two weeks.
• Integrations. Standalone is cheapest. Every system it must read or write — CRM, e-commerce, booking, accounting — adds real work, and vendor API quality varies wildly.
• Template volume and languages. Ten is routine. Sixty across four languages is its own workstream.
• Compliance surface. Consumer engagement is straightforward. Once health, financial or minors' data is involved, the guardrails and audit trail change the build's shape — and its price.

What usually goes wrong

Three failure modes. All boring, all expensive, all avoidable.

Template approval arrives late. The team fixes a launch date, then finds Meta rejected four templates and the copy needs rewriting. Submit templates early — before the build finishes, not after.

Contact data shows up last. The database is the spine, and projects stall waiting on a clean export from a system nobody owns. Ask for the data in week one even if you won't touch it until week four.

Nobody owns the content. Automation delivers messages; it doesn't write them. Programs die quietly when marketing runs out of things to say in month three. Agree the content calendar before building the machine that delivers it.

Does your business actually need this?

A rough test. You need all three:

• You're sending scheduled, segmented messages to more than a few hundred contacts.
• The segmentation depends on something that changes per person over time — not one flat list.
• Someone is doing it by hand right now, and it's visibly straining.

All three, and the economics are obvious. Miss the third — nobody's doing it manually because nobody needs it done — and you're buying a machine for a problem you don't have. We'd rather tell you that on the call than four weeks into a build.

If your problem is answering inbound questions rather than sending outbound campaigns, that's a different build: see WhatsApp Business API chatbot Malaysia for the conversational side, or AI chatbot Malaysia for the wider picture. If both sound partly right, that's normal — most businesses need a bit of each, and the split is worth twenty minutes on a call.

And if you genuinely don't know which problem you have, that's what the digitalisation audit is for. We look at what your team does by hand and tell you honestly whether automation is the answer or whether a cheaper change solves most of it. Sometimes the answer is a better shared inbox and two templates.