Sell it by refusing to oversell it.

The number is our proof. It is not our position.

Nobody in this category prints a dose. We will. But we should be honest with ourselves about what that buys.

We measured 440 Malaysian prebiotic ads. None of them prints a gram figure. Roughly one advertiser in 153 across Malaysia and Singapore does, and not one is a brown-rice drink. That is a real gap and we should walk into it.

But an empty space can mean two things. It can mean nobody thought of it. It can also mean somebody tried it and it did not sell. We are measuring what advertisers supply, not what buyers want — and a number with nothing to compare it against persuades a nutritionist, not a man who started reading the back of packets last Tuesday. He does not yet know what a good number looks like.

It is also free to copy. Any competitor can print their gram figure next week at no cost, and a better-funded one can print theirs next to a published clinical trial.

Practically: keep 7 g, printed on the pack as the creative hook. It is cheap, it is legal, and it self-selects the people who read labels. But run it in week one against a moment-led hook — your doctor told you to change your breakfast; here is what to drink tomorrow — and let the two compete. If the hook is the problem, we need a plan that can discover that.

One cell carries this launch. Everything else is a test.

Before any strategy, the arithmetic — because it changes what the plan should be.

97% of month-one contribution comes from one cell. And the English cell loses money at RM284 margin against a RM366 cost per order — while sitting in the plan at full weight. Any recommendation that does not confront those two facts is decoration.

What it costs before anyone pays you

There is no easy lane.

We got this wrong the first time and it is worth showing why, because the corrected version changes what each language argues.

Our first read said English buyers were shown a median price of RM253, making our RM258 entry a comfortable, at-market ask. That was wrong. The figure sitting exactly on the English median was a property listing — "Prices start from an affordable RM253k" — read without its thousands suffix.

Three more faults sat underneath it: one advertiser owned about 40% of the English price pool; 61% of all Malay price tokens were gift values, not prices (one advertiser's free-ebook manifest — Ebook — RM59); and roughly a fifth of English price mentions were not ingestible products at all, but property, furniture, air-conditioning and pet food.

English is the hardest lane for our price, not the easiest. Not one English advertiser in the sample asks RM258 or more. Forty of a hundred-and-fourteen Chinese advertisers already do. Chinese is the only lane where our price is ordinary.

The person who was told to change.

Lead with one segment. Not the one with the loudest identity — the one whose price we actually meet.

Primary — recently told by a doctor to change their diet

Roughly 1.1 million reachable. Measured willingness to pay RM3.20/day, the highest urgency of any segment we modelled, and the least price-sensitive.

The hero rung fits them and nobody else. Combo C is RM430 over 135 days = RM3.19/day — RM0.44 a month inside their stated willingness to pay. Run the same arithmetic on chronic gastric sufferers at RM2.70/day and no rung on the public ladder clears, not even the gated one.

They are also the only segment that survives the compliance wall intact. We can never connect the word gastrik to a benefit. This buyer needs no claim at all — the ad says a doctor told you to change your breakfast, here is exactly what is in the sachet, in grams, with its limits. Every word of that is legal.

And they are portable. Doctor visits happen in all three languages; pesakit gastrik is Malay identity language only.

All six segments, and what each is worth

Who we deliberately ignore in month one

Weight-managers, diabetics, caregivers and general wellness. Not because they are wrong — because below roughly RM4,000 per cell the numbers are noise, and we cannot afford eighteen underfunded cells. Two segments across three lanes gives readable results. Six gives guesses.

Five programmes, named by days.

Priced off a real, countable shelf price — not an invented "usual price".

The "was" price is real. Competitors multiply an inflated single — one prices a 1-box at RM299 above its own 2-box at RM288 purely to manufacture a 70% discount. Ours is boxes × RM86, and RM86 is a price a buyer can verify by counting. That is why our discounts read 40–50% rather than 70%: we are anchoring to something true.

Combo E is gated — offered only to someone who has finished a shorter programme. Selling nine months of an unfamiliar taste to a stranger costs more in refunds and bad reviews than the order is worth.

Chat-only rungs

The RM42 seven-day trial and the RM86 single box exist only inside a conversation, never on the cold page. In chat they raise conversion from 3.6% to 8.7%, which more than pays for the drop in average order. On a cold page they would simply pull the average down.

The same ladder, written natively three times.

Never mixed on one surface. Same prices everywhere — only the language changes, and each is written rather than translated.

Bahasa Malaysia

Formal pronoun, colloquial particle. anda, then break the sentence with je.

Line that ships with every Malay tier — Semua kotak: sama nombor batch, sama tarikh luput. Setiap kotak boleh discan sendiri.
The per-day reframe — RM3.19 sehari je. Kira kotaknya sendiri.

中文

Lead on 早餐. Never 养胃, never 菊粉 — use 益生元.

Every Chinese tier — 全部同批号、同有效期。每盒均可独立扫码验证。 Chinese buyers read 赠 / 免费 as secondary, informal, which is a risk signal on a health purchase — so the same-batch line is mandatory here, not optional.
The reframe — 9 盒原价 RM774。你付 RM430。自己算。

English

The most sceptical, label-reading lane. Show the working.

Every English tier — 9 boxes × RM86 = RM774. You pay RM430. Difference: RM344. RM430 ÷ 135 mornings = RM3.19. Set the arithmetic out with its working — this lane will do the division anyway, and doing it for them removes the suspicion that the maths is hiding something.

The naming test

Day-named beat quantity-named by 36% in simulation. It has never been tested against outcome-naming, which is what the category's largest advertiser uses. So run it — same prices, same gifts, same creative, names only:

perut tenang — a settled stomach — surfaced on its own in testing as how people describe the result. Judge on revenue per thousand reached, not conversion rate: the two move in opposite directions, and a higher-converting version sitting on a cheaper rung can still lose. Run it in Malay first — three languages × two arms at this budget is 17 orders a cell, which is noise dressed as insight.

Three lanes, three surfaces, no mixing.

A Malay buyer never sees a Chinese word. Prices are identical everywhere; only the language and the surface change.

Malay is the most WhatsApp-native lane, and this is not visible from the click data. Only 9.3% of Malay ads use an inbox button — but 97% of those go to WhatsApp, and another 20% of its website ads carry a WhatsApp number in the copy. Counting buttons alone understates WhatsApp's role in Malay by about three times.

Chinese is not WhatsApp-dominant. Of Chinese inbox ads, 14% clearly signal WhatsApp and 35% clearly signal Messenger. Chinese gets both surfaces; Malay gets WhatsApp only.

English creative dies four times faster — 38 days against 151 and 166. Budget for replacement, not just production.

Meta will not see more than half your sales.

The Malay cell — the one carrying 97% of contribution — will read 0.97× in Ads Manager while actually running 3.48×. On the numbers on the screen, it looks like the failure.

What has to exist before the first ringgit

• A unique wa.me reference per ad, so a conversation can name the ad that started it
• A mandatory order-source field in whatever the desk types into — not optional, not free text
• One reconciliation sheet joining ad → conversation → payment, reviewed weekly by a named person

Three items. None of them technically hard. All three are the difference between managing this launch and guessing at it.

The page, section by section.

One page per product, per language. Awareness → comprehension → conviction → action, in that order, with the uncomfortable part left in.

The conversation, and where it declines the sale.

8.7% of conversations become paid orders. The triage is not a formality — it is the part competitors cannot copy.

Language notes, from the ads themselves

Malay is formal pronoun, colloquial particle. We counted anda against awak across the lane and it runs 16 to 1 — but the je / lah particle appears in a fifth of all ads. So: address them as anda, then break the sentence with je. Writing awak reads wrong for this category.

"Gastrik" is not the word this lane leads with. We had assumed it was. Ranked by how often each appears: angin 33.8% · kembung 32.2% · sendi/lutut 29.8% · gula 26.2% · usus 25.7% — gastrik is eighth at 15.3%. The shared substrate is angin and kembung: folk-physiology that the whole lane assumes without explaining. It has no clean English equivalent, and a Malay ad written by a non-speaker will miss it entirely. note we cannot attach any of these words to a benefit — but they tell us how this buyer describes their own body, which is what the triage question must sound like.

Chinese uses 益生元 for the prebiotic — never 菊糖, and never 养胃, which is both claim-shaped and the shortest-lived frame we measured. Lead on 早餐, the safest and longest-compounding word available.

The desk, and the two questions nobody has answered.

Capacity is not the constraint. One agent handles about 5,040 conversations a month; at RM12,000 the desk runs at 12% load. A second agent only pays for itself from roughly RM60,000 a month. You need one person, early, trained on the IBS screen — not a team.

Build order

What we run, and what we refuse to make.

Structure — month one

This is a deliberate departure from equal thirds. Equal thirds buys three lane verdicts and zero creative verdicts — and funds a cell our own numbers say loses money. For a brand that has never advertised, creative variance is larger than lane variance. So: fund the hero properly, halve the losing lane, and spend the difference on testing two hooks against each other.

The eight formats, and what each is for

Twelve concepts, ready to shoot

Four per language, each mapped to a surface and a segment. Every one carries the dose and its limits as a single indivisible unit.

Hooks, in the language they run in

What we do not make

This list is a gate, not a preference. Each line is either a legal exposure or a measured mistake.

Seven things, and four of them stop the launch.

Decision rules, written before we have feelings about the results.

Every threshold below names which numbers and which view — because ROAS alone will point the wrong way in this account.

What gets replaced by real numbers first

Everything in this document is a prior. Nothing here is this brand's own data — you have never run an ad. In sensitivity order, replace: reply time, then cost per thousand impressions per lane, then the conversation-to-paid rate, then payment completion. The first two will be real within 72 hours.

After the sale — currently missing

Nothing in this plan runs past the first order, and adherence is our stated weak point. Four messages close that: delivery confirmation, the day-3 half-sachet check, a day-12 reorder prompt, and a finished-your-programme trigger. That last one is what makes Combo E sellable — it is gated behind "someone who finished", and right now nothing knows who finished.

If this fails, here is what the post-mortem says.

1. "Nobody answered the phone." Reply time was the dominant variable, the buying window was the morning, and the desk was not hired or scheduled before launch. Cost per order went from RM47 to RM93–153 and the best lane read as a failure.
2. "We switched off the only profitable cell." Meta saw 37 of 84 orders. Someone looked at the ROAS column in week two and killed Malay.
3. "We sold a box that did not exist." Every price assumed a 15-sachet pack that was still a proposal.
4. "The trade found out." We advertised 44% off shelf price direct to consumer, using retailers' own price as the "before" number, in the manufacturer's name.
5. "It did not taste good enough to finish." A third of testers reached for sugar. No simulation can answer this one.

If you say yes today.

Four decisions only you can make

1. Does the RM5.00 cost include the sachet film and fill, or only the outer carton — and is it ex-factory or delivered? Every margin here moves on the answer.
2. Will the trade carry a 15-sachet box at RM86? If retail exists, the ladder needs a channel answer before it goes live.
3. Who answers the WhatsApp, and from what hour?
4. Is the specification sheet marked "outdated" the current formulation?