What a customer costs in Vietnam, and why the number keeps moving

A regional review opens with one slide: cost per acquisition by market. Vietnam sits at the wrong end of it, and the discussion that follows is about Vietnam.

It is usually about the slide. Two markets belong on the same chart only if both numbers were built the same way, and in Vietnam they rarely are. What follows is not a defence of the figure. It is a list of the places where the two numbers separate, so that the next version of the slide compares something real.

The two numbers in that comparison are not the same number

An acquisition cost is a fraction. Both halves of it are decisions, and both get made locally.

The denominator is whichever event the team agreed to count. An order placed. An order delivered. A payment collected. A registered account. A qualified lead. Each is defensible, and each produces a different figure from identical activity.

The numerator is whichever costs the team agreed to include. Media alone. Media and agency fee. Add marketplace commission, affiliate payouts, livestream fees, production. Each is defensible too.

The error at a regional review is not that one market picked wrong. It is that nobody wrote down which pair each market picked.

And the direction of the error surprises people. If another market counts a completed card payment while Vietnam counts a confirmed order, the Vietnamese denominator is inflated with orders that will never become revenue. Vietnam then looks cheaper, not more expensive. That version of the mistake gets celebrated rather than investigated, and budget moves toward it.

Where the Vietnam figure gets built differently

Four local habits widen the gap more than a foreign team expects.

• Payment on delivery. A confirmed order is not yet money. Everything upstream of the doorstep has already counted the sale; the bank account has not. The acceptance rate is not uniform, either — it varies by the source that produced the order.
• Marketplace sales are recorded inside the marketplace. Spend goes in there and revenue comes out there, and neither reliably lands in the same sheet as the website numbers. A company can run two acquisition engines and only see one of them properly.
• A large share of buying ends in a conversation. Cost attaches to the channel that started the chat. Revenue attaches to nothing at all unless the conversation is linked to the order record. Channels whose job is to open conversations then look weak, and budget drifts away from them for a reason that is not real.
• Affiliate and livestream payouts sit outside the ad account. They are genuinely variable acquisition costs. Leave them out and the figure understates; pull them in from two systems at once and it double-counts.

None of this is disorder. Each system is reporting accurately on what it can see. The work is in rebuilding what each number excludes and stating it once.

One month, counted four ways

The numbers below are invented to show the arithmetic. They are not drawn from any company, survey or study, and they are not a benchmark. Media spend is set to 100 so the rows can be compared to each other rather than to anything external.

Same month. Same campaigns. Same country. The last row is eighty per cent above the first, and every row is arguable.

So the question at a regional review is never which row is correct. It is which row the other market is standing on.

Why the number moves even when nothing is wrong

Three forces move the headline figure in Vietnam without any change in how well the work is being done.

The sale calendar is fixed and everyone shares it. Tết, and the repeating double-digit dates through the second half of the year, concentrate demand and competition into the same days. Cost per result climbs into those days and falls after them. A month containing one of them is not comparable to a month that does not, and comparing them anyway produces a quarterly story that never happened.

Channel mix changes the unit being measured. Move budget toward a channel that closes in conversation and reported cost per acquisition rises, because part of the revenue stops being visible. Move it back and the figure improves. Neither movement is performance.

Acceptance rates drift with source mix. If the sources that grew this month are the ones with a lower delivery acceptance rate, cost per collected order rises even though cost per confirmed order was flat. On a chart with one number per market per month, none of this is visible.

The ceiling you can calculate, and the benchmark you cannot

There is no reliable national figure for a reasonable cost per lead, and it is worth being precise about why rather than treating it as missing data. Such a figure would have to hold deal value, margin, closing rate and channel mix constant. None of those is constant even inside one category, let alone across a market.

What is calculable is your own ceiling, this month, from your own records: the average value of an order you actually collected, multiplied by the rate at which a lead becomes a collected order. That is the most you can pay for a lead and still break even before your other costs.

The word doing the work there is *collected*. Run the same arithmetic on orders placed and the ceiling comes out above what the business can afford, and the error is largest exactly where volume is easiest to buy.

Two limits on that calculation are worth stating plainly. The margin input is a figure your finance team supplies, not one that marketing oversight derives. And independent oversight can rebuild profit after advertising cost; it cannot rebuild true profit, because that needs cost of goods, and cost of goods sits outside the scope of marketing work. A structured audit can establish the first and should be honest about the second.

When "not scaling" is the market, and when it is the setup

The two have different shapes, and the shapes are easy to tell apart once the data is split.

A market limit looks like this: cost rises roughly in proportion to spend, it rises across every channel at once, and it stays risen after the sale calendar has passed. That is auction competition against a finite audience for the current offer. More budget into the same offer will keep buying worse marginal results.

A setup problem looks different: cost rises in one place while the rest holds. One source degrades and the average follows it. Or reported cost rises while collected revenue per order is flat — which is a counting change wearing the costume of a market change.

On a monthly chart the two are indistinguishable. Split the same chart by source, restate it on collected revenue, and hold the definition fixed for one full cycle, and they stop resembling each other at all. Where an agency is running the channels, this is also the reporting the oversight relationship should produce without being asked twice.

Leads arrive and do not convert: where to look first

When volume is healthy and closing is not, there is an order to the checks that saves most of the argument.

1. Split closing rate by source. Usually one or two sources are pulling the average down while the rest behave normally. That makes it a targeting question, not a sales-team question, and it changes who owns the fix. 2. Measure response time. In a market where much of the purchase happens in a conversation, the interval between a form arriving and a human answering is one of the largest and cheapest variables available. 3. Check that follow-up happens where the lead lives. A lead generated inside a chat thread and followed up by email behaves like a lead that was never contacted. 4. Check whether the word changed. A broader form, a new lead-generation placement or a lower-friction offer all raise volume and lower quality at the same moment. Cost per lead falls; cost per closed deal rises.

That fourth check is the one most often skipped, and it is the reason an imported cost-per-lead target can do quiet damage. A target set in another market is easiest to hit by changing what counts as a lead.

The definitions to settle before the next budget review

One page, agreed once, ends most of this permanently.

• What counts as an acquisition: order placed, order delivered, or payment collected.
• Which costs are inside the figure: media, agency fee, marketplace commission, affiliate and livestream payout, production.
• Which record is authoritative when the platform and the order system disagree.
• Which revenue figure the affordability ceiling is calculated on.
• Which months are comparable to each other, given the sale calendar.
• Who is allowed to change any of the above, and where that change is written down.

It is the cheapest document in a Vietnam plan and the one most often absent. Companies still preparing to enter can settle it before there is any history to restate, which is considerably less work than settling it afterwards.

Where this work stops

MWY rebuilds how the number is constructed and states what it excludes. MWY does not run the advertising, produce the content, or take commission from any platform or agency — that independence is the reason the restated figure is worth anything.

The boundary is worth naming precisely. Profit after advertising cost is reconstructable. True profit is not, because it requires cost of goods, and that lies outside marketing scope. Any margin figure in the calculation is a number the business supplies.

Training or restructuring a sales team also sits outside this. What independent advisory and oversight connects is marketing to sales outcomes at the level of numbers: where the leads came from, how many became collected orders, and what each collected order cost to acquire.

The deliverable is narrow and specific. One written definition of an acquisition, the cost base that belongs with it, and the history restated on that basis — so that this month can be compared with last month, and with the market on the other side of the slide.