A calculation that looks reasonable
Last month, a shop spent VND 50 million on ads and brought in VND 400 million in GMV. ROAS was 8. The natural next question: so if we spend VND 500 million, will we bring in VND 4 billion?
On paper, this linear math makes sense. But markets don’t scale linearly. Advertising follows a curve, and the further along you go, the flatter the curve gets.
The saturation curve
Every additional dong spent yields a lower return than the one before it. This is called the saturation curve: steep at first, then gradually flattening.
| Monthly ad budget | GMV from ads | Average ROAS | Marginal ROAS |
|---|---|---|---|
| VND 50 million | VND 400 million | 8.0 | - |
| VND 100 million | VND 650 million | 6.5 | 5.0 |
| VND 250 million | VND 1,050 million | 4.2 | 2.7 |
| VND 500 million | VND 1,400 million | 2.8 | 1.4 |
Illustrative example, hypothetical figures.
The budget grows 10 times; GMV grows 3.5 times. There’s nothing unusual about that. The marketing measurement tools that Google (Meridian) and Meta (Robyn) have released as open source both model advertising as a saturation curve, not a straight line.
The linear calculation fails because it takes the efficiency of the first VND 50 million, the steepest part of the curve, and assumes that efficiency holds at every budget level.
Why the curve bends
- High-intent buyers are reached first. The first VND 50 million usually reaches people who are searching for exactly that product and are close to buying. Expanding the budget forces the algorithm to reach “colder” audiences who are less ready to buy.
- Bids rise as you scale. To get more impressions, the shop has to pay more for the same keywords, or expand into broader, less relevant ones.
- Ad fatigue sets in. The probability of conversion on the fifth impression is rarely higher than on the second.
- Demand in any given month is finite. The number of people who actually need the product in a month is limited. Ads can capture existing demand, but they cannot manufacture it.
Average ROAS and marginal ROAS
These are two different metrics, and confusing them is the root cause of most misguided decisions to scale the budget.
- Average ROAS = total GMV from ads / total budget. This is the number you usually see in reports.
- Marginal ROAS = additional GMV / additional budget. It tells you what the additional spend brought back.
Look back at the table, at the VND 250 million level. Average ROAS is 4.2, which still looks good. But the extra VND 150 million over the VND 100 million level brought in only VND 400 million of GMV, meaning marginal ROAS is down to 2.7. The blended figure still looks healthy, while the added budget may no longer be profitable.
Decisions to raise the budget must be based on marginal ROAS, not average ROAS.
The table above uses GMV from ads to make the shape of the curve easy to see. When you calculate it for your own shop, use the shop’s total GMV, not the ad GMV the marketplace attributes. Marketplace attribution can claim credit for orders that would have happened anyway, for example a customer who already intended to buy and clicked the ad only because the product appeared first.
A stop threshold is discipline, not a lack of ambition
Before raising the budget, you should know the ROAS below which ads start losing money. A quick estimate:
Break-even ROAS ≈ 1 / contribution margin before ad spend.
Example: after cost of goods, marketplace fees, payment fees and shipping subsidies, every VND 100 of GMV leaves VND 35, or 35%. Break-even ROAS is about 1 / 0.35 ≈ 2.9. Any part of the budget with a marginal ROAS below 2.9 is paying VND 1 to get back less than VND 1. In the table above, the extra VND 150 million at the VND 250 million level, with a marginal ROAS of 2.7, has already fallen into this zone.
Illustrative example.
Setting a stop threshold in advance, and scaling back the additional budget when marginal ROAS falls below it, is not a lack of ambition. It’s financial discipline: refusing to buy revenue at a loss.
When raising the budget is the right call
The saturation curve doesn’t mean you should never raise the budget. Raising it is right when:
- Marginal ROAS is still above the threshold. The shop hasn’t reached the flat part of the curve yet.
- The curve has just shifted up. The product has gained many more good reviews, more people search for the exact brand name, the product page is more persuasive. The same budget now brings back more.
- Peak season. During major sale events or the category’s peak season, demand rises temporarily, and the curve temporarily shifts up with it.
Sliding along the curve and shifting the curve
There are two ways to get more out of advertising, and they are very different.
- Sliding along the curve: raising the budget on the same curve. The further you go, the less each dong brings back.
- Shifting the curve up: making every budget level bring back more. This doesn’t come from the ads themselves; it comes from what the ads amplify: the trust buyers see on the page, brand familiarity, a clear reason to choose.
Most “budget doubling” plans merely slide along the existing curve. Most lasting change comes from shifting the whole curve up. The article Advertising doesn’t create demand, it amplifies it describes the layers beneath advertising and how to find the weakest one. For brands that sell at a high price and can’t cut it, shifting the curve means moving up the brand axis of the brand-price matrix.
Calculate your shop’s marginal ROAS
- Pick two comparable periods: same season, with no major sale event in only one of them. Two consecutive months are usually enough.
- Record the ad budget and the shop’s total GMV for each period.
- Calculate: marginal ROAS = (new GMV - old GMV) / (new budget - old budget).
- Compare it with the break-even threshold calculated above.
The result is only an estimate, because other variables also change between the two periods: prices, promotions, competitors. For a cleaner read, run a controlled incrementality test: raise the budget 20-30% for two weeks, keep everything else unchanged, then recalculate.
Conclusion
Scaling the budget 10x rarely yields 10x the revenue, but that doesn’t mean the ads are failing. They are doing exactly what they are designed to do: amplify what the brand already has. To get more out of advertising, a brand needs to build up more for it to amplify.