One ROAS target for a whole shop is a mistake you only notice when you grow
While a shop is not scaling, a single return target works. The moment you add budget, money flows to where things sell most easily — not where they earn. How to split targets by margin without rebuilding the account.
- Break-even return differs per category: it equals one divided by that category’s gross margin.
- With a single target, budget flows to the categories that sell most easily regardless of profitability.
- It can be deployed without rebuilding the account — through margin bands in the feed’s custom labels.
Most shops run advertising on one number: a cost-of-sale or return target for the whole account. While the budget was not growing, that was a reasonable shortcut. The moment you start scaling, it becomes the most expensive simplification in the account.
Why it stops working as you grow
An automated strategy meets the target you give it and does not care where it gets it from. When the whole account has a ROAS target of 5, the system will look for the cheapest route to five units of revenue per unit of cost. It usually finds that where things sell most easily — in cheap goods with high turnover and low margin.
In numbers it looks like this. Two categories, the same ROAS target of 5:
| Cosmetics | Large electronics | |
|---|---|---|
| Gross margin | 55 % | 9 % |
| Break-even ROAS (1 ÷ margin) | 1.82 | 11.1 |
| Single ROAS target | 5.0 | 5.0 |
| Result when the target is met | High profit, but the target is needlessly strict — it holds growth back | A loss on every order, even though the target is met |
So one category is braking needlessly and the other is quietly losing money. Both are meeting the brief. The report shows a ROAS of 5 and everything looks fine.
At a small budget the money sits in the cheapest auctions and the differences do not show. The more you add, the more it matters where the system moves the marginal unit — and it sends it where the target is easiest to meet. This connects directly to marginal return: the gap between the average and the edge widens as the budget grows.
How to calculate a target for a category
- Find the category’s gross margin. After deducting cost of goods, shipping, packaging, payment fees and returns. Returns are the most commonly forgotten and in fashion they change the answer fundamentally.
- Calculate break-even: ROASbreak-even = 1 ÷ margin. At a 35 % margin that is 2.86.
- Add headroom. For fixed costs, management and fluctuation. How much depends on what share of costs advertising should cover — commonly the target sits a fifth to a third above break-even.
- Factor in commercial intent. For a category where you want to grow share you may deliberately go closer to break-even. For slow movers, stricter.
How to deploy it without rebuilding the account
The good news: you do not need a campaign per category. Three or four margin bands are enough.
- Split the range into bands. For example high margin (over 40 %), medium (20–40 %), low (under 20 %) and clearance separately.
- Write the band into a custom label in the feed. Typically
custom_label_0. See the article on feeds. - Build a campaign or asset group per band, not per category. Three campaigns instead of thirty — and each has enough conversions for the strategy to learn from.
- Give each band its own target derived from its break-even.
- Update the labels when buying prices change. At least quarterly; a product can move between bands.
An even better option: send margin directly
If your shop can do it, send gross margin on the order as the conversion value instead of its price. Then you need neither bands nor different targets — the system optimises directly for profit and prefers the profitable range on its own.
There are two conditions worth weighing honestly:
- You have to know the margin at the moment of the order, at line-item level. For shops with shifting buying prices that may not be trivial.
- All your numbers change with it. ROAS will no longer be 5 but perhaps 1.8 — because it is counting margin, not revenue. The whole team and the board have to know, or there will be confusion.
A commonly used intermediate step: send revenue excluding VAT and shipping. It is not margin, but it is closer to the truth than the gross order total.
What else to consider
- New versus returning customers. A first order may cost more than a second. If you can distinguish them, set a different target for acquisition — Performance Max has a new-customer goal for exactly this.
- Returns. In fashion easily in the tens of per cent. A margin calculated without them is fiction and the target derived from it is too loose.
- Free shipping over a threshold. It reduces margin on orders just above the line. Worth including in the calculation.
- Repeat purchase. For consumables the argument for a looser acquisition target is legitimate, but it has to rest on measured data, not an estimate. See LTV and its traps.
What to take away
A single cost-of-sale target is not a mistake while the shop is not growing. The moment you start adding budget, that neutral number becomes a decision: you are sending money where things sell most easily. Three margin bands in the feed and three different targets is a few hours of work and usually the fastest way to raise profit without raising spend.
Common questions
How do I calculate a cost-of-sale target for a category?
Break-even cost of sale equals the category’s gross margin, and break-even ROAS is one divided by that margin. At a 35 per cent margin the zero point is therefore a 35 per cent cost of sale, or a ROAS of 2.86. Set the target above break-even with headroom for fixed costs, management and fluctuation — commonly a fifth to a third higher. Remember to include shipping, packaging, payment fees and above all returns in the margin, as returns are most often overlooked and change the answer substantially in fashion.
Do I have to rebuild the whole account because of differing margins?
No. It is enough to split the range into three or four margin bands, write the band into a custom label in the feed, and build a campaign or asset group per band rather than per category. That gives you separate return targets while keeping enough conversions in each campaign for automated bidding to learn from. Thirty campaigns split by category would fragment the data instead.
Should conversion value be revenue or margin?
Margin is better, if you can determine it at the moment of the order at line-item level — the system then optimises directly for profit and prefers the profitable range without you having to split campaigns by band. Be aware though that all your numbers change: return will no longer be calculated from revenue, so the figures will be markedly lower and the whole team and board need to know. A sensible intermediate step is to send revenue excluding VAT and shipping.