Pricing and margin
How do you know when to stop a campaign that is losing money?
The four numbers that decide it, why ROAS on its own is misleading, and what a holdout group tells you that attribution never will.
How MarketWar handles it
Break-even ROAS is computed from your contribution rather than assumed at a round number, and there are 2 thresholds rather than 1: the ratio at which spend equals contribution, and the lower one at which spend starts eating the margin you said you would protect. A campaign between them is profitable and still breaking the rule you set, which is the state nobody has a number for.
Most campaigns are stopped too late, and the reason is that the number people watch — ROAS — cannot tell them when to stop.
Why ROAS alone misleads
Return on ad spend compares revenue to spend. It says nothing about what the revenue cost you to deliver.
At 70% contribution, 2:1 is comfortable. At 25% contribution, 2:1 loses money on every order. The same number, two opposite decisions.
The two thresholds
Break-even ROAS, where spend equals contribution. Below this the campaign is losing cash outright.
Minimum permitted ROAS, where spend equals the acquisition pool — contribution minus the margin you protect. Between the two, the campaign is technically profitable and is eating the margin you told yourself was untouchable.
Most businesses only have the first number, which is why the protected margin quietly disappears.
The four things worth watching
Cost per acquisition against the ceiling your margin supports. Refund rate, because a refunded sale is a paid acquisition with no revenue. Fraud rate. And the trend in conversion rate against its own baseline, which turns early.
The one that changes decisions
A holdout. Hold back a comparable slice of the audience, show them nothing, and compare.
Attribution tells you which touchpoint came before a sale. Only a holdout tells you whether the sale needed it. The gap between attributed and incremental performance is often large, and almost always in the direction nobody wants.
What stopping should look like
A stated threshold, decided before the campaign starts, that triggers a stop rather than a discussion. A campaign that has to be argued about is a campaign that runs another fortnight.
Related: how much you can afford to pay and products that cannot carry a commission.
The threshold to write down before you start
One number and one date: the cost per acquisition above which this stops, and the day you check it.
Both decided while nobody is invested. A campaign judged after it has run is a campaign judged by somebody who wants it to have worked.
What to do instead of stopping
Not everything failing needs to be killed. In order of what to try:
Cut the audience, not the budget. Most underperforming campaigns are spending across people who were never going to buy.
Change the offer before the creative. A weak offer with beautiful artwork stays weak.
Fix the page it lands on. Run the audit — sending paid traffic to a page that does not convert buys a more expensive version of the same nothing.
Only when those have been tried does the number decide it. And when it does, it decides — that is what writing it down beforehand was for.
What this does not do
It measures what it can count. Brand effects and long consideration cycles are real and largely invisible to this, which is why the kill switch reports what it measured rather than declaring the campaign worthless.
Common questions
What is a good ROAS?
There is no good ROAS without a margin. 3:1 is comfortable on a product with 70% contribution and catastrophic on one with 25%. Compute break-even from your own numbers before comparing to anybody's benchmark.
How do I know if my ads actually caused the sales?
Hold out a comparable group and compare. Attribution tells you which touchpoint preceded a sale; only a holdout tells you whether the sale needed the ad. The difference is often large and usually unflattering.
When should a campaign be stopped rather than tuned?
When the measured cost per acquisition is above the ceiling your margin supports and the trend is not moving, or when the refund and fraud rates have made a profitable-looking campaign unprofitable in cash.
Related
How much can you afford to pay an affiliate?
The arithmetic that decides an affiliate rate: contribution, the margin you protect, and the ceiling above which a sale costs you money. With the numbers worked through on a £100 product.
Why some products cannot carry a commission at all
Thin-margin products cannot fund a percentage, however small it looks against the price. What to do instead of quietly cutting the rate — and why a headline rate that shrinks is worse than no programme.
What should you pay a creator who has no followers?
Most programmes turn away anyone under 10,000 followers and lose the person who was about to be big. The two-door model: 0.5% with no gate at all, 0.75% and 1% for a verified audience.