Short answer
Spend and revenue rising together fits a world where your ads work. It also fits a world where they do nothing. Budgets go up during good periods — busy seasons, launches, strong demand — so the two lines rise together either way. Only changing spend on purpose, or modeling the baseline properly, tells them apart.
Spend goes up. Revenue follows. The two lines hug each other for four quarters. It feels like proof.
Story one: the spend drove the sales. The chart fits.
Why does that chart prove nothing?
Because the spend was not set randomly. It was set by people who knew what was coming.
- Budgets go up before the busy season. Sales were going to rise anyway.
- Budgets go up around launches and promotions, which move sales by themselves.
- Budgets go up when business is good. Good performance caused the spend, not the reverse.
- Budgets get cut when things are tight, which is usually when demand is soft too.
What evidence actually settles it?
| Evidence | How strong |
|---|---|
| Spend and revenue rose together | None |
| Platform-reported conversions | Weak — double-counted across platforms |
| Model with a real baseline, tested on hidden months | Strong |
| Match-market test | Strong — measures cause directly |
| Repeated tests at several spend levels | Strongest |
A model gives you every channel but rests on assumptions. A test gives you one hard number. Together, the test checks the model and the model extends the test.
What changes in a budget meeting?
- 01Say the claim exactly: "channel X returned Y at the margin, not on average."
- 02Say where it came from, and its error on data the model never saw.
- 03Say what would change your mind, and when the next test reads out.
Is this just "correlation is not causation"?
Applied to a very common case. What makes it sharp is that the correlation is not an accident — spend is deliberately timed to demand, which manufactures it.
Does more data fix it?
No. More data measures a biased relationship more precisely. The fix comes from variation you created, not volume you collected.
Our agency shows a strong model fit. Different?
Only if it was tested on months the model never saw, and the baseline was built properly. Otherwise it is the same problem in better clothes.
Find your wasted third
Want this run on your own numbers?
The two-week audit shows you where each of your channels sits on its own curve — what to cut, what to grow, and how much budget is sitting in the wrong place.
Book the two-week auditKeep reading
- How to find the point where your channel stops paying offWhat data you need, which method fits your budget, and how to check the answer against the real world before you move money.
- What a 3.87% error on unseen data actually provesOne of our models predicted unseen months with 3.87% error. What that proves, what it does not, and why almost nobody publishes the number.