Scaling
Scaling without burning margin: when more budget is the wrong answer
Every extra euro in Google Ads brings less than the one before it. The question is not whether your ROAS still looks good. The question is what the last thousand brought.
Published on · 7 min read
- 1Average ROAS hides what happens at the edge. What matters is marginal contribution margin: what did the last budget step bring on top?
- 2Beyond a certain point revenue keeps rising while profit falls. That point is different in every account and can only be found through step tests.
- 3Raise budget in steps of 20 to 30 percent, hold each step for at least two weeks, calculate against the comparison period.
- 4If impression share on brand searches is already high or conversion rate falls with budget, more budget is the wrong answer.
Scaling is the word that comes up in every second first call, and almost always it means the same thing: more budget at the same ROAS. That sounds reasonable and is the most common way to lose money in Google Ads without noticing. Because the ROAS you see in the dashboard is an average over everything you spent. It says nothing about what the last euro brought.
The average lies, the edge decides
Google shows your ads first where the probability of a purchase is highest. With every additional euro you move into auctions that fit slightly worse: more generic queries, less decided users, more expensive clicks. That is not a flaw of the algorithm, it is its job. But it means every budget step brings less contribution margin than the one before. At some point it brings less than it costs.
Schematic curve. It falls differently in every account, but it always falls. The red mark is the point where the last step exactly earns back its own cost.
The insidious part is that the average hides this curve. If the first 20,000 euros bring a ROAS of 6 and the next 10,000 a ROAS of 2, the dashboard shows 4.7. That looks like a healthy account. In reality you just spent 10,000 euros to sell below margin.
The calculation missing from the dashboard
That is why we never evaluate budget steps with ROAS, but with contribution margin after ad costs. That is revenue times margin, minus returns, minus media spend. This number does not automatically rise with revenue. It can fall while revenue rises, and that is exactly what happens in most scaling efforts beyond a certain step.
| Step | Budget | Revenue | ROAS | CM before ad costs | CM after ad costs |
|---|---|---|---|---|---|
| 1 | €20,000 | €80,000 | 4.0 | €28,000 | €8,000 |
| 2 | €30,000 | €110,000 | 3.7 | €38,500 | €8,500 |
| 3 | €40,000 | €130,000 | 3.3 | €45,500 | €5,500 |
Invented worked example with 35 percent margin after returns. Step 2 still just pays off, step 3 makes more revenue and less profit. ROAS looks acceptable in all three steps.
Look at the last column. From step 1 to step 2, 10,000 euros of extra budget bring 500 euros more profit. Not much, but positive. From step 2 to step 3, another 10,000 euros bring 3,000 euros less profit. The store has 20,000 euros more revenue and 3,000 euros less in the bank. And the ROAS in step 3, at 3.3, still sits above the break-even of 2.9, so no alarm goes off.
+€20,000
more revenue from step 2 to step 3
€3,000
less profit in the same step
3.3
ROAS that looks completely unremarkable
How to test budget steps cleanly
- 1
One step of 20 to 30 percent
Bigger jumps throw Smart Bidding back into the learning phase, then you measure the learning phase and not the step. Smaller jumps drown in noise.
- 2
Hold for at least 14 days
Shorter is not reliable, because weekdays and attribution windows distort. With long purchase decisions correspondingly longer, until the conversions of the step have fully arrived.
- 3
Calculate against the comparison period
Not against the previous month, but against the same 14 days before, adjusted for promotions and season. Where possible with a geo split: one region gets the step, one does not.
- 4
Decide on marginal contribution margin
The step stays if contribution margin after ad costs went up. Not revenue, not ROAS. If it fell, the step goes back, and you have found the edge of your account.
When more budget is the wrong answer
Sometimes a look into the account already shows that the next step will bring nothing. These four signals are the ones we see most often:
- Impression share on brand searches is already above 90 percent. More budget then only buys generic queries that convert far worse.
- Conversion rate falls measurably with every budget increase. The algorithm has reached the edge of the audience.
- Stock or assortment is the bottleneck. Whoever cannot restock the bestsellers is scaling sold-out messages.
- The extra revenue comes from existing customers who would have bought anyway. Visible in a new-customer share that does not grow with the budget.
The question before every budget increase
Not: is our ROAS still good? But: what did the last step bring on top, in euros after ad costs? Whoever does not know that number raises budget on a hunch. Whoever knows it knows whether the next step is an investment or a donation to Google.
And once the edge is reached? Then growth is not in the budget but in the structure: segment the feed by margin, take brand searches out of Performance Max, close tracking gaps. Each of these steps shifts the curve upwards. After that the next step pays off again, and you can fire it with a clear conscience.
Scaling does not mean spending more. It means knowing the point beyond which spending more earns less, and staying in front of it.