Why did my ROAS drop when I scaled spend?

Is a ROAS decline while scaling actually a bad sign?

Not by itself. Every account has a pool of the cheapest, most responsive impressions, the people already primed to buy your thing. Spend a little, and the algorithm finds those people first. Spend more, and it has to reach further into the audience: people slightly less ready, slightly less warm, slightly more expensive to convert. ROAS falling as spend rises is what that reach looks like in the numbers. It's arithmetic, not a verdict on the creative.

The mistake is treating any decline as proof something broke. The better question is whether the decline is steeper than the spend increase justifies, and whether the account's total contribution, revenue minus spend, is still climbing even as the ratio comes down. A ROAS that falls from 55x to 22x while spend nearly doubles can still mean more dollars in the bank than a flat 30x at half the spend.

What does this actually look like inside a real account?

The Moveable Coops account is the clearest version of this available here, because it's a genuinely strong account and its own numbers still do this. Across five consecutive weeks, weekly ROAS ran 71.7x, then 32.7x, then 49.2x, then 55.3x, then 21.9x, while weekly spend climbed from $224 to $861. That's not a smooth curve down. It goes up, down, up, up, down, while spend rises almost the whole way.

This account's lifetime badge is 36.0x, and the same account had a week that closed at 21.9x, less than a third of that badge. Both numbers are true. The badge is the retargeting layer's lifetime figure; the weekly table is the same layer's actual week-to-week movement underneath it. If a single bad week meant the account was failing, this one would have been declared broken in week two.

Why shouldn't I read the Jun 17 week the same way as the Jul 15 week?

Because $224 of spend and $861 of spend aren't the same kind of evidence. A 71.7x on $224 is built from a small number of purchases; a couple of orders shifting either direction would move that ratio dramatically. A 21.9x on $861 rests on nearly four times the spend and is a steadier read of what the audience is actually doing at that budget level. Comparing them as if they're equally reliable numbers is the error, not the direction either one moved.

This cuts against the instinct to panic at any single-week dip, and it cuts against the instinct to celebrate any single-week spike. A strong small-spend week tells you less than a strong large-spend week, and a weak small-spend week tells you almost nothing at all.

So what should you actually check when ROAS drops on scale-up?

Two things, and in this order. First, whether the drop is proportionate: is spend up by a similar amount to how much ROAS came down, or has ROAS collapsed far faster than spend grew? Second, whether total contribution, the actual dollars of revenue over spend, is still positive and still growing at the new budget level. If both hold, a falling ROAS number can be exactly what a healthy scale-up looks like. If contribution is falling too, that's the real signal, and the ratio was just where you noticed it first.

Neither check requires new tooling or a benchmark you don't have. It requires pulling your own weekly spend and revenue side by side, the way the table above does for one account, and reading the two lines against each other instead of reading the ratio alone. Most dashboards show ROAS as the headline number precisely because it's one figure instead of two, which is also why it's the easiest one to misread in isolation.