Why does my ROAS drop when I increase budget?

The marginal buyer is not the average buyer

ROAS is an average, and averages hide the thing you need to see.

When you raise the budget on an ad set, Meta has to find more people, faster, inside the same audience. It does that by relaxing toward people less likely to convert — because the most likely ones are already being reached. The last euro you added performs worse than the first euro you were already spending.

Your reporting blends the two together, so you do not see a cheap segment and an expensive one. You see a single number drifting down and it looks like the account is degrading.

Why raising budget slowly only delays it

The standard advice is to increase budgets 20% every few days so you do not disturb delivery. That advice is about volatility, and on volatility it is fine.

It does nothing about depletion. Pacing changes how fast you arrive at the ceiling, not where the ceiling is. If a pocket contains a certain number of reachable buyers, spending faster gets you there sooner and spending slower gets you there later. Either way you arrive.

This is why brands that "scale carefully" still hit the same wall, just a few weeks further out, and conclude that the account is fragile.

Vertical scale versus horizontal scale

Vertical is more budget through the same angle. It works until the pocket runs thin, and then every additional euro buys a worse impression than the one before it.

Horizontal is more validated angles running at once, each addressed to a different pocket. Total spend rises, but no single pocket absorbs the increase, so no single pocket is pushed past its productive depth.

MER holds under horizontal scale for an unglamorous reason: you added buyers rather than pressure.

What to watch instead of ROAS

Ad-set ROAS across a budget change tells you very little, because the budget change is what moved it. Three things are more useful.

• Blended MER at the account level. The question is whether the account got more efficient overall, not whether one ad set did.
• CPM and CTR on the ad set you scaled. Rising CPM with falling CTR means you have reached the edge of that pocket. This is the same signature as ordinary saturation, arriving early because you forced it.
• Downstream lift. A new top-of-funnel angle does not only produce its own sales, it refills your retargeting pool with people who did not previously know you existed. Check whether retargeting conversion rate and overall conversions improved while the angle was live. An angle can look mediocre in isolation and still be paying for itself.

How to add budget without buying worse impressions

Put it somewhere new rather than on top of the winner.

• Fund a new angle in its own ad set rather than raising the winner's budget.
• Budget each new angle at roughly 2–4× your target CPA per day. At a €50 tCPA that is about €150/day.
• One ABO campaign, one ad set per angle, ads grouped underneath. This stops Meta from quietly funnelling budget to whichever angle is cheapest on day one, before any of them have been properly tested.
• Leave it 3–7 days without edits. Most decisions are clear by day five.
• When you do want more from an existing winner, extend it into new formats and funnel stages aimed at the same pocket rather than simply raising its budget.

A test that tells you which situation you are in

Raise the budget on one winning ad set by 30–50% and hold it for a week.

If CPM climbs and CTR falls, you were already near the edge of that pocket and the answer is another pocket, not more money in this one.

If CPM and CTR hold and CAC holds with them, you genuinely had headroom, and the earlier ROAS drop was pacing rather than depletion.

That is a cheap week, and it settles an argument that otherwise runs for a quarter.