Why does our ROAS drop every time we scale budget?
By Dan BartleyUpdated 3 minute read
The short answer
Because each extra dollar reaches people less likely to buy than the last. Meta spends your first dollars on the likeliest buyers; as budget rises it moves to colder ones, and if the same few ads carry the extra spend, frequency climbs and they fatigue faster. Scale with more distinct creative, in steps, and judge it on profit and blended MER rather than platform ROAS alone.
Is some drop in ROAS normal when we scale?
Yes. Average ROAS almost always falls as spend rises, because the marginal sale costs more than the average one. The question is whether the extra spend still adds profit, not whether ROAS held.
Measure the increase on its own: marginal ROAS is the change in store revenue divided by the change in spend, over comparable weeks. If marginal ROAS stays above breakeven, the scale-up made money even though the average fell.
Why does creative limit how far we can scale?
Because under Andromeda the ads themselves decide who sees them. Meta's retrieval system narrows tens of millions of ad candidates to a few thousand for each person, so more distinct creative gives it more routes to new buyers, while a handful of similar ads keeps finding the same people.
The signs that creative is the ceiling: spend piling onto one or two ads, frequency rising on them, CPM climbing as budget grows, and new ads failing to take spend because they look like the old ones.
Raising volume and quality together lifts that ceiling. For an e-commerce brand Dan ran creative for, new ads went from under 5 a month to 50 to 100, the ads got better as well as more numerous, and ROAS went from 2.5x to 5.5x while the account spent 300% more. That is one account, not a forecast.
Do big budget jumps make the drop worse?
They can. Frequent or large budget changes can send an ad set back into Meta's learning phase, which usually ends after about 50 results in the week after the change, and CPA tends to run higher until it does.
Raise budgets in steps and give each step several days of data. With Advantage+ campaign budget, Meta moves spend between ad sets itself, so the campaign budget is the lever rather than each ad set.
Does market size matter in Australia and New Zealand?
Yes. A smaller market runs out of new people sooner at the same spend, so frequency rises faster and fresh creative matters earlier. A brand selling only in New Zealand will usually hit this point at a lower budget than one selling across Australia.
Watch reach as well as frequency when you scale. If reach flattens while spend keeps rising, the extra budget is buying repeat impressions on the same people, and new creative or a new market will do more than a bigger number.
How should we measure a scale test?
On store numbers over the same period: total revenue, new customers and blended MER, alongside Meta's ROAS. Platform ROAS can fall while blended revenue rises, because Meta only credits the sales inside its attribution windows.
Agree the measures before the budget moves, and read them weekly rather than daily. There is more on reconciling the two in platform ROAS, blended ROAS and MER.
What is the safest way to scale Meta budget?
Add creative before you add budget, then raise spend in steps and check marginal returns after each one.
- Launch a batch of new, distinct concepts a week or two before the budget rises
- Raise the campaign budget in steps rather than one jump
- After each step, check frequency, CPM and marginal ROAS
- Keep winners running and add new ads beside them
- Hold, or step back, if marginal ROAS drops below breakeven
Other questions people ask about this
Should we duplicate winning ad sets to scale?
Rarely now. Duplicates chase the same people and each one restarts learning; raising the budget on a consolidated campaign and adding new creative to it usually scales more cleanly.
Read the full answerWhy does ROAS recover when we cut the budget?
Cutting spend pulls delivery back to the likeliest buyers. That shows diminishing returns, not that the higher budget was wrong: the test is whether the extra spend made profit.
Should we expand to a new country to keep scaling?
It can work once the home market is saturating. Treat the new country as a launch, with creative made for that market where it differs, and judge it separately.
Still have a question? Ask us