How to scale a Meta campaign without killing your ROAS
Why ROAS drops when you raise your Meta budget, and how we scale step by step: budget, creative, new markets and watching the whole funnel.
The campaign runs well for a week or two, you decide to double the budget, and the next day ROAS is cut in half. It’s the most common story we see when scaling on Meta, and it can usually be avoided.
Why ROAS drops with a bigger budget
There are three causes, and they often act together.
First, the learning phase starts over. Meta trains each ad set until it collects about 50 optimization events (purchases, for example) within 7 days. A big change to budget, audience or creative restarts that learning, and results get worse for a while.
Second, the audience runs out. On a small budget, Meta shows the ad to the people most likely to buy. On a bigger budget it also reaches less interested people, and the cost per sale goes up.
Third, creative wears out. The same people see the same ads more often. Click-through drops and each click costs more.
Vertical scaling: more budget, gradually
We don’t double the budget in one go. We raise it in steps, usually around 20% every few days, and watch how cost per sale reacts. If it holds, we keep going. If it jumps, we pause for a few days.
One more rule: we don’t change budget and creative at the same time. If results drop, you need to know which change caused it.
Horizontal scaling: more creative and more markets
Often the safer path is giving the campaign more material.
New creative: new openings for winning videos, new static banners, new formats. A banner modeled on the best performer but leading with a different benefit often has a lower cost per sale than a completely new idea.
New markets. A product that sells in one country often sells in the next one over. Translation is where most campaigns lose. So we make a base translation, rewrite it the way a local would say it, and then combine the best of both. Calls to action, the form of address and the tone are adapted to the market.
Testing and scaling in separate campaigns
We test new creative in campaigns with ad set budgets (ABO), where every variation gets a fair chance. Winners move into Advantage+ campaigns, where Meta allocates the budget itself. That way tests don’t eat into the budget of the campaigns that bring in sales.
Watch the whole funnel, not just ROAS
When ROAS drops, we first check where exactly the drop is:
- how many people reach the page;
- how many start checkout;
- how many complete it.
If clicks stay the same and orders fall, the problem isn’t the ad. It could be a slow page, a changed price, a size that sold out or a payment issue. Then we look at the page, including heatmap recordings, to see how far people scroll and where they stop.
Don’t react to one bad day
Meta reports sales with a delay, and results vary from day to day. We make decisions on at least three to seven days of data. The most expensive scaling mistake is switching off a campaign that just had a bad day and restarting its learning from zero.
