There is no single right way to structure Meta campaigns. This is the setup I use, and I’ve seen it work for other marketers up to roughly $10,000–$15,000 in daily spend. If another structure is working for you, I wouldn’t change it. If you want to try this approach, start with a small test budget.
I’d spend more time on creatives than account structure. With Meta’s Andromeda and GEM updates, your creative helps determine who sees your ad. Test different hooks and angles, daily if you can. At the creative level, track hook rate, hold rate, CTR, CPC, and ROAS to see what’s working and what needs another iteration.
With that in mind, here’s how I set up my campaigns.
Campaign setup
I run one campaign for each country. It is a campaign budget optimisation (CBO) campaign, so the budget sits at the campaign level and Meta decides how much to spend across its ad sets.
Each campaign has two kinds of ad sets: one for winners and others for experiments. Every new creative starts in an experiment ad set. The experiments test new ideas; the winners compete for spend once they have shown they can convert.
Testing creatives
I usually put 25–30% of the daily budget into experiments. The rest goes to winner ad sets.
For an experiment, I plan around 4,000–5,000 in daily spend when our average CPA is 1,000, all in the same currency. In a CBO campaign, I use the ad set’s minimum-spend setting to give the test room to run; the budget itself still sits at the campaign level.
I normally let the experiment run for three days, putting the planned test spend around 12,000–15,000 in this example. Check the data daily, though. Sometimes a winner appears early. Sometimes an ad has spent enough to make the decision clear.
We usually test three to five ads in each experiment ad set. That gives Meta several chances to find a winner without forcing the full budget through one creative.
The trade-off is speed. Once Meta finds one promising ad, it tends to spend most of the budget there. Copy that ad into a winner set and pause its original in the experiment so the remaining ads get a shot. An ad that has barely spent is still untested; it is not automatically a loser.
Teams with larger experimentation budgets can use ad set budget optimisation (ABO) with one ad per ad set. Each ad then has its own allocated budget, so you can learn about each creative faster. That costs more because every ad gets funded. For a bootstrapped company, grouping ads reduces the amount at risk, even if finding the second or third winner takes longer.
Making the call
An ad qualifies as a winner for us when its CPA is no more than about 120% of our average CPA. If the account average is 1,000 and a new ad is getting conversions at 1,200, I copy it into a winner ad set and pause the original.
That 15–20% leeway gives a new ad some room while it is still learning. Its CPA may improve with more delivery, but that is not guaranteed.
Pause ads that have used their test spend without hitting the CPA goal. Give ads that have not had enough delivery more room before making the call.
Scaling winners
Moving an ad into Winners does not guarantee more spend. It gives the ad a chance to compete with proven creatives. If Meta finds opportunities for it, it will spend.
Different creatives can find different audiences. As long as the campaign keeps hitting its goal, I let Meta decide which winner deserves more of the campaign budget.
Increasing the total campaign budget is a separate decision. A new winner alone is not a reason to increase spend; the campaign still needs to meet its business goal as spend grows.
As a winner set fills up, I create another one. Winner 1, Winner 2, and Winner 3 can sit inside the same country campaign, with Meta allocating the campaign budget across them.