ABO vs CBO for Meta ads, and how to use both to scale
You don't have to choose between ABO and CBO. Run an ABO campaign for testing, with each batch of 5 to 10 new ads in its own ad set budgeted for around 50 conversions a week, and a CBO campaign for scaling that holds your proven winners. Keep the testing budget flat, move winners across each week, and only raise the budget on the CBO.
Manu Verma, founder of Nuver Digital22 October 2026 · 4:52 video · 6 min read
4:52Key takeaways
- ABO puts you in control of the budget, which testing needs. CBO puts the algorithm in control, which scaling needs. Run both, at any budget.
- Each testing batch gets its own ABO ad set with 5 to 10 genuinely different ads: different angles, messages and formats.
- Work the testing budget back from your CPA so the ad set can reach around 50 conversions a week. At a $30 CPA, that's $210 per day.
- Use push delivery to send 5% of the ad set's budget into each ad over seven days, so every ad gets real data. We don't recommend it on low-budget accounts.
- Keep the testing budget flat. Its job is to find winners, not revenue. Move winners into the CBO each week, and put iterations of proven winners straight into scaling.
- The CBO can hold up to around 50 ads. Clear out ads with little spend when new winners move in, and raise the budget about 20% every two to three days.
One of the most common questions we get is whether to use ABO or CBO. It’s the wrong question. They both have a job, and if you pick one and ignore the other, you’re either wasting your testing budget or capping your scale. This is how we run them together across the brands we work with, from a few hundred dollars per day up to 8 and 9 figures a year.
What’s the difference between ABO and CBO?
CBO (campaign budget optimisation) means the budget sits at the campaign level, and Meta decides how to spread it across your ad sets and ads.
ABO (ad set budget optimisation) means the budget sits at the ad set level, and you decide.
That’s it. CBO gives the algorithm control. ABO gives you control.
Should you use ABO or CBO?
Both, and that difference is exactly why.
- Testing needs control. Every new concept has to get enough spend for a clean read.
- Scaling needs the algorithm. Once you’ve got proven winners, Meta is better than you are at deciding where the next dollar goes.
So the structure is an ABO campaign for testing and a CBO campaign for scaling, regardless of your budget. Your budget changes how much you can test at once, not the structure.
How do you set up the ABO testing campaign?
Every batch of new ads goes into its own ad set. Keep each batch to 5 to 10 ads, and make them genuinely different:
- Different angles
- Different messages
- Visually different, with a mix of statics, carousels and videos
Ten variations of the same hook on the same static isn’t ten tests. It’s one test, ten times.
How much budget does a testing ad set need?
Work backwards from your CPA. Meta wants around 50 conversions a week for an ad set to learn. That’s about 7 conversions per day, so multiply your CPA by 7 to get the daily budget.
| Your CPA | Testing ad set budget |
|---|---|
| $30 | $210 per day |
| $50 | $350 per day |
| $125 | $875 per day |
The $125 CPA example comes from one of the brands we work with.
Only add a second testing ad set when you’ve got another one of those budgets available. If you split what you have across five ad sets, nothing exits learning and you get no read at all.
If that number is more than you can spend right now, the structure doesn’t change, only how much you can test at once. Our guide to scaling Meta ads on a small budget is the entry point: one CBO, plus a smaller fixed testing ad set where Meta decides where the spend goes, until you can work up to the budgets above.
What is Meta’s push delivery, and why does it matter for testing?
The biggest problem with testing has always been that some ads never get any spend. We’ve seen batches where 24 ads went in and only 3 got meaningful spend, so you learn nothing about the other 21.
Push delivery is a Meta feature that fixes this, and it’s now rolled out to all accounts. It’s set at the ad level, and it lets you tell the algorithm to push a share of the ad set’s budget into each ad over a seven day period, for example 5% per ad.
So with 10 ads, half the budget gets spread evenly across all of them, and the algorithm allocates the rest on performance. Every ad gets real data, and the algorithm still backs the winners.
Push delivery is available in every account, but we don’t recommend it on low-budget accounts. At a smaller budget, let Meta decide where the testing spend goes, as in our guide to scaling Meta ads on a small budget.
It gets more powerful with an attribution platform on top, because you can see which ads drive first clicks, which drive last clicks, and which contribute in the middle of the journey. Winners become obvious. (That depends on your tracking being complete: see fixing your tracking before you scale.)
How long should you test before moving winners into the CBO?
Give each batch about a week for the first real read. Then:
- Kill the clear losers.
- Let the next batch compete against whatever survived.
- Move the winners into the scaling campaign. This happens every week.
One rule we hold on every account: the testing campaign budget stays flat. Its job is to find winners, not revenue, and Meta has confirmed that to us directly. You scale the scaling campaign, not the testing one.
Do new versions of a winning ad need to be tested again?
No. Only new concepts need testing.
If you take a proven winner and change the hook, the creator or the first three seconds, that’s an iteration. It goes straight into the scaling campaign, with no retest, because the concept is already validated. This one distinction saves a lot of testing budget.
How many ads can a CBO scaling campaign hold?
More than you’d think: up to around 50 ads, all of them proven winners. Meta rotates through them as individual ads fatigue, which is part of what keeps the campaign stable.
Over time you’ll notice some ads getting very, very little spend, and it’s okay to turn those off. Low spend is the algorithm telling you the other ads in the campaign are currently better.
The exception is fresh winners you’ve only just moved in. They can take a couple of weeks to start picking up spend, so give them that time before you judge them.
For example, say you’ve moved four weekly batches into scaling. You’d turn off the low spenders from batches one and two, because they’ve had their time, and leave batches three and four alone.
When should you turn ads off in the scaling campaign?
The natural time is when you’re moving new winners in. Each time a batch comes across, turn off:
- The ads that aren’t getting spend (other than the fresh winners)
- Anything that’s had plenty of spend and is clearly missing your CPA target
New winners in, dead weight out, and the campaign stays focused on what’s working.
Beyond that, judge performance at the ad set level, not the ad level. If the ad set is hitting your CPA target, leave it alone, even the ads that look expensive on their own, because some of them are assisting everything around them. Only cut an individual ad on performance when the ad set is missing its target and that ad is clearly dragging it down.
How fast should you increase the CBO budget?
When the scaling campaign is performing, raise its budget by around 20% every two to three days, rather than doubling it overnight. The testing budget stays where it is.
What else decides whether this works?
Getting this structure right is important, but it only performs as well as what you feed it: the creative going into your testing batches, the offer behind them, and whether your tracking is giving the algorithm clean data. The whole framework runs on complete data, and most brands’ tracking is missing 20 to 30% of it. Start with fixing your tracking before you scale, and see the two-campaign account structure for how testing and scaling sit together in the account.
Questions people ask
Should I use ABO or CBO for Meta ads?
Both, because they do different jobs. ABO keeps the budget at the ad set level, so you control it, and that's what testing needs. CBO keeps the budget at the campaign level and lets Meta decide where it goes, and that's what scaling proven winners needs. We run an ABO testing campaign and a CBO scaling campaign regardless of budget.
How much budget should a Meta ads testing ad set have?
Work backwards from your CPA. Meta wants around 50 conversions a week for an ad set to learn, so multiply your CPA by 7: at a $30 CPA that's $210 per day, and at a $125 CPA it's $875 per day. Only add a second testing ad set when you can give it that same budget.
What is push delivery in Meta ads?
A Meta feature, set at the ad level, that tells the algorithm to push a set share of the ad set's budget into each ad over a seven day period, for example 5% per ad. With 10 ads, half the budget is spread evenly across all of them and the algorithm allocates the rest on performance, so every ad you test gets real data. It's available in every account, but we don't recommend it on low-budget accounts.
Should I increase the budget on my Meta testing campaign?
No. Keep the testing campaign budget flat, because its job is to find winners, not revenue. Scale the CBO scaling campaign instead, by around 20% every two to three days.
How many ads can you put in a CBO campaign?
More than most people think: up to around 50 proven winners. Meta rotates through them as individual ads fatigue, which helps keep the campaign stable. Ads that end up with very little spend can be turned off, except fresh winners, which need a couple of weeks to pick up spend.
Manu Verma
Founder of Nuver Digital, an ecommerce paid ads agency in Sydney. Close to a decade in paid ads and more than 300 ecommerce brands. On YouTube he breaks down what's working in Meta, Google and TikTok ads, pulled straight from the accounts Nuver manages.
