What's the best creative testing setup: an ABO test campaign, CBO, ASC, DCT or Meta's creative testing tool?
By Dan BartleyUpdated 4 minute read
The short answer
For most e-commerce accounts in 2026, test inside the campaign that will scale the winners, using Meta’s creative testing tool or a new ad set per batch, and leave the winners’ ad sets alone. Use a separate ABO test campaign when you need even spend per concept or a bid strategy the tool does not support. Advantage+ shopping campaigns (ASC) have become Advantage+ sales campaigns, and the flexible ad format has replaced dynamic creative.
How do the testing setups compare?
They trade clean data against keeping delivery learnings. The closer a test runs to where the winner will scale, the more of its result carries over.
| Setup | How it works | Best for | Watch for |
|---|---|---|---|
| Meta’s creative testing tool | 2 to 7 copies of an ad tested inside an existing campaign, with a set share of budget | Testing where winners will scale | Highest volume bidding only, and no confidence level |
| ABO test campaign | A separate campaign, one ad set per concept, each with its own budget | Even spend per concept | Winners moved to the scaling campaign start learning again |
| New ad set in the scaling campaign | Each batch goes into its own ad set under Advantage+ campaign budget | Adding tests without touching the winners | Budget can favour proven ad sets and starve the test |
| New ads in an existing ad set | Tests added alongside the winners | Simplicity | Adding an ad sends the ad set back into learning |
| Dynamic creative (DCT) | Replaced by the flexible ad format for new sales ad sets | Packaging the assets of one concept | Not a way to compare concepts |
How does Meta’s creative testing tool work?
You set it up at the ad level of an existing campaign: choose 2 to 7 copies of an ad, swap in the new creative and set how much of the budget goes to the test. Meta suggests no more than 20% and aims to spend that amount on the test ads each day.
Because the test runs in the campaign where a winner will keep running, delivery learnings are kept; Meta notes that moving ads into another campaign would reset them. After the test, the ads keep running alongside the existing ones until you turn them off.
Two limits matter. The tool only works with the Highest volume bid strategy, and its results name top performers without a confidence level, so set your own threshold before you call a winner.
When is a separate ABO test campaign still the right call?
When you need each concept to get the same spend, or your scaling campaign uses a cost per result goal, bid cap or ROAS goal, none of which the testing tool supports.
The cost is the move. A winner proven in a test campaign has to be relaunched in the scaling campaign, where it starts learning again and does not always repeat its result. Relaunch it in a new ad set rather than in the winners’ ad set.
How do we add tests without disturbing the winners?
Keep new creative out of the winners’ ad sets. Meta lists adding a new ad to an ad set as a significant edit that sends it back into the learning phase, while adding a new ad set to a campaign with Advantage+ campaign budget does not reset the others.
So each batch gets its own ad set or goes through the testing tool, and the winners’ ad sets are left alone. Advantage+ campaign budget is Meta’s current name for CBO, and the same logic applies inside Advantage+ sales campaigns, which have replaced ASC. Account structure as a whole has its own answer.
How much spend and time should each test get?
Enough to judge it on your target metric. A common rule is two to three times your target CPA per creative before you turn off one that has not converted, over at least a few days so weekday swings even out.
For high-priced products where purchases are rare, judge early on add-to-carts or initiated checkouts, then confirm on purchases. Set the rule before the test starts, so a good first day does not crown a winner early.
What matters more than the setup?
The creative going into it. A clean test of five near-identical ads tells you less than a rough test of five genuinely different concepts.
Plan each batch as distinct concepts, as covered in concept testing against iteration, and the setup has something worth measuring.
Other questions people ask about this
Is ABO or CBO better for creative testing?
ABO guarantees each concept its share of spend, which makes comparisons cleaner. CBO, now Advantage+ campaign budget, moves spend to early leaders, which is efficient but can starve a concept before it has had a fair test.
What replaced dynamic creative?
The flexible ad format. It holds several images or videos in one ad and lets Meta choose what to show for each person and placement, which suits the assets of one concept rather than a test between concepts.
Should test ads use the same audience as the scaling campaign?
Yes, broadly. A winner found on an audience you will never scale into tells you about the wrong buyers.
Still have a question? Ask us