Been wrestling with how to set up ad structure when testing multiple angles in ABO. Here's where I'm at:

1 campaign, 3 ad sets (each ad set = one angle)
Each angle has 3 native image variations and 3 primary text options

The part that trips me up is how many ads to put inside each ad set for clean learning.

Two schools of thought:

Option A - 1 ad per ad set (1 image + 1 copy) => Find the winning angle first, then break out image and copy tests separately.

Option B - 2-3 ads per ad set, each ad with all 3 copy variations as multiple text options => Faster testing upfront, but you lose sight of what drove the result.

i've been leaning toward Option A because it's easier to scale later. Yesterday, someone in the thread pointed out that using multiple text options fragments social proof. When it's time to pull the winner into a scaling campaign, you can't grab a clean existing post ID - the engagement gets scattered across permutations Meta served in the background. That hit home.

But Option B is tempting when you're impatient to find that first winning angle.

Also - the ABO vs CBO debate came up. Some argue CBO lets Meta starve weak angles naturally, which reduces wasted spend by 20% or more. i'm still on ABO for control, but open to switching if it accelerates the test.

One thing I haven't decided: do you optimize for purchases from the start, or use earlier metrics as proxies until you have conversion data? A colleague mentioned that if you're early in the test, optimizing for ATC might not align with eventual profit. Worth thinking about.

What's the consensus? Are you running 1:1:1 ad structures for angle testing, or do you throw multiple combos at the walls? And for those who've migrated to CBO: did it actually speed up finding winners, or did you lose too much control?