Thread by @IstvanicMarin

My ABO testing is not necessarily about testing; it's also about scaling

It's the exact setup I'm using on EVERY ad account of mine, regardless of the Andromeda update

Here's my most technical breakdown of my ABO campaigns

Each new test = new ad set = new concept

Each concept has 3-6 variations.

When it's a video, it's mostly hook; when it's an image, it's either a different layout or text overlay

All of my tests start with a $100 or $200 daily budget, and I launch all of my tests on Wednesday. On Friday, I evaluated them for the first time.

I scale good ones, and in bad ones, and I either decrease the budget or kill the variation that got most of the spend and terrible results in bad ad sets .

I then let them run through the weekend, and on Monday, I do the same thing.

By that time, I probably
> killed 1/3 of the ad sets,
> scaled 1/3,
> and the remaining 1/3 is running with a lower budget & fewer variations than on Wednesday (after a week from testing)

After a week (next Wednesday), I take the winners with the post ID and carry them to ASC with CC, and I repeat this every week

BUT I DON'T KILL THE ADS I CARRIED TO ASC IN THE TESTING CAMPAIGN.

I continue to scale them there as well.

I know people said they had issues with running the same ads in different campaigns, but I didn't notice that.

ASC CC operates on a higher budget, and the CC target is around my target CPA.

Even if ASC spends over the target CPA, CPA is never "that bad" that I would be worried, because it's spent the ad on the winning ad

Even if the newly imported ad doesn't get much spend in my ASC "Hunger Games" campaign, I don't care, because I'm still scaling in testing

In terms of how soon I move the winner to ASC?

> I don't care about the existing learning phase.

> I'm not looking for statistical significance, nor do I care about it when identifying something as a winner

> If I managed to increase the budget a few times a week in testing, and performance is still good, that's a good enough sign for me to move it to ASC

P.S. ABO testing works if your tests are good enough. If you usually launch 10 tests a week, and you get only 1 winner, you'll do better sticking with CBO.