1 ad per ad set is better than multiple ads per ad set for testing.
Before I would test with 5 or more ads per ad set and let Meta decide the spend.
Spend would only go to 1-2 ads and I just assumed the other ads weren't winners as meta didn't spend on them.
But I've found so many winning ads recently from forcing spend on ads that originally got no spend.
I don't know why it took me this long to realise that you can't determine if an ad is a winner with only $0.41 spend.
The downside is you now spend more on testing, but it's worth it as you'll find more winning ads this way.
Now I test only with 1 ad per ad set.
I put the winning ads into a new campaign with multiple winning ads in the ad set.
I'll still leave the original testing ad set on, and scale this ad set too.
@frichdandy spending on ad level - I dont understand, this is what my tweet was about.
min is $5 a day for me, if roas still bad on this I'll turn it off (assuming it was profitable in the initial test)
max is unlimited. If it keeps doing well it keeps scaling.