Hi all,
When testing angles, do you split test them at the adset level or the ad level? The reason I am asking is because based on Zeno's guide, he would create separate adsets for each demographic targeted. So if I'm split testing angles at the adset level, that'll mean I'll have a shit ton of adsets?
i.e. Angle 1-US-Male-25-30-Desktop, Angle 2-US-Male-25-30-Desktop, etc
Yes, ad sets.
You generally want these to have independent budgets from each other for better parallel testing.
Having lots of ad sets is not something to worry about, not when you have something like the power editor and efficient filtering.
If your budget is very restrictive then yeah, you could test in one ad set, just be mindful of the evenness of delivery.
I agree, I think its a bit complicated in Zeno's guide (but if you read it carefully you get it there). If you create them all in one adset, most of the money will go to one ad quite fast, and you dont get a real split test.
Although FB Account Manager recommended me a different strategy, but Zeno's Guide made more sense for us
I have just finished the FB curriculum for the AMC, so STM Forum's will see an update shortly (finally...).
There have been a lot of updates to the way Facebook does things since the guide was written.
Typically we launch 10 angles, each as its own adset. Each ad set has just one ad in it where the creative is the same, only the text (aka angle) is changed.
Then we run it for 2-3 days. Running it for less is not good, as it takes time for FB algo to kick in.
Offer payout x3-4 per adset. So if you are running a $10 payout, you want each adset to be $30-40 - this is needed else FB algo can't do its job to test, then optimize.
Hope that helps.
I usually just throw in the different angles in as different ads (with everything in its own adset).
There are a number of ways to go about this, one important factor is whether you are using the FB pixel and what event data you can send - both the event nature and the number of events per day per ad set.
Thanks for the reply guys. Everything is starting to make more sense now. 