Facebook Ads Targeting Lesson 101: How to test, validate and optimize your audience
A lot of people think lookalikes are everything when it comes to targeting.
Right? Right.
Little do they know that interests can have similar (if not better) results too.
So here’s what you can do to test interests:
When TESTING interests, I usually separate them into their own ad sets because I want to isolate which ones would work or not.
But if you have hundreds of interests to test and you don't want to test each one of them individually…
Then group them by commonality!
(Side note: I usually just test by commonality if the audience size of an interest is below 1m.)
Once the audiences are qualified, I'll move them to my validation campaign.
I combine the highly related interests that require similar ads to have an audience size of at least 5m.
And that's how I test interests.
“But Sasa, what about lookalikes?”
When testing lookalikes, I keep them separate when testing as well.
1 Ad set for:
1% Web Visitors
1% View Content
1% Saved Posts
1% Video Viewers
1% Purchase
1%-2%
2%-3%
3%-4%
etc.
Just like testing interests, keep them all separated when testing so you can see which ones work or not.
And when moving them to your main campaign (or second validation campaign, if you have one), that’s when you should stack the 1% audiences together.
There are 2 reason for this...
If you have a group of high performing 1% audiences, then:
1. You have a bigger audience size, which ‘generally’ works better with Facebook.
2. You're squeezing the BEST of all audiences and giving more juicy data for the Facebook algorithm to work with.
(Just gonna put an emphasis on the word: “testing”)
Don't worry too much about the overlap when TESTING audiences.
Just have the right exclusions.
>> Here's an example:
Say there's a 70% overlap on 2 US 1% lookalike audiences.
There are 2.2million people in a 1% US lookalike audience.
70% of 2.2 million is 1.54 million.
Which means there's still 660,000 people that's not covered by the other lookalike audience.
You're just testing, you're not gonna reach 2.2million people right away.
PLUS, 660k is still a lot of people.
And a certain percentage of that audience size is gonna be your target audience as well.
So don't worry about overlap.
And that's it!