Facebook media buying has always been data-driven. However, post-iOS14, we are seeing huge signal-loss, meaning that FB’s algorithm is weaker than it was before the update.

The platform just doesn’t seem to be able to make as robust and intelligent connections from the fewer data points than it could when there was more data flowing to leverage. To put it simply, there’s less precision.

Based on the information that FB have been producing and testing that we have conducted ourselves, we can see that broadening audiences is the way to navigate some of these tougher waters right now.

A lot of extremely niche brands are struggling with their FB marketing right now in terms of being able to validate and use interest audiences. This is just because they just can’t reach their desired audiences as efficiently as they previously could.

When and where you can, you should test broad audiences and interests with more people in them, ideally with a higher budget. The extra budget will help to get you in front of more users and ultimately, haul in more data- essential considering the data-driven nature of testing on FB.

For example, what we find to work well is if you can validate an interest in a test, try stacking that interest with another successful one in the same ad set. 

We would also recommend testing broad as an audience; no age or gender parameters and no interest/custom audiences- especially if you have no previous winning audiences.

When you validate larger audiences, this will also help you as you begin to scale. This way, you can increase the budget without saturating the audience and you can also use creative to reach new segments of the audiences.

Another tip would be to ensure that you’re leaving your tests on for long enough to be able to gather that data. With broader audiences and the signal loss, it can take a little longer than it used to, in order to get in front of the right people. 

Try to test for at least 7 days before making any significant changes, unless things are definitely not working- in which case, take things back to the drawing board. 

It’s possible that the campaign may have false-started; if you hear talk of FB playing around behind the scenes or you see odd-looking data and you want to try restarting a test, you can try this.

A longer timeframe for testing, proper communication and the correct expectations set with clients are going to be integral for navigating data-dumps and helping you make data-driven decisions.

Hope this helps!