How to find the “best” Lookalike Audience
Published: January 26, 2016
Author: Kendra Pennington
Using Lookalike Audiences is (almost) a surefire way to reach your ad goals, but only if you know how to find the best Lookalike audience. Here are our best practices for getting the right Lookalike audience.
Recommended target seed audience size
Social marketers at 3Q Digital used to recommend that new folks start with audiences at ~75K matched users, but we’ve found that hyper-targeted audiences from 1-5K matched users work extremely well. You can find your audience size per custom audience from the audience section in your ads manager.
If your current lists are larger than the recommend size, we suggest scaling list down to result in a stronger seed audience.
How many points of commonality Facebook looks at (it’s larger than you think)
Facebook looks at over 10K characteristics to develop the Lookalike Audience. Using an audience with a larger sample size dilutes the results that Facebook can produce. Hence: our new recommended hyper-targeted seed audiences.
Methods for testing LALs against each other
Want to make sure you have the best audience? Test against another! Start with a 1% audience with segments 1, 2, and 3 and leave all other attributes the same (bid amount, bid type, creative, etc.). Increase lookalike %’s when looking to scale your top-performing group.
Use the a/a/a/a/a test within each lookalike: test the same ad, in the same ad sets, multiple times. Facebook’s algorithm decides which ad is ‘top performing’ fairly quickly; this is attributed to the ad ID rather than the image ad copy combination.
Imagine you only have one version of an ad within a lookalike audience. Facebook shows it to 100 random people in the audience, and they don’t react favorably. Facebook’s algorithm decides to show that ad less often based on that small sample size. Conversely, if you have a random 100 people who react favorably, that ad will be shown more often. Using multiple iterations with the same creative allows you to get a range of “first impressions” that will more accurately represent the effectiveness of the audience as a whole.
Make sure you’re updating exclusion lists constantly; for instance, exclude current customers by asking for updated exclusion lists periodically from your client. Note that lookalike audiences are refreshed by Facebook every 3-5 days, but Facebook itself has no way of refreshing your exclusion list. Make sure to manually refresh to optimize performance.
Got any other social topics you’d like to see covered? Let us know in the comments!