An Introduction to Split Testing on Facebook
Published: April 25, 2017
Author: Zenia Johnson
Via John Towner
One of the most exciting additions to the Facebook UI in 2016 was the ability to conduct clean A/B/C tests using Split Testing. Prior to this update, designing tests was somewhat complicated in setup: multiple ad sets had to be created, you couldn’t ensure an even ad rotation, and there was no definite way to prevent audience overlap. Now, using the Split Testing tool, digital marketers can identify and test the variables that will have the most impact on campaign performance.
A quick guide to setting up split tests to take your campaign performance to the next level:
Currently Facebook Split Testing is only available for four Business Objectives: Traffic, App Installs, Lead Generation, and Conversions. Within those objectives, you have the option to test three variables (although only one variable can be tested within one single campaign).
Below are the variables that you have the option to test:
- Placements: This allows you to test the efficacy of ad placements. Facebook recommends testing custom placements against automatic placements, but not testing custom placements against one another.
- Audiences: This allows you to test up to three audiences against one another. One thing to note is that you must save your audiences prior to setting up this test. This means that your test would be Saved Audience 1 versus Saved Audience 2 versus Saved Audience 3.
- Optimizations: This is my favorite variable to test! This allows you to test the myriad of delivery levers that impact performance. Some of the things that you can test are ad delivery optimizations (impressions, link clicks, conversions, daily unique reach), conversion window, and automatic vs manual bids. As an example, I’m currently designing a test for a client that will give insight into how Average bid affects performance compared to Maximum bid.
Once you have your testing variable, it’s time to set your budget. Facebook provides a suggested budget, but it’s only a suggestion so just be sure that the budget you set is large enough to give you results that clearly identify a winner. Finally, set your campaign schedule: Facebook recommends a minimum of 3 days and a maximum of 14 days to allow your test to run. And now you’re all ready to run your first split test!
Have you used Facebook’s Split Testing option before? Tell us about your experience below in the comments!