Facebook Split Testing: A Quick-Start Guide
Published: November 13, 2018
Author: Tamer Rizk
Facebook recently released a split testing feature that allows you to test different versions of your ad or ad sets within your existing campaigns without having to create a campaign specifically for split testing. The split test feature will divide your audience into non-overlapping groups, which helps ensure that the test is conducted fairly and that each ad set is given a fair and equal chance in the auction. It fills a need advertisers have lacked for years in the Facebook platform.
Getting Started
To begin, simply duplicate an existing ad set and pick the below option:
Once complete and launched, the ad sets running the split test will be indicated with the icon below:
Variables You Can Test
- Delivery Optimizations
- Target Audience
- Placements
- Creative
- Products Sets
Ideas to Split-Test
Optimization Events
- Ad set optimizing for conversions vs. link clicks
- Ad set optimizing for link clicks vs. landing page views
- Ad set optimizing for view content conversion event vs. add to cart conversion event
- Ad set optimizing for add to cart conversion event vs. value
- Ad set optimizing for add to cart conversion event 7 day click / 1 day view vs. add to cart conversion event 1 day click or view
Audiences
- Lookalike audience based off of purchase pixel
- Lookalike audience based off of Facebook pixel customer value
- Note: this is a new feature where Facebook creates a lookalike based off of your high-value customers. It uses conversion events like purchase to dynamically identify these customers to create stronger-performing lookalikes.
- Demographics: Men age 21-30 vs. men 31-40
- Top 25% of site visitors last 30 days vs. last 30 day site visitors
Viewing Your Results
Once there is enough data to report on your test, you’ll receive a notification in Ads Manager. Facebook will help determine which test won and will provide you with a confidence level in the results. You can further look into the results of the test by clicking on “View Results”:
Happy split testing!