To create an A/B test, go to Experiments in your All Tools menu in Ads Manager.
You can test campaign groups, campaigns, or ad sets. With campaign groups, you can group similar campaigns together to compare their performance to another group.
In this video, I tested three ad sets. They were set up identically (targeting, placements, and creative) except for the performance goal. Each ad set was optimized differently.
What’s nice about creating A/B tests in Experiments is that you select ad sets that are currently running. The original method of split testing required you created a split test with a set start and finish. The campaign then stopped.
In this case, you determine when the split test will run. The ad sets will run normally outside of the test. There will be no overlap in targeting while the test is running. You can choose to end the test early if a winner is found.
Choose a key metric to determine your winner and up to nine additional metrics that you’ll want to see in the results.