Last Updated on 4 weeks by School4Seo Team
The best practice when creating a campaign experiment is to isolate one variable to focus on for each test and use different tests when assessing the impact of more than one change.
- Choose two or three metrics that can be used to reliably gauge campaign performance and determine the winner of a particular test.
- After the end of an experiment, assess performance over a period of time that includes ramp-up.
- As a means of expediting experiments, new ads aren’t subject to the same approval process during time of the experiments.
- Isolate one variable to focus on for each test and use different tests when assessing the impact of more than one change.
The correct answer is: Isolate one variable to focus on for each test and use different tests when assessing the impact of more than one change.
Explanation: When creating a campaign experiment, it’s essential to focus on one variable per test. If you want to assess the impact of multiple changes, use different tests for each. This way, you can clearly determine the effect of each individual variable. Also, design your tests to reach statistical significance as quickly as possible, typically achieved through a 50/50 split. Be mindful of the ad approval process, which can take about a day, and always allow for this when scheduling your experiment. Utilize the experiment sync feature to automatically propagate any optimizations made to your original campaign to your experiment campaign. Finally, select one metric to be the determinant of test outcomes and ensure to exclude the ramp-up period while evaluating results.