- After an experiment has ended, examine performance over a timeframe that includes a ramp-up period.
- Select two or three metrics to assess campaign performance and determine the winner of a test.
- Use a different approval process for new ads during experiments so that experiments can be conducted more expeditiously.
- Focus each test on one variable at a time and use different tests to assess the effects of more than one change.
The correct answer is: Focus each test on one variable at a time and use different tests to assess the effects of more than one change.
Explanation: Creating a campaign experiment necessitates focusing each test on one variable at a time, and using different tests for examining the impacts of multiple changes. This helps clearly delineate the influence of each variable. Tests should be designed to attain statistical significance swiftly, usually through a 50/50 split. Bear in mind the ad approval process that creatives undergo, and factor this into the experiment’s start date. To keep your experiment and original campaign aligned, use the experiment sync feature, which is typically enabled by default. This ensures any original campaign optimisations are automatically copied over to the experiment campaign. Always choose one metric for determining test results and evaluate these results, excluding the ramp-up period, which is typically about one week.