Final answer:
The best practice for creating a campaign experiment is to isolate a single variable and choose specific metrics to measure performance. All new ads must go through the usual approval process during the experiment, and performance assessment should exclude the ramp-up period to better gauge impact.
Step-by-step explanation:
When creating a campaign experiment, a best practice is to isolate one variable to focus on for each test. This approach ensures that you can clearly attribute any changes in performance to that specific variable. By controlling the experiment to change only one variable at a time, you eliminate confusion over what may have caused the outcome.
Additionally, while conducting the experiment, two or three key performance metrics should be chosen that can effectively measure the campaign's success. This allows for a clear understanding of which variation is more effective. It is a common misunderstanding that new ads don't need approval during experiments, but in fact, all ads must adhere to the standard approval process regardless of the phase. Finally, when assessing the performance after the experiment, it's essential to consider a timeframe that does not include the ramp-up period to get a more accurate measure of the impact.
Experimental design should aim to test hypotheses about cause and effect with minimal influence from any lurking variables, providing measurable results that support a determination of whether the hypothesis is validated.