下列何者是進行實驗時的最佳做法？

Testing one variable at a time ensures that any observed changes in performance can be directly attributed to that specific variable. This allows for clear, unambiguous conclusions about the impact of the change. If multiple variables are altered simultaneously, it becomes impossible to determine which change, or combination of changes, caused the results. Proposing multiple hypotheses initially is fine for brainstorming, but each experiment should isolate one. Assigning two campaigns per experiment group or segmenting audiences into too many groups unnecessarily complicates the experiment design and can dilute statistical significance, making it harder to draw reliable conclusions.