Start with one variable
A useful experiment changes only one thing at a time. If you adjust the setup, keep the context steady enough that the difference is still visible. Write down what you are changing, what you are leaving alone, and how long the trial will run. Choose something simple enough to undo. A new routine, an alternate layout, or a different order of tasks can be enough. Avoid combining several changes, because then you cannot tell which one mattered. The smaller the move, the clearer the lesson.