By making your performance tests controlled experiments, you understand which variables change and which are constant. This lets you know what is effective and what is not.
This doesn't mean you have to make changes one at a time. You can still throw in a whole slew of changes; just make sure you can control them individually. Put each change under the control of a Boolean flag, either local to the object or global to the class. Then you can turn each flag on and off individually or in combination and measure the results.