What does it mean?
Use the same DI and segment, with clear boundaries and routing; randomize comparisons when useful. Do not infer pedal differences from different performances or song sections.
ExampleRender A and B from the same take, then match perceived loudness before judging timbre.