Gain-match before deciding
A small level difference can distract from the change you intended to hear.
Set a repeatable comparison
Loop a representative passage and keep the monitor level steady. Include both a busy phrase and a quieter moment so a setting is not chosen from a single transient. Save the original state before making a change.
Listen to the change in context first. Solo can help identify an artifact, but the arrangement determines whether a change is useful.
Match perceived level
Adjust the processed output until switching between the two states does not create an obvious loudness jump. A loudness or RMS meter can help, but compare the actual passage as well: peak matching alone does not match perceived loudness.
Keep the comparison short enough to repeat accurately. If the brighter or denser version still feels louder, revisit the output trim before adding more processing.
Choose the useful difference
Check the change at a quieter playback level and through the complete section. Keep it when it serves a clear purpose, such as making a lyric easier to follow or leaving room for another instrument.
If neither version wins consistently, keep the simpler state and move on. Save an alternative rather than stacking more processing to justify the first choice.