A/B a compressor with makeup gain matched to the bypass level to hear compression without loudness bias

Compression applies gain reduction but also changes loudness, and louder almost always sounds ‘better’ even when nothing else improved. Gain-matched A/B neutralises this bias: raise the makeup gain until the compressed signal is the same perceived volume as the bypassed signal, then toggle. Only then can you judge what the compression is actually doing — transient shaping, sustain, glue — without being fooled by a volume increase. As a rough guide, a compressor showing about 4 dB of gain reduction needs roughly that much makeup gain to match, though matching by ear is the goal.

Examples

In Ableton Glue Compressor on the drum bus: set Threshold so the meter shows ~-4 dB gain reduction, Attack ~10 ms, Release short. Raise Makeup until the on/off toggle sounds equally loud, then A/B and listen for ‘what changed’, not ‘which is louder’.

Assessment

Why does comparing a compressor bypassed vs engaged without gain-matched makeup give a misleading impression of whether it sounds good? How do you set the makeup gain for a fair comparison?