Because it does. Louder is perceived as fuller, brighter and closer, and the effect is strong enough to overturn a tone judgement on its own. Every comparison made at unmatched level is a comparison of loudness wearing the costume of a comparison of quality.
This is the single most reliable way to make a bad decision in a mix, and it is invisible while you are making it. A plug-in that adds a decibel on the way through will win a bypass test against a better-sounding plug-in that does not. An engineer who trusts their ears and does not match level is trusting a measurement instrument with a known, large, systematic error.
Gain-staging is the boring fix: set every stage to a sensible operating level so that nothing downstream is being flattered or starved, and so that a plug-in with a level-dependent character is seeing the level it was designed for. Done once, at the start, it removes a whole class of problem that is otherwise debugged one track at a time at 2am.
The comparison problem needs its own tool. A loudness-matched A/B measures both states and equalises them before you hear them, so what is left when you flip is the actual difference. It is unglamorous, it settles arguments, and it is the reason our plug-ins can make claims that are checkable rather than adjectival.
The practical check is embarrassingly simple and almost nobody does it: bypass a plug-in and measure whether the level changed. If it did, every A/B you have made through that plug-in was a loudness test. Do the same for a whole chain. A staging tool earns its place if it can tell you the number rather than asking you to trust that the two states are comparable.
Every RTM plug-in runs a full 14-day trial with every feature unlocked and no card. An RTM account is required to activate; the trial needs no card. One-time purchase, 2 machines, no subscription. Updates are free for as long as we sell the plug-in.