Match the level before you decide
A fair before-and-after on any track, so you keep the changes that help and drop the ones that were only louder.
The problem
You add an EQ boost or a saturation stage, switch it on and off, and it sounds better. Then you come back the next day and it doesn't. Most of the time the processed version was simply a little louder, and louder almost always wins at first.
What to do
Loop eight bars that include a busy phrase and a quieter one, so the decision isn't made on a single hit. Keep the monitor level where it is for the whole comparison.
Make the change, then trim the plug-in's output until switching it on and off makes no obvious jump in loudness. A loudness meter gets you close. Finish by ear, because matching the peaks doesn't match how loud something feels. A brighter or denser version still feels louder at the same peak level.
Keep the switching quick. Flip between the two states every few seconds and listen to one thing at a time: the words, the kick, the space around the snare. I close my eyes for this, so the plug-in window can't vote.
On an EQ, the free Band Practice makes this easy. Set its Bypass Mode to Matched and the bypass is loudness-matched, so switching it off doesn't flatter either side. Auto Gain holds the loudness steady while you shape the curve, which keeps every move honest from the start.
When to stop
If neither version wins after a few honest switches, keep the simpler one and move on. A change you have to strain to hear isn't worth the time it takes to defend. Save the other version if you can't let it go, but don't stack more processing to justify the first choice.
The test
Come back after a break and play both versions at a quieter level, with the whole section running. The one you keep should make a lyric easier to follow, or leave more room for another part, at the same loudness. If the only difference was level, you've just saved yourself a plug-in. Band Practice is free for good.
Rory