Learning to properly A/B your mix is the skill that separates engineers who improve quickly from those who spin their wheels. A/B-ing simply means switching between two things and comparing them: your mix versus a reference, or a processed signal versus its bypass. The catch is that our ears are easily fooled, and the biggest culprit is loudness. Louder almost always sounds better for the first few seconds, regardless of whether it is actually better. This guide shows you how to compare fairly, what to listen for, and how to build A/B checks into your workflow so you catch problems before they ship.
There are two kinds of comparison you will do constantly. The first is A/B against a commercial reference track to judge your overall balance and tone. The second is bypass comparison, switching a single plugin on and off to hear what it actually did. Both only work if the levels match.
Rule one: level-match before you A/B your mix
If you do nothing else from this guide, do this. Whenever you A/B your mix against a reference or toggle a plugin’s bypass, the two options must be at the same perceived loudness, ideally within 0.1 to 0.5 dB. Otherwise you are just choosing the louder one every single time.
- Reference tracks are mastered and loud. A finished commercial song is 6 to 12 dB louder than your unmastered mix. Pull the reference down with a gain plugin or the channel fader until it matches your mix, not the other way around.
- Compensating makeup gain on plugins. A compressor that adds 3 dB of makeup gain sounds “better” purely because it is louder. Turn the output down to match bypass before you judge whether the compression itself improved anything.
- Use a loudness meter, not just your eyes. Match by LUFS (short-term or integrated over the same section) rather than peaks, because peaks do not track how loud something feels. A free LUFS meter from our audio tools hub makes this quick.
How to use reference tracks well
A reference track is a professionally mixed and mastered song in a similar genre that you trust. Import it directly into your session on its own track, drop the level to match your mix, and switch back and forth. A dedicated referencing plugin can auto-level-match and let you jump between specific song sections, which removes the manual gain-matching step, but a simple imported audio track works fine.
When you switch, do not listen to everything at once. Pick one question per pass:
- Low end: Is my bass and kick as full and controlled as the reference, or boomy and loose?
- Vocal placement: Is my vocal as forward and present, or is it buried or too far out front?
- Top end: Does mine sound as open and airy above 10 kHz, or dull, or harsh?
- Width: Is my stereo image as wide, or narrower and more mono?
Pick two or three references, not one. Chasing a single song leads you to copy its quirks. Two or three give you a range that reveals what is genuinely standard versus what is one mix’s taste.
Bypass, mono, and instant-switch checks
Beyond referencing, build these quick A/B habits into every session:
- Bypass every plugin at least once. After dialing in an EQ or compressor, toggle it off and on at matched level. If you cannot hear a clear improvement, you probably do not need it. This is the fastest way to stop over-processing.
- The mono check. Collapse your mix to mono and A/B it against stereo. Phase-cancellation problems, over-widened elements, and a vocal that vanishes in mono all jump out. Anything that must translate on a phone or a club sub needs to survive mono.
- The instant switch. When comparing two settings, switch quickly and immediately, not after listening for 20 seconds. Auditory memory is short; a fast switch lets you compare the two states directly instead of comparing the second one to a fading memory of the first.
- Blind it when you can. If a plugin has an A/B snapshot function, set two versions and switch without looking at which is which. Removing the visual bias of “the expensive one” or “the one I spent an hour on” is startlingly clarifying.
Be aware of confirmation bias in every one of these checks. If you expect a plugin to help, you will tend to hear it helping. The defense is to make the comparison as fair and as blind as possible: match the level, switch fast, and where you can, hide which option is which. When two settings genuinely sound the same after a fair comparison, choose the simpler one, because fewer plugins mean less phase smearing, less latency, and a cleaner signal path overall.
A/B across systems and over time
The final layer is comparing your mix on different playback systems. Bounce it and A/B on your monitors, headphones, a phone speaker, and a car. A mix that holds up everywhere is balanced; one that only works on your monitors is tuned to your room, not to music. Note where it falls apart, then go back and fix the underlying issue rather than EQ-ing for one speaker.
Also A/B across time. Print today’s mix, sleep on it, and compare it fresh tomorrow against yesterday’s version. Overnight distance strips away the fatigue and attachment that hide problems. For more on why quiet, matched listening reveals so much, see our guide to mixing fundamentals and our companion piece on EQ and compression basics. Master the discipline of fair comparison and every other mixing decision gets easier, because you will finally be hearing what your processing actually does rather than what you hope it does.
