LUFS, Explained: How Loud Your Master Should Be for Streaming (and Why Louder Loses)
Somewhere near the end of every home recording project comes the question that launched a thousand forum wars: how loud should the master be? The modern answer involves three letters, LUFS, and a piece of good news that hasn’t fully reached bedroom producers yet: the loudness war is over, the streaming platforms won it for you, and squashing your song to compete is now actively counterproductive. Here’s what LUFS means, what the platforms do with it, and the numbers to actually aim for.
What LUFS measures
LUFS stands for Loudness Units relative to Full Scale, and it measures perceived loudness: how loud a track feels to human ears over time, not how tall the waveform is. That distinction is the whole subject. Peak meters (the dBFS meters in your DAW) tell you the height of the tallest single sample; LUFS integrates level over the whole song, weighted toward the frequencies our ears care about. A track of quiet verses and one loud snare hit can peak at 0 dBFS while measuring very quiet in LUFS; a brickwalled wall of sound peaks in the same place while measuring very loud. Your DAW or a free metering plugin will show you “integrated LUFS,” the whole-song number, and that’s the one this article means.
What the platforms do with it
Every major streaming service measures your track’s loudness and turns it down to a reference level so listeners don’t get blasted between songs. Spotify’s default reference sits around -14 LUFS, and the other major platforms cluster in the same neighborhood, give or take a couple of dB. The precise numbers shift with platform policy, but the mechanism is what matters, because of its brutal implication: if you master to -8 LUFS to sound huge, Spotify simply turns you down about 6 dB, and what remains is your crushed, pumping, dynamics-free master played at the same volume as everyone else’s. You paid the cost of the loudness war (squashed transients, fatigue, no punch) and the platform confiscated the winnings.
Meanwhile the track mastered around -14 with its dynamics intact plays at the same perceived volume but feels louder, because the drum hits still jump out of the mix. In a normalized world, dynamics are the new loudness.
The numbers to actually use
For streaming releases, aim your integrated loudness somewhere around -14 to -10 LUFS, erring louder only if the genre demands density (heavy rock and electronic tend to live -10 to -8 even post-war, accepting some turn-down as a stylistic choice; acoustic music happily sits -16 to -14). Set your true-peak ceiling at -1.0 dBTP, because lossy encoding (the compression streaming uses) can push peaks slightly higher than your master’s and clip playback; a -1 ceiling leaves room. And stop there. Do not chase a target to the decimal: the platforms measure whole-catalog loudness fine either way, and -13.2 versus -14.0 is inaudible politics. If your mix sounds punchy, translates on both your monitors and headphones, and meters in the neighborhood, it’s done.
One honest caveat: normalization can be turned off by listeners, and a few contexts (DJ tools, some stores, sync briefs) still want louder files. If a specific outlet gives you a spec, follow their spec. For general streaming release, the neighborhood above is correct.
What this changes about how you mix
The liberating part: loudness is no longer something you buy with a limiter in the last five minutes; it’s something you build in the mix. Arrangement space, controlled low end, and a vocal recorded clean in the first place contribute more perceived loudness than any mastering plugin. Use a limiter at the end to catch peaks and set your final level, sure. But if you find yourself pushing it until the mix flattens, back off and fix the mix instead, because the platform will undo your limiter’s damage-free gains and keep only the damage.
The loudness war’s peace dividend goes to home recordists: for the first time since the 90s, the amateur with restraint beats the amateur with a brickwall limiter. Take the win.