Few questions come up as often in mastering, and few have such an unsatisfying answer: there is no number you have to hit. Since every major platform levels playback, louder no longer wins. What counts is how a song sounds after the platform has brought it to its own reference level.
What normalisation actually does
Every streaming service plays all tracks back at roughly the same volume. To do that it measures how loud a song is on average and turns it up or down on playback. The measurement is in LUFS, a loudness scale that corresponds roughly to what people perceive as loud – unlike raw peak level.
The direction is what matters: louder tracks get turned down. Quieter ones are only lifted as far as their peak levels allow. So squeezing a master as loud as it will go achieves exactly one thing – the platform turns it back down.
What the platforms aim for
The figures sit close together and rarely change. Spotify normalises to about −14 LUFS, although listeners can choose −19 or −11 in the settings. Apple Music works at around −16 LUFS. YouTube, Tidal and Amazon Music sit at roughly −14 LUFS, Deezer at about −15.
Two things follow from that. First, there is no shared number a master could hit for all of them at once. Second, the spread between services is smaller than the difference a well-made master makes anyway. Optimising a master towards one of these figures therefore achieves less than most people assume.
Why louder gains you nothing
Take two versions of the same song. One is squashed to −7 LUFS, the other sits at −11 LUFS and has kept its dynamics. Spotify turns the first down by seven decibels, the second by four. In the end both play back equally loud – but the squashed version has no peaks left, no push in the chorus and no air in the quiet passages. It sounds smaller, despite being "mastered louder". That is precisely why the loudness war is over.
True peak: the real risk
What can genuinely go wrong is peak level. Streaming is delivered in lossy formats, and that conversion creates peaks that were not in the original. A master pushed right up to 0 dB distorts audibly afterwards – everything is clean on your own machine, and the stream crackles.
So we leave headroom, usually down to about −1 dBTP. It costs nothing in perceived loudness, because the platform normalises anyway, and it removes the one risk that cannot be repaired after the fact.
What we do instead
We master to the song, not to a number. A singer-songwriter track tolerates and needs more dynamic range than a metal track, and a live recording depends on the loud section actually being louder than the quiet one. Whatever LUFS value comes out of that, we note it, but it is a measurement and not a target.
What we always do is check how the song sounds after the platform has normalised it, and compare it directly against references from the same genre. If it holds up there, the master is finished. Everything else is cosmetics with numbers.
In short
- Every major service levels playback – mastering louder no longer gives you an advantage
- Spotify, YouTube, Tidal and Amazon sit at around −14 LUFS, Apple Music at roughly −16, Deezer at about −15
- Louder tracks are turned down, quieter ones only lifted within their peak headroom
- Leave roughly −1 dBTP of headroom, otherwise conversion to the streaming format distorts
- Master to the song and check against references instead of chasing a LUFS number
Frequently Asked Questions
If you want to know how your current mix holds up after normalisation, send it over – we will listen against references and tell you honestly whether a new master is worth it or whether everything is fine as it is. More on our page about mixing and mastering at punchline studio.