How loud should my master be?

The honest answer: it depends where the track is going to play. A club system and a streaming app treat your master in completely different ways, and one number can't serve both. Here's how to pick a target on purpose instead of by superstition.

The short answer

Those streaming numbers are the platforms' published reference levels, and the club range is common practice for dance genres — melodic techno and house masters typically land around −8 LUFS, harder tech house and big-room styles closer to −7, and drum & bass often pushes toward −6. None of this is law. It's convention, and it exists for physical reasons worth understanding.

What LUFS actually measures

LUFS (Loudness Units relative to Full Scale) is the loudness measurement defined in the ITU-R BS.1770 standard. Unlike a peak meter, which only reports the tallest sample, a LUFS meter filters the signal to approximate how human hearing weights different frequencies, then averages energy over time. Three windows matter:

• Momentary — a 400 ms window. Useful for watching individual hits.
• Short-term — a 3-second window. Tells you how loud the drop feels right now.
• Integrated — the whole track, gated so silence and quiet intros don't drag the number down. This is the figure people mean when they say "my master is −8 LUFS."

One LU equals one decibel of level change, so a −8 LUFS master is roughly 6 dB louder, perceptually and electrically, than a −14 LUFS one. That's a big difference — and it has to come from somewhere, usually by limiting away transient headroom. Loud and dynamic pull in opposite directions; the whole job is choosing the trade.

Streaming normalization: the part people get wrong

Major streaming platforms apply playback normalization. Spotify and YouTube reference around −14 LUFS integrated; Apple's Sound Check sits near −16. If you deliver a −7 LUFS master, the platform simply turns it down about 7 dB on playback. The listener hears it at the same loudness as everyone else's track — except yours arrives with its transients shaved off by heavy limiting, and the quieter, punchier master next to it can actually sound bigger.

Two honest caveats before you swing the other way:

• Normalization can be off. Some players and modes don't normalize, and DJ software and download stores generally don't. A dramatically quiet master is a liability everywhere normalization isn't running.
• Turning quiet tracks up is not guaranteed. Platforms turn loud material down reliably; boosting quiet material is more conservative because it risks clipping. Don't master to −20 expecting a free ride up.

The practical takeaway: for streaming, chasing extreme loudness buys you nothing after normalization. Somewhere in the −10 to −8 LUFS region, most dance mixes still keep their punch while staying competitive in un-normalized contexts. What you should never do is push to −5 for Spotify — you're paying for loudness the platform immediately refunds.

Why club masters run hot anyway

A DJ booth has no normalization. Your track plays in a mix next to commercial releases mastered at −7 to −5 LUFS, and if yours sits at −12, the energy in the room visibly drops the moment it comes in — the DJ reaches for the trim, the gain structure of the night wobbles, and your track gets remembered as "the quiet one." Density is also part of the sound: the sustained, saturated character of a modern techno or DnB master is an aesthetic, not just a level.

So for club-destined masters, common practice really is −7 to −5 LUFS integrated, with short-term loudness in the drops hotter still. The craft is getting there without audible pumping or a flattened kick — which is mostly a mix question. If the low end isn't tight before the limiter, no amount of mastering rescues it; see mixing low end for club systems.

One master or two?

Reasonable engineers disagree. One camp cuts a single master — club-loud — and lets streaming turn it down; the level is fine after normalization, and the dense character is the sound of the genre anyway. The other camp renders two: a club master at −7 or hotter, and a streaming master around −10 to −9 with a little more transient life. Both are defensible. What matters is that the decision is deliberate and that both versions come from the same mix with the same tonal balance, so the record sounds like itself everywhere.

Loudness is not the ceiling

Whatever integrated target you choose, the peak ceiling is a separate control. Set the limiter's true-peak ceiling to −1.0 dBTP, not 0 dBFS — lossy encoding for streaming can otherwise push inter-sample peaks past full scale and cause audible clipping on cheap playback gear. That mechanism is its own topic: true peak explained.

How to measure it

Any BS.1770-compliant meter will do — most DAWs ship one. Measure the whole track for integrated loudness, not just the drop, and check short-term loudness in the loudest section separately. When comparing against a commercial release, gain-match first: the ear reliably prefers whichever version is louder, even by half a dB, so an unmatched comparison tells you nothing about quality.