Why Is My Master Quieter Than Commercial Tracks? What 349,404 Measured Records Show

Your master is quieter than commercial tracks for two reasons: commercial masters are denser, and they spend less of their energy between 60 and 500 Hz. Across 349,404 records measured through our analyzer, the median release projects to about -9.4 LUFS. The loudest tenth have a median crest factor of 8.9 dB and put 24.9% of their spectral energy below 500 Hz. The records sitting around -12.7 LUFS have a crest factor of 13.1 dB and put 32.4% below 500 Hz. When I hold crest factor fixed, the leanest quarter of records still measures about 2 dB louder than the heaviest quarter, in every band I checked. Density is the bigger lever, roughly 5 dB across the range we measured against about 2 dB for the balance, but the balance is the one nobody checks, and it decides how much loudness the limiter can buy you.

The version of this question I hear most is some version of "I've got the limiter at -1 and it's still not as loud as my reference." That's a reasonable complaint, and the usual answer is "limit harder." After 20 years, 19 Billboard Top 20 albums and records past 3.3 billion streams, I've watched that advice wreck a lot of good kicks. So I pulled the measurements to see what the loud records actually have in common.

First, one distinction. If your song only sounds quiet on Spotify, that's normalization, and I've written it up separately in why is my song quiet on Spotify. This post is about the gap you hear in your own session, loudness meter up and normalization off, sitting next to a commercial reference.

How Loud Commercial Records Actually Are

90.7% of everything we measured is louder than -14. If you mastered to -14 because a tutorial told you to, the gap you hear against your reference is real, and it's about four to five dB. The genre-by-genre breakdown is in the average LUFS by genre study.

What Changes as Records Get Louder

I sorted all 349,404 records into tenths by loudness and averaged their spectral balance and crest factor. Crest factor here is the peak level minus the RMS level, in dB. It's the simplest measure of how dense a record is.

The mids barely move. Energy leaves the bottom and lands on the top. And it isn't the sub: the band under 60 Hz stays between 3.5% and 4.8% the whole way, and the loudest tenth carries more sub than the quietest. What falls is the 60 to 250 Hz bass band (16.5% down to 12.8% from tenth two to ten) and the 250 to 500 Hz low-mids (11.5% down to 8.1%). That's the kick body, the bass guitar's fundamental and the pad and guitar buildup, the same region I wrote about in the tonal balance by genre study.

Same Density, Different Loudness

Crest factor and loudness are tied almost by definition, so the obvious objection is that bass-heavy records are just less limited. I checked that. I grouped records into 1 dB crest-factor bands, and inside each band I compared the quarter with the least energy below 500 Hz against the quarter with the most.

Same peak-to-RMS, same amount of squashing, and the lean records still read about 2 dB louder. The leanest quarter sits near 19% to 20% below 500 Hz and the heaviest between 40% and 47%, so this is a wide comparison, but the direction never flips.

The reason is how loudness is measured. The LUFS meter uses a weighting curve that adds about 4 dB to everything above roughly 2 kHz and slightly discounts the deep low end, and the curve exists because your ears work the same way. Low-frequency energy takes up the waveform's peaks without being heard as proportionally loud. So a record that spends its headroom on 60 to 500 Hz hits the ceiling earlier and gets less loudness back for it. You can't limit your way out of that. The limiter just starts grabbing the kick.

Your Lane's Lines

The balance moves a lot by genre, so here is the same picture per lane. "Loud-10% crest" is the crest factor that the loudest tenth of each lane gets down to. "Gap" is the leanest-vs-heaviest difference at matched crest factor.

Every lane points the same way. Hip-hop and R&B carry a lot of low end and still get loud, but they do it with lower crest factors, and their gaps are smaller (1.3 and 1.2 dB), and my read is that those lanes have spent years learning to keep a big low end controlled. Afrobeats is nearly flat at 0.6 dB. Ambient is the extreme at 3.2 dB, which makes sense for a lane where 40% of the energy lives below 500 Hz.

Three Checks Before You Touch the Limiter

1. Loudness-match first. Pull your reference down until its integrated LUFS matches yours, then listen. If yours still sounds smaller, it's balance, and more limiting won't fix it. Picking the right reference matters more than people think.
2. Read your crest factor. Compare it with your lane's median and loud-10% columns. If you're at 13 dB in a lane whose loud records sit near 8.7, density is part of your gap. Meters calculate this differently (some use true peak minus integrated LUFS), so compare your own master against your reference on the same meter rather than against my numbers to the decimal.
3. Read your energy below 500 Hz. If you're well above your lane's median, fix it in the mix: the kick and bass fighting over 60 to 150 Hz, and whatever is stacking up between 250 and 500. That work belongs before mastering, which is why it's the first item on my is-my-mix-ready-for-mastering checklist.

I'm not telling anyone to scoop their record. The leanest quarter isn't a target, and a thin record at -7 is still a thin record. What I've noticed in my own sessions is that the masters that finally got loud without falling apart usually got there in the mix, a few dB out of the low-mids and a cleaner kick and bass relationship, with the limiter doing less work at the end.

Method and Catches

Every number comes from audio measured by the Sonic Analyzer pipeline at Freshly Baked Studios: 349,404 tracks, excluding six rows with impossible loudness readings and any row still carrying the extractor's placeholder band values. Loudness is the analyzer's projected whole-track integrated estimate. The raw measurement is taken on a captured segment and reads about 3 dB quiet, so we correct it with a fitted model. A live-capture cross-check put the projection about 1 dB conservative. Crest factor is sample peak over RMS on that capture, in dB. Spectral shares are sub (under 60 Hz), bass (60 to 250), low-mid (250 to 500), mid (500 Hz to 2 kHz), and 2 kHz and up, as shares of spectral energy. Crest factor also enters the loudness model, but with a very small weight (0.05 per dB). The spectral bands don't enter it at all, which is why the matched-crest comparison is a real test and not circular. Genre lanes come from the analyzer's own classification, with regional and case variants merged and lanes under 800 tracks dropped. This is correlation across released records, not a controlled experiment. Lean records may also differ in arrangement, and I can't separate that here.

If you want these numbers on your own file instead of a guess, run it through the Sonic Analyzer. It's free and reports loudness, crest factor and the same bands against your genre. And if the answer is that you want someone else to do the mastering, here's what mastering costs and what mixing and mastering together cost, itemized honestly, alongside what a producer costs if the real fix is the arrangement. Or skip the ranges and get a number calibrated to where you actually are as an artist from the rate calculator.