How to Mix a Song (A Practical Order of Operations)

You've been on the same song for five hours. It sounded great two hours ago. Now the vocal is buried, the kick disappeared somewhere around the third plugin, and you can't tell whether you fixed anything or just got used to it. You bounce it, play it in the car, and it's a completely different song, and not in a good way.

That's not a gear problem. That's an order-of-operations problem. Almost every rough home mix goes wrong in the same place: plugins went on before the balance was right, so every move after that was a correction to a correction. Here's a working order that keeps that from happening.

Before You Touch a Fader: Fix the Session

Mixing can't rescue a session that's disorganized or badly recorded. Spend twenty minutes here and save five hours later.

• Name and color everything. "Audio 12" is a tax you pay every time you look for the snare. Kick, SnareTop, SnareBottom, OH_L, OH_R, BassDI, BassAmp, GtrRhythm_L, VoxLead, VoxDouble. Group by instrument family and color the groups.
• Gain stage before anything else. Get every track peaking somewhere in the -18 to -10 dBFS range using clip gain or a trim plugin, not the fader. Plugins, especially anything modeling analog gear, behave differently at different input levels. If your tracks slam in at -1 dBFS, every compressor you insert is going to overreact and you'll blame the compressor.
• Clean up the obvious. Cut silence between vocal phrases, fade edits so nothing clicks, comp the takes you're actually keeping, tune what needs tuning, and fix timing that distracts. Do the surgery now. Doing it mid-mix means re-balancing afterward.
• Check phase on multi-mic sources. Kick in and kick out, snare top and bottom, DI and amp. Flip polarity on one and listen for which version has more low end. This single check recovers more low-end weight than any EQ move you'll make later.

Build a Static Balance First

Bypass every plugin. Set all faders to a starting point of nothing, then bring elements up one at a time until the song reads. No EQ, no compression, no reverb. Just volume and pan.

Start with whatever the song is actually about. In most rock and pop that's drums and bass, then the lead vocal, then everything else around them. In a singer-songwriter track it might be voice and acoustic first. The test is simple: with nothing but faders, can you hear every part you need to hear, and does the chorus feel bigger than the verse?

If the answer is no, that's an arrangement problem, not a mix problem, and no plugin will fix it. Two instruments playing the same rhythm in the same octave will fight forever. Mute one, move one up an octave, or change the part. Fixing it at the source takes a minute; fixing it with EQ takes an hour and never fully works.

Pan with intent while you're here. Anything that carries the song's weight — kick, snare, bass, lead vocal — stays centered. Doubles, rhythm guitars, percussion, and pads go out to the sides. Width comes from what you move off center, not from a stereo widener.

EQ: Subtract Before You Add

The instinct is to boost what you want to hear. The better first move is to remove what's in the way, because a mix only has so much room and most of it is being eaten by content nobody can hear anyway.

• High-pass what doesn't need low end. Vocals, guitars, cymbals, most synths and keys. Sweep the filter up until the source thins out, then back it off. You're clearing space for the kick and bass, which are the only two things that usually belong down there.
• Find and cut the ugly resonances. Boost a narrow band hard, sweep it across the source, and listen for the frequency that makes you flinch. Then cut it, narrowly, by a few dB. Boxiness usually lives around 200–500 Hz. Harshness lives around 2–5 kHz.
• Carve for the vocal, not against it. If the vocal is fighting the guitars, a small cut in the guitars where the vocal lives usually beats a boost on the vocal. The one you cut is the one nobody misses.
• Boost broadly, cut narrowly. Wide gentle boosts add character. Narrow surgical cuts fix problems. Reversing that is how mixes end up brittle.

Compression: Control, Then Character

Compression has two jobs and they get confused constantly. The first is control — making a performance sit consistently so it doesn't duck under the mix on quiet words and jump out on loud ones. The second is character — the pump, the glue, the attitude.

Do control first, and do it modestly. On a lead vocal, a few dB of gain reduction on the loudest phrases with a medium attack and a release that breathes with the tempo gets you most of the way. If you need 10 dB of reduction to make the vocal sit, do it in two stages (a light compressor into another light compressor) rather than one compressor working overtime — it sounds far more natural.

Attack and release are the actual controls, and most people ignore them. A slower attack lets the transient through, so the snare keeps its crack. A faster attack tames the transient, so the bass sits flatter. Release should let the compressor recover before the next hit, roughly in time with the song.

Bus compression across the drums or the whole mix is character work, and it goes on last and light. One or two dB of movement is plenty. If you can clearly hear it working on the master, it's probably doing too much.

Effects: Depth, Not Decoration

Reverb and delay put things in a space. Used badly, they push everything to the back of the room and turn a mix into fog.

• Use sends, not inserts. One or two reverbs on aux buses that multiple tracks feed into puts everything in the same room. A different reverb on every channel puts every instrument in a different building.
• Short for glue, long for drama. A short plate or room ties a drum kit together without you noticing it. A long hall is a specific effect you use on a specific moment, not the default.
• High-pass and low-pass the reverb return. Reverb with full low end muddies everything; reverb with full top end hisses. Filtering the return is the difference between depth and wash.
• Time the delay to the track. Eighth or dotted eighth for movement, quarter for space. Feed it from a send so you can push the echoes up in the gaps and pull them back under the lines.
• Saturation earns loudness honestly. A little harmonic distortion on bass, drums, or a vocal makes them read louder without moving the fader, and it survives streaming compression better than a limiter does.

Automation Is What Separates a Mix From a Balance

A static mix is one compromise applied to a song that changes. The verse balance is never the chorus balance. Automation is how you stop compromising.

Ride the lead vocal line by line so every word sits — this is the single highest-impact thing most home mixers never do, and it does more for vocal clarity than any compressor. Pull the reverb send up on the last word of a phrase. Drop the guitars a dB under the vocal in the verses and let them back up in the chorus. Push the whole chorus up half a dB. Small moves, many of them.

Referencing and the Checks That Catch Problems

You cannot trust your ears after three hours in the same room, and you cannot trust the room itself unless you've treated it. Build checks into the process instead.

• Reference at matched loudness. Import a commercial track in your genre, turn it down until it's as loud as your mix, and A/B. Louder always sounds better, which is why unmatched comparisons teach you nothing.
• Mix quiet, and check loud briefly. Low volume exposes balance problems: if you can hear the vocal and the kick at a near-whisper, the balance is working. Turn it up occasionally to check the low end, then go back down.
• Check in mono. Collapse the mix and listen for anything that disappears. What vanishes in mono has phase problems, and mono is still how your song plays on a phone speaker, a Bluetooth speaker, and half the venues you'll ever play.
• Take real breaks. Ten minutes of silence resets your ears more than another plugin will. Come back and the problem is usually obvious in the first ten seconds.
• Sleep on it, then check three systems. Headphones, phone speaker, car. Note what's wrong on each, then fix in one focused pass rather than chasing each system in turn.
• Get outside ears. You've heard this song two hundred times and you no longer hear it. Someone who hasn't will catch the buried vocal in eight seconds. On Bandry, Showcase is built for exactly this: you post a track and get private, anonymous feedback from other musicians, and you give feedback to get it. Anonymous matters here, because people are honest when their name isn't attached.

Exporting, and Knowing When to Hand It Off

Export at the session's sample rate and bit depth, 24-bit minimum, with the mix bus peaking around -6 dBFS and no limiter on the master. That headroom is what a mastering engineer needs to work with. If you also want a loud version to send to friends or a venue, bounce that separately.

And be honest about the ceiling. Mixing your own songs is worth learning — it makes you a better writer and a better tracker, and it means demos don't cost you anything. But there's a point where the same problem keeps returning on every pass, and the fix isn't another tutorial: it's someone who has mixed a hundred songs shaped like yours. If the release matters, that's a good place to spend money. Our guide to finding a mixing engineer without getting ripped off covers rates, vetting, and what to send. When the mix is done, mastering is a separate job with a separate person.

Bandry is where a lot of that hiring happens. Mixing engineers, mastering engineers, and producers post their own listings and link straight to their work, and because those services work remotely, the posts are visible everywhere, not just in your city. When a listing fits, you tap 🔗 and your contact details go to them directly — no middleman on the booking, and no messaging layer to get lost in. If you'd rather post the need yourself, a "seeking" post describing your genre, track count, and deadline puts it in front of the engineers who want that work. See Bandry's find-a-mixing-engineer flow for how it fits together.