The master comes back on a Tuesday. Thursday morning your client emails: the low end feels different than the mix, did something happen? You did not touch it. Someone else did, at the client’s request, using a chain you never saw. But you are the one answering the email, because you are the one they have a relationship with. Preparing a mix for mastering is not really a technical problem. It is a handoff problem, and the technical part is only the half you can control.
Most advice on this topic stops at a headroom number. That number is the least important thing in the exchange, and it is usually quoted wrong anyway.
Get the handoff right and mastering makes your work better. Get it wrong and it produces a week of email you did not budget for, plus a client quietly wondering whether the mix was the problem.
The Headroom Number Everyone Quotes Is the Wrong Number
Ask a forum how much headroom to leave and you will get minus six from a dozen people within an hour. It is not bad advice, but it is arbitrary, and it survives mostly out of habit. The convention traces back to the days when mixes were captured to DAT machines and two-track recorders, where clipping the input was permanent and ugly.
In a modern digital chain, a mastering engineer can turn your file down. If a mix peaks at minus one and it is genuinely clean, with no clipping and healthy dynamics, it can be mastered without any drama. The reason the minus six habit persists is that most mixes peaking near zero are not clean. They are already limited, already squashed, or bounced from a session where the mix bus was overloaded.
Put differently, the number is a proxy for a habit. Engineers who leave a few decibels of room are usually engineers who were not pushing the mix bus in the first place, and that underlying restraint is what actually makes a mix for mastering easy to work on.
So the rule is worth keeping as a workflow default, not as a commandment. Peaks somewhere in the minus three to minus six range, nothing touching the ceiling, and you have removed an entire category of problem.
Dynamic Range Is the Thing You Cannot Give Back
Here is the part almost nobody says out loud. Peak headroom is easy to fix. Dynamic range is not.
If you have been mixing into a brickwall limiter that is pulling eight decibels of gain reduction, that is eight decibels of transient information the mastering engineer will never see. Turning the file down afterward does absolutely nothing to help, because the damage happened before the fader. You have handed over a quieter version of an already-flattened mix and called it headroom.
The same logic applies to heavy clipping on the mix bus, which has become fashionable in a lot of modern production. A little is a sound. A lot is a decision that cannot be walked back at the next stage.
This is the single most common reason masters come back sounding lifeless. The mastering engineer had nothing left to work with. Every move they make is fighting a decision you already baked in.
What to Bypass Before You Print a Mix for Mastering
The distinction that matters is between processing that shapes the sound and processing that exists purely to make things louder.
Gentle bus compression doing one or two decibels of glue is part of your mix. Leave it. Mix bus saturation or EQ that defines the record is part of your mix. Leave it. A maximizer, a loudness limiter, or a clipper you added at the end to make the bounce competitive on the client’s phone is not part of your mix. Bypass it.
Sometimes the limiter genuinely is the sound. Certain aggressive electronic and hip hop productions fall apart without it. In that case, print two versions, label them clearly, and say so in your notes. Let the mastering engineer decide which one gives them more room. Sending both takes four minutes and prevents a full round of confusion.
A useful test: bypass the chain and ask whether the mix still sounds like the record. If it does, the processing was loudness. If it collapses, the processing was arrangement, and it stays.
If you are printing with any processing pushing hot, a 32-bit float export preserves information above zero that a 24-bit file would clip and discard for good.
File Settings That Actually Matter
Keep this boring and consistent:
- Print at the session’s native sample rate. Do not convert on the way out, and let the mastering engineer handle any conversion with better tools.
- 24-bit minimum, or 32-bit float if the chain is running hot.
- Keep true peaks below roughly minus one dBTP, and never let them touch zero.
- Skip dither on your end. Final dither belongs at the last stage, and applying it twice is worse than not at all.
- Leave a little silence at the head and tail rather than trimming tight to the waveform.
- Name files so they sort correctly and identify themselves: artist, song, version, date.
None of this is clever. It is just the difference between a file that gets worked on and a file that generates three clarifying emails first. The same instinct applies to everything you hand over, which is why a standing file delivery checklist is worth more than remembering.
Send Notes With Every Mix for Mastering
A file with no context invites guesswork, and guesswork is what produces a master that sounds nothing like what the artist imagined. A short note prevents most of it.
Tell them what the record is going for in plain language. Name the reference the artist keeps citing, and be specific about what they actually respond to in it, whether that is the weight of the low end or how forward the vocal sits. Flag anything you already know is imperfect, like a noisy guitar take you cleaned up as far as it would go, so nobody spends an hour rediscovering it.
Include the practical details too: release date, where it is going, whether vinyl is planned, and who has final approval. If the artist, the manager, and the label all have opinions, the mastering engineer needs to know whose note is the deciding one.
The Alternate Versions Nobody Asks For Until It Is Too Late
Print these at the same time, from the same session, at the same fader positions. Coming back three weeks later to match a vocal-up is a genuine headache and it is entirely avoidable.
The standard set is the instrumental, a TV mix with vocals down rather than out, an acapella, and a clean version if there is language to handle. Some projects also want stems for sync or live playback. Ask once, at the front of the job, rather than assuming. Handing all of it over in a single organized batch is also what separates a professional final mix delivery from a trickle of follow-up emails.
When They Want Stems Instead of a Mix for Mastering
Stem mastering comes up more often now, particularly on projects where the artist wants a second opinion baked into the process. Instead of one stereo file, the mastering engineer receives grouped buses: drums, bass, music, vocals, effects.
Be careful about agreeing to it casually. Stem mastering hands part of your balance decisions to someone else, and if the vocal comes up two decibels at the master stage, that is your mix that changed. It can work well with an engineer you trust and it can quietly undo weeks of work with one you do not.
If you do send stems, print them from the same session at unity, with all bus processing intact, and make sure they sum back to your approved stereo mix exactly. Say so in your notes, and state plainly which decisions are open for adjustment and which are not. The stem handoff rules you use with clients apply here too.
When the Master Comes Back Sounding Different
It is supposed to sound different. That is the job. The problem is that clients almost always compare the two at wildly different volumes, and louder wins every time regardless of quality.
Before you agree that something went wrong, level-match and listen again. Pull the master down to sit at the same perceived loudness as your mix and compare. Nine times out of ten the change the client heard was three decibels of gain, and the actual tonal difference is small and usually an improvement.
Ask the client to listen on the system they know best rather than whatever is nearest. A comparison made on laptop speakers at two different volumes is not information, and acting on it is how a good master gets revised into a worse one.
It also helps to remind everyone that streaming platforms normalize playback anyway, which means the loudness arms race matters far less than it did a decade ago. If your client is fixated on volume, the fuller version of that conversation is in the post on why clients keep asking for louder mixes.
Who Takes the Blame When the Client Hears a Change
Structurally, you are exposed. The client hired you, and if they hired the mastering engineer separately they still treat you as the person responsible for how the record sounds. When something shifts, the email comes to you.
Two habits reduce that risk. First, ask to be copied when the master is delivered, so you hear it before the client forms an opinion and writes to you about it. Second, offer to listen together. A ten-minute call where you play the level-matched comparison resolves nearly every one of these, and it positions you as the person managing the process rather than defending yourself.
It also helps to set the expectation before anything is sent. One line in your delivery email, saying that the mastered version will sound louder and slightly different by design and that you are happy to walk through it together, defuses the entire conversation in advance.
If you recommended the mastering engineer, you own the handoff completely. Choose people whose work you know and whose communication you trust, because their turnaround and their email manner become yours in the client’s memory. This is also a strong argument for being deliberate about whether you bundle mixing and mastering together in the first place.
Make the Mix for Mastering Handoff Part of Your Process
Write the checklist once. Bypass the limiter, print at native rate, check true peak, print alternates, write the notes, send. It takes twenty minutes at the end of a project and it removes the two most common failure points: a squashed file and a mastering engineer working blind.
Keep it as a saved note, not something you reconstruct from memory at eleven at night on a deadline. A mix for mastering that goes out wrong at midnight costs a full day to fix.
Every clean handoff is the same underlying skill, and the one you control most is at the front of the job rather than the back. That is what session.trackbloom.com is for. It is an upload link you send your client, and their tracks arrive grouped by instrument: vocals with vocals, drums with drums, keys with keys. Think WeTransfer built for audio. Files that show up organized are files you can actually work from, and a project that starts clean is a project that ends clean.
Ask Before You Guess
The fastest improvement available here costs one email. Before you print anything for a mastering engineer you have not worked with, ask what they want. Sample rate, bit depth, headroom preference, whether they want the limited version, how they want alternates labeled.
Most will answer in a paragraph and be glad you asked, because it saves them the same round trip. Save that answer, build it into your template, and the next twenty projects go out right the first time. For a deeper read on why the six decibel rule persists despite being mostly folklore, The Pro Audio Files breaks it down well.