How do I proof a video like a pro before release?
Proofing is not watching your edit again and hoping. It is a deliberate pass against targets, in the right order, designed to catch the flaws viewers punish. Here is how the pros run it, and how to run it yourself.
By Thomas, founder of CutScore · Updated June 2026
I have released videos I was proud of that turned out to be quietly broken. The audio was four decibels too low. A caption sat half under the progress bar on mobile. I did not catch any of it, because by export day I had seen the cut so many times that I was not watching it anymore. I was remembering it. That is the trap, and taste does not save you from it.
Print and broadcast figured this out decades ago. Nobody ships a magazine because the writer read it twice. A proofreader reads it cold, against a standard, hunting for mistakes rather than enjoying the prose. Proofing a video is the same idea. You are not admiring the thing. You are trying to break it before a stranger on a bad phone does it for you, in public, in the comments.
The difference between watching and proofing is the difference between a vibe and a target. A vibe adapts to your gear and your fatigue. A target does not. A pro proof has six of them, run in a deliberate order, and most of the work is boring. Here is the order, and what each pass is actually looking for.
The six-pass proof, in the order pros run it.
Each pass has a target you can hit and a reason it matters. Run them top to bottom. Sound first, because it is the flaw viewers forgive least, and export last, because it can undo all the rest.
| Pass | Target to hit | What slips through if you skip it |
|---|---|---|
| 1 · Loudness | ≈ −14 LUFS | A timid mix that feels weak the instant the next clip plays louder. |
| 1 · True peak | ≤ −1 dBTP | Crackle and distortion that only appear after the platform re-encodes. |
| 1 · Voice vs music | voice on top | Music swallowing the words, the loudest amateur tell there is. |
| 2 · Exposure + colour | neutral, not clipped | Dark, flat or green-tinted footage that reads as ungraded and unfinished. |
| 2 · Focus + stabilisation | sharp, no drift | Soft or shaky shots that look like a mistake, not a style. |
| 3 · Pacing | fits the genre | Dead air that lets the scroll win, or a frantic cut that tires people out. |
| 4 · First 3 seconds | one reason to stay | A slow open, where most of your drop-off quietly happens. |
| 5 · Captions + safe zones | readable, in-frame | Text too small, too low-contrast, or hidden behind the interface on mute. |
| 6 · Export + recheck | platform spec, re-watched | A crisp file that arrives soft and blocky after upload, unnoticed. |
Six passes on every video, every time, is real work. CutScore runs the whole proof in one pass and hands back the fixes, so the time goes into the next video instead.
What a pro is actually looking for.
Pass 1. Sound: the proof that catches the most
Pros proof sound first, because a viewer will tolerate a soft shot and bail on bad audio. Two numbers do most of the work. Loudness, which you want near −14 LUFS for YouTube so your video does not feel timid the second a louder clip follows it, and true peak, which you keep at or below −1 dBTP so nothing crackles after the platform squashes your file. Then the human test: play it on the worst speaker you own and see if you can hear every word over the music. If the music is winning, pull it down four or five decibels and do not get sentimental about it.
Pass 2. Picture: exposure, colour, focus, stability
Drop your screen to a normal brightness, not the heroic level you grade at, and look for two failures: shadows crushed to solid black with no detail, and highlights blown to pure white. Then ask whether your whites read as white rather than blue or orange, because a drifted white balance is the fastest way to look like nobody finished the shot. Check the subject is genuinely sharp, and that no shake or rolling-shutter wobble is pulling attention off what you are saying. This whole image family is half of what we analyze, because it is the first thing a viewer reads before a word is heard.
Pass 3. Pacing: proof it for the platform, not your patience
You have seen this edit so many times that it feels fast to you. It almost never is. The clearest single number for pace is average shot length, how long a shot holds before you cut away. A tutorial can breathe; a short cannot. When a section drags, it is usually one shot held three seconds too long, repeated twenty times, and a well-placed jump cut removes that dead air with no reshoot. The honest proofing question: would you keep watching this if it were not yours?
Pass 4. The first three seconds: proof the hook
Watch your first three seconds as if you were thumbing past it on a crowded feed. Is there one clear reason to stay, or do you open with a logo sting and a throat-clear? Most of your drop-off is decided right here, before anyone reaches the work you are proud of. If your strongest moment lives at 0:40, a piece of it belongs at 0:01. Proof the opening as ruthlessly as you proof the audio, because a great video nobody watches past second two is still a video nobody watched.
Pass 5. On-screen text: read it like a stranger on mute
Roughly half your audience watches with the sound off, so for them the captions are the video. Read your text on a phone, held at arm's length. If you have to squint, the font is too small or the contrast is too weak against the footage. Check nothing important drifts into the platform's interface, the progress bar, the username, the buttons, where it gets clipped or covered. And while you are reading, count the filler words in your delivery: a few "ums" are human, a dozen a minute quietly says you are not sure of yourself.
Pass 6. Export, then proof the upload, not the file
This is the pass everyone forgets and everyone regrets. Export at the platform's preferred resolution and a healthy bitrate, upload, and then watch the published version inside the actual app. Platforms re-compress everything, so a file that looked crisp on your drive can arrive soft and blocky. A proof is not finished when the file exports. It is finished when you have watched the thing your audience will actually see, on the device they will actually use.
Here is a real CutScore proofing report for an everyday vlog: every pass above, scored, with timestamps and the exact fixes to make before release.
Proof these three first.
A full proof is six passes, but most of the gap between "homemade" and "this person knows what they are doing" hides in these three. Fix them before anything else.
By eye, by meter, or in one pass.
By eye and ear
Free, and better than skipping the proof entirely. The catch is the one we opened with: your senses adapt and your gear flatters. It works best on someone else's video, or yours after a day away from it. Use the six passes above so you are proofing against targets, not against a mood.
With scopes and meters
Accurate and honest. A loudness meter, a waveform, a scope for exposure. The cost is time and knowledge: you have to know the targets, open three tools, and read them correctly for every single video. Great if you genuinely enjoy this part. Most people do not, and skip it under deadline.
With a coach in one pass
Hand the file or a link to CutScore. It runs all six passes and more, against the right standard for your genre, and returns a 0 to 100 score with timestamped evidence and the fixes. No scopes to read, no targets to memorise. See a sample report.
Frequently asked.
Proof it before you release it.
CutScore runs the whole proofing pass for you and tells you exactly what to fix, with the evidence to back it up. Join the waitlist for early access.
Join the waitlist