Module 4: The Edit
Finding the moment was Module 3. This module is what happens after — the build, the captions, the sound, and the short list of mistakes that get clips denied. The edit is where a good moment becomes an approved clip, and where a sloppy one quietly earns nothing.
The standard clip build
Every clip starts as horizontal footage and has to live vertical. Open a 9:16 project in CapCut or your editor of choice and reframe before you cut anything else. Scale the source until the speaker fills the frame in a way that keeps their face and gestures in shot. When two people trade lines — a podcast pair, a streamer and a guest — don't shrink both into one wide strip at the top of the screen. Cut between two framings, one per speaker, or stack them: speaker on top, reactor below. The viewer should always know whose moment this is.
Then set tight in and out points. Your in-point is the first word of the moment — not the breath before it, not the end of the previous sentence. Your out-point is the payoff line, plus a beat. Everything before the moment is a reason to scroll, and everything after it is a reason the clip feels loose. If the moment needs one line of setup to make sense, use the shortest one that works, and check the campaign brief's source rules before you reorder anything — most briefs allow a setup line pulled from the same segment, some don't.
Length follows the moment type. Reactions and quotable lines run 10–25 seconds — they're a single beat and they die if you pad them. Reveals and payoffs run 20–45 seconds, because the build is part of the value. A conflict or a short story can hold up to about 60 seconds, but only when every second earns the next one. When you're torn between two lengths, post the shorter one. A clip that ends early gets rewatched; a clip that runs long gets abandoned at the same second either way.
Export clean, every time: 1080×1920, straight from the editor, with no editor watermark and no end card stamped on the file. If your editor adds a closing logo clip by default, trim it or switch it off in the export settings — that stamp marks the file as a draft. Save the project file, too. When a campaign manager asks for a re-export or a small fix, the project is the difference between two minutes of work and starting over.
You've found a strong 18-second moment in a podcast: one guest's answer, start to finish. What's the right build?
Captions and on-screen text
Most people meet your clip with the sound off, which makes the captions the clip. Word-by-word captions — the kind that light up as each word is spoken — hold attention better than a static block of text, and every standard editor can generate a first pass. Treat that pass as a rough draft. Go through it word by word and fix what the machine got wrong, because auto-captions mangle exactly the words that matter most: names, numbers, slang, and the payoff line itself. One wrong name in a caption is the error a reviewer remembers.
Emphasis is a ration, so spend it carefully. Bold or color the one or two words that carry the clip — the dollar figure, the name, the twist word at the end of the sentence. If every third word is highlighted, nothing is. The viewer's eye should land where the moment lands, and that only works when the rest of the caption stays quiet.
Before you export, hold your phone at arm's length and read the captions off it. That's the real test. If you have to squint, the size goes up or the line gets shorter — break long lines in two rather than shrinking the type. Give the text a dark outline or a backing shade so it survives bright footage, and position it clear of the platform furniture: the right-side buttons, the bottom caption and progress zones. Captions buried under the app's own text might as well not exist.
Beyond captions, the only on-screen text that belongs is a short context line when the moment needs one — the score, the stakes, who's speaking — placed at the top and gone within a few seconds. What never belongs is a promise the clip doesn't keep. Text that hypes a payoff the footage doesn't deliver reads as a trick, viewers leave early, and reviewers deny it. The caption's job is to make the real moment legible, not to sell a bigger one.
You auto-generated captions for a clip. The guest's name is spelled wrong twice and the dollar figure in the punchline is off by a zero. A friend says nobody reads captions that closely. What's the professional call?
Sound first
Viewers forgive grainy footage. They leave muddy audio in about two seconds, usually without knowing why. So when edit decisions compete, sound wins: a slightly awkward cut with clean dialogue beats a smooth cut where the key line is buried. Everything in this lesson follows from that one priority.
Start by leveling. Stream and podcast audio swings hard — a quiet setup line, then the room erupts. Bring the quiet parts up until every word is easy to hear, and pull the peaks down so the loud moment doesn't distort. Your editor's normalize or loudness tools get you most of the way; your ears finish the job. Aim for a clip that plays at one comfortable volume from first second to last, and keep that level consistent across all your clips so your account sounds like one place.
Then cut the dead air. Long pauses, room tone with nothing happening, the breath-and-shuffle before a sentence starts — each one is a small invitation to scroll. Tighten gaps down to a natural beat. But don't strip every breath out: a half-second of quiet right before a payoff line is timing, and timing is often what makes the line land. Cut the air that does nothing; keep the air that loads the moment.
Music goes under the moment or stays out. If you use a track, duck it well below the dialogue — a viewer straining to hear words over music is a viewer leaving. When the moment's own sound is the point — crowd noise, a room full of people reacting, a game sound cue — skip the music entirely. And where the campaign brief restricts which music you can use, the brief wins, full stop. Final check, always: one full listen on your phone's speaker, not your headphones. That's the speaker most of your audience owns.
Your clip's setup is a near-whisper and the payoff is the whole room shouting. In the export, the whisper is inaudible and the shout distorts. What's the fix?
The rejection reel
Clip denials are rarely mysterious. They cluster around a short list of mistakes, and once you've seen the list you'll recognize almost every rejection you ever get. Learn it now, because each of these is preventable in the sixty seconds before you post.
Wrong shape. A horizontal upload, or a vertical frame with the footage letterboxed into a strip between black bars, tells the reviewer you didn't build the clip for the platform. Some campaigns reject it on sight. The buried moment. The clip opens on eight seconds of wind-up — greetings, setup, dead air — with the moment nowhere in sight. Reviewers stop watching, and so does everyone else. If the moment is at second twelve, the clip starts at second eleven, at the latest. Unapproved overlays. Your logo in the corner, your handle watermarked across the middle, a text banner the brief never mentioned. Anything on screen that the brief didn't ask for or allow is a denial waiting for a reviewer to notice it.
Someone else's watermark. The source clip you worked from was itself a repost, and the original clipper's handle or a platform watermark is still visible in a corner. That reads as recycled work — because to the reviewer, it is. Work from the cleanest source the brief points to, and check the corners of the frame before you export. The same family of mistakes covers the paperwork errors: required tags missing, the campaign's mention left out, the clip posted to a platform the brief didn't list, footage pulled from outside the approved source. None of these are editing talent problems. All of them are checklist problems.
So run the checklist below on every clip, out loud if you have to, before anything posts. Sixty seconds. It protects the only reputation number you have — your approval rate — and Modules 6 and 7 will show you exactly what that number is worth. And when a denial does land, read the reason, fix the pattern it points to, and move on. Reposting the identical file and hoping for a friendlier reviewer is not a strategy; it's the same mistake with postage.
A clip comes back denied with the reason "unapproved overlay." Looking at your export, what most likely caused it?
What you can now do
- Build the standard clip: reframe the horizontal source to 9:16 before you cut anything, keeping the speaker's face and gestures in frame when two people trade lines.
- Caption word by word, then proof the auto-captions line by line — most viewers meet your clip with the sound off, and the machine mangles exactly the words that matter.
- Put sound ahead of picture in every close call, and run the rejection checklist in the sixty seconds before you post.