Most long-form videos warm up. There’s a greeting, a bit of context, a question, and then, a minute later, the line everyone remembers. That order works when someone has chosen to watch your episode. It doesn’t work in a feed, where the viewer didn’t choose you and is already halfway through the swipe.
So when I cut a short, the first question I ask isn’t “where does this moment start?” It’s “what’s the most interesting thing in this clip, and how close to the first frame can I put it?” That’s what people mean by a cold open: you start in the middle of the action, or on the payoff, and let the context catch up.
This post covers how I actually do it, where it goes wrong, and one honest caveat from my own research that made me treat it as a tool rather than a law.
What a cold open is in a short
The term comes from TV and film, where a show jumps straight into the story before the title sequence or opening credits. In a short, it usually takes one of three forms:
- Trim the run-up. The moment is already in the right order; you just start later. No “so, um, anyway”, no host re-reading the question. The clip opens on the first word that matters.
- Lift the strongest line to the front. You take a line from later in the clip, play it first, then cut back to the setup and let the exchange play through to that line again.
- Open on the question. If the clip is an answer, start on the question that makes the answer interesting, and drop everything before it.
The first one is the most common, and it’s the one I use on almost every clip. The second is the one people usually picture when they hear “cold open”, and it’s the one that’s easiest to get wrong.
Why the first seconds carry so much weight
Someone scrolling past your short knows nothing about you. In the first couple of seconds they’re deciding whether this is worth their time, based on whatever is on screen and whatever they hear. A greeting or a slow lead-in gives them nothing to decide with.
I saw this clearly when I went through every reel on one public short-form account. The ones that opened on a greeting or a calm, “nothing is happening yet” introduction tended to do badly. The ones that opened on something odd, tense or surprising tended to do well. That was one account in one niche, so I don’t treat it as proof, but it matched what I’d already seen while editing: the clips that felt flat in review were nearly always the ones that took too long to get going.
Client feedback points the same way. On a local interview podcast I edit for, the very first note on my first version was that there was about half a second of silence before the first word. Half a second sounds like nothing. In a feed, it’s long enough to look like a video that hasn’t started yet. Now every clip I deliver starts on the first spoken word.
Payoff first doesn’t mean spoiler first
The title of this post says “start with the payoff”, and I want to be careful with that, because done literally it can kill a clip.
If the payoff is the answer to a question, and you play the answer first, there’s nothing left to wait for. The viewer has the punchline and can leave. What you want at the front is the part of the payoff that raises a question, not the part that settles it.
Three made-up lines to show the difference:
- Say a guest says, “And that’s when I realised I’d been charging half what everyone else was.” That works as a cold open. It makes you want to know how they found out. What doesn’t work is opening on the number they ended up charging, because that answers the question before it’s asked.
- A commentator says, “This is the worst decision in the entire fight.” Good opening line. The explanation of why it’s the worst decision is the reason to keep watching.
- A host laughs and says, “You can’t say that on here.” Strong opener, as long as the clip then shows what was said.
The rule I use: open on the line that makes a stranger ask “wait, what?”, and make sure the clip answers it before it ends. The tension and the payoff both have to sit inside the clip.
What my research did and didn’t show
Here’s the caveat. After studying that one account, I worried I was building rules from a single page. So I looked at a much wider sample: the top and bottom posts from 33 creators who clip their own long-form content, across YouTube Shorts, TikTok and Instagram Reels, 325 posts in total. Each clip was labelled against a fixed checklist without the labeller seeing its views, and only then compared against how it performed.
A cold open, on its own, made no measurable difference. Hits and flops used it about equally.
What did separate them, modestly:
- A question in the first three seconds. Hits opened with one more often.
- Moments a stranger can fully follow. Clips that relied on inside knowledge of the show did worse.
- Two people in frame at the start, rather than a lone close-up.
The labels came from software, not people, and the effects were small, so I treat this as a ranking hint rather than a formula. But it changed how I think about cold opens. Moving a line to the front isn’t what helps. Opening on tension, a question or something a stranger can immediately follow is what helps, and a cold open is often the easiest way to get there. If the clip already starts on a good question, rearranging it buys you nothing.
The mechanics that make it work
When I do lift a line to the front, these details decide whether it feels deliberate or messy:
- Start on the first word, with no air before it. Even a few frames of silence read as a slow start.
- Never clip a word. If the lifted line ends on “insane”, let “insane” finish. A cut-off word at the start sounds like an editing mistake, not a hook.
- Make the jump back obvious. A hard cut back to the setup is usually enough. Some creators like a quick flash, a whoosh or a zoom; that depends on their style, not a rule.
- Let the line play again in context. The repeat is the payoff. If you cut it out the second time, the viewer never gets the full version.
- Keep the whole thing short. A lifted opener should be a sentence, not a paragraph. If it takes five seconds to say, it’s a second clip, not a hook.
When I don’t use a cold open
Some moments are better in their natural order:
- Jokes built on timing. If the laugh depends on the setup arriving first, playing the punchline early just spoils the joke.
- Clips that already start strong. If the first line is a question or a bold claim, leave it alone.
- Lines the creator already used in the episode’s own cold open. Their regular viewers have heard it twice already. I check for this before picking moments.
- Moments that only make sense with context. A cold open won’t rescue a clip that a stranger can’t follow. I wrote more about choosing moments that stand on their own in how to find short-form moments that work without the full episode.
A quick checklist
Before a clip leaves my hands, its opening has to pass these:
- The first frame has a face or an action, not a transition or a blank moment.
- The first word is the first frame, with no dead air before it.
- Within about three seconds, a stranger has a reason to stay: a question, a claim, a surprise.
- Whatever that opening promises, the clip delivers before it ends.
- The ending lands right on the point. Getting the start right and then trailing off at the end wastes it. Length plays into this too, which I covered in how long should a YouTube Short be.
None of this is complicated. It just takes the time to listen to the whole moment, find the one line that carries it, and decide whether it belongs at the front.
If you’d like to see how your own videos open when they’re cut this way, I can make a free short from one of your videos. You keep it either way.