the short version
- Pause at two seconds. If a stranger can’t say what the video is about, the rest of it doesn’t matter yet.
- The first frame needs a person or a result, words you can read with the sound off, and movement. A logo does none of these.
- The same test applies to a launch film, a creator video and an App Store preview. The shapes of those are different; the opening isn’t.
- When a video is quiet, change the opening, not the whole thing. Three hooks on one body teach you more than three new videos.
In this post6 sections
People don’t decide to watch your launch video. Their thumb decides, in about the time it takes to breathe in, and the decision is made with the sound off.
That’s uncomfortable if you spent two weeks on the middle of the film. It’s also useful, because the opening is the cheapest part to change. This is the test we use on a cut before it goes anywhere: two seconds, muted, on a phone.
It is not a second copy of the other two posts. How to structure a 45-second launch film is the shape of the whole piece. How to brief creators is the five hook formats and the one-page brief. Use those to make the video. Use this to find out if the start of it will be seen.
Pause at two seconds
Put the cut on your phone. Turn the sound off. Play it. Pause as soon as you could have scrolled past, which is about two seconds, not at a neat edit point.
Now answer, without pressing play again:
- What is this about, in one sentence?
- Is there a person, or a result, on screen?
- Can you read every word without squinting?
- Did anything move?
- Could you tell which app it was, or is it still a pretty picture of nobody in particular?
If you hesitate on 1, the opening is a mood. Moods get scrolled. If you fail 3, the type was set on a laptop. If you fail 4, autoplay looks like a poster that forgot to start. If you fail 5, you made a video for people who already care, and they aren’t the people in the feed.
Do this with someone who doesn’t work on the app, if you can. Your team will hallucinate the missing sentence, because you already know it. A friend won’t.
What the first frame has to be
Frame zero is what shows before the tap, and the first second is what shows if they never tap. They are almost the same picture. It has to work as a still and as a moving image.
A person, or a result. A face mid-sentence. A finished thing (the meal plan, the tidied tab, the photo after). Not your logo, not a wordmark on a colour, not a slow fade up from black. Logos are for the end, when someone has a reason to remember the name. At the start they look like an ad, and ads are what the thumb is trained to skip.
Words, big, few, still long enough to read twice. One short phrase. High contrast. Away from the edges, where the platform’s own buttons live, and away from the bottom, where captions and interface chrome sit. The film post’s rule holds here too: if it looks a bit too big in the edit, it’s about right on a phone. Hold it long enough to read twice. Reading once is a glance. Twice is a memory.
Movement in the first second. A head turn, a card landing, a thumb tapping, a hard cut. Not a slow Ken Burns zoom on a still, unless the still is so strange that the stillness is the point (it usually isn’t). Feeds start playback on their own. A frame that doesn’t change looks paused, and paused looks skippable.
The product, soon. By the end of those two seconds, a person should have a hint of the app: a screen, a result the app just made, or the name said in the words on screen. You don’t have to open with the UI. You do have to arrive. A beautiful two seconds that could be an advert for a trainer is a beautiful two seconds for the trainer company.
The opening doesn’t have to explain the app. It has to make the next second feel like a loss if they leave.
The same test in three places
The test doesn’t change. What you’re allowed to do in the opening does.
| Place | What “stop” looks like | What to watch |
|---|---|---|
| Launch film | Someone stays for the problem, not just the logo | The poster frame is the first frame. No three-second ident. The six beats start immediately |
| UGC video | The first line could be a person, not a brand | One of the five formats, in their words. No “download now” yet |
| App Store preview | The hero moment, muted, with a poster that works as a still | Up to 30 seconds, and it replaces your first screenshot, so it has to sell like frame one |
A launch film can take a second to push in, if something is already moving and the words are already up. A creator video usually can’t. The creator’s face or the result wants to be there at frame zero, because the surrounding feed is other faces and other results. An App Store preview has no luxury at all: it’s standing in for a screenshot, in a row of screenshots, with the sound off by default.
Export a poster frame for each cut and look at it next to the moving version. If the poster is a blank colour and the movement only starts later, the poster is the thing people see in the grid. Fix the poster, not the caption under it.
Re-cut the opening, leave the body
When a video doesn’t hold, the instinct is to reshoot. Usually the middle is fine and the first two seconds are doing nothing. That’s good news.
Keep the body. Cut three different openings onto it:
- The claim. One sentence, to camera or in type, that a person could disagree with. Then the body, unchanged.
- The result first. Open on the finished thing, then cut back to how it happened. The body already shows how. You’re just starting at the end.
- The specific moment. “It’s 11pm and the fridge has been opened four times.” Smaller and more exact than “meal planning is hard”. Then the same demo.
Same middle, same end, three doors. Post them as separate videos, not as a collage, and not all in the same hour. Give each one a name in the folder that says which opening it is (split-claim, split-result, split-pov), so that next week you can remember which door worked without rewatching twelve files.
You won’t learn anything from one quiet post. You will learn something from three openings on one idea. That’s the whole trick, and it’s the same one in the creator brief: judge a batch, and when something works, make more doors, not a new house.
A test you can run tonight
You don’t need a new edit to start. Take whatever you already have, including a rough phone video.
If you have no footage yet, the test still tells you what to shoot first: the face, the result, and one line of words. Shoot those before you shoot the tour of every screen. You’ll use the tour in the middle, if you use it at all.
What we won’t pretend
We can’t tell you a video will travel. Nobody honest can, from a thumbnail. What we can say, because it’s the promise on our UGC work and not a story about a past launch: if a video doesn’t hit 10k views within 2 weeks, you get your UGC money back. The way there is more openings, tested, with the winners kept. Not one precious cut.
A launch film is 30–60 seconds, two weeks, from £3,000. A UGC batch is 10–20 videos, one to two weeks, from £1,500. Both are on the menu, and either one can be the thing you start with.
If you already have a cut, book a free 15-minute call and we’ll watch the first two seconds with the sound off. That’s the meeting. Bring the file, or just the first line.
