Your launch video will be judged by people who never chose to watch it. It autoplays, muted, into a feed they were scrolling for something else. You get roughly three seconds before the thumb moves on. Everything else in the video — the pacing, the proof, the call-to-action — only exists if those three seconds do their job.
This is the part most founders under-invest in. They spend days polishing the middle of the video and thirty seconds choosing the opening frame. It should be the reverse. The hook is not the intro to the work; it is the work.
The math of the first three seconds
Two facts shape everything about video in feeds. First: most of it plays silent. Industry reporting has put muted viewing on Facebook as high as 85% (a publisher-reported figure Meta never confirmed), and Meta's own tests (back in 2016) found captioned video ads were watched about 12% longer on average. The direction is not in doubt: mute is the default state of your launch video.
Second: the drop-off is front-loaded, brutally. Facebook's research (published back in 2016, still the best public data we have) found that 65% of people who watch the first three seconds of a video go on to watch at least ten seconds, and 45% keep watching past thirty. The first three seconds are not an opening — they are a filter. Win them and most of your audience follows you deep into the video. Lose them and nothing you placed at second eight will ever be seen by anyone.
The same physics apply wherever founders launch — X, LinkedIn, or a Product Hunt launch: silent autoplay, a moving thumb, a three-second trial. So a video hook has exactly one job: give a muted, mid-scroll stranger a reason to stop. Four patterns reliably do that. Three reliably don't.
Four hook patterns that survive the scroll
1. The bold claim
Open with the strongest sentence you would dare put on your homepage — full-screen, high-contrast, no preamble. "Your CI is wasting 40 hours a month." "Invoices should not require software." A claim works because it forces a reaction: agree, disagree, or "prove it". All three keep the thumb still.
The test for a good claim: someone could screenshot your first frame and it would still be interesting on its own. If the opening frame needs the next twelve seconds to mean anything, it isn't a hook — it's a setup.
2. The big number
"14 hours a week." "$30,000 per bad hire." "73% of carts, abandoned." Numbers stop scrolls because they are specific in a feed where everything else is vague. The pattern is number first, context second: the digits land in the first second, the caption explains them in the next two.
One rule: the number must be true and defensible. A shock stat you can't source gets you dragged in the replies faster than it gets you watched — and on launch day, the replies are the whole point.
3. The product doing the impressive thing, frame one
Skip the setup entirely and open on the money moment: the report assembling itself, the deploy going green, the design appearing from a prompt. If your product's core action is visually satisfying, it is the hook — no copywriting required.
This works because a real interface in motion reads as "product" instantly, where stock footage reads as "ad". It's also why building the video from your actual page beats any template intro: your own UI, in your own brand colors, is the most credible opening frame you own.
4. The question that stings
"Still copy-pasting between five tools?" "How much did your last bad hire actually cost?" A pointed question recruits the viewer's own situation into the video before they realize it. The sting matters: polite questions ("Looking for a better way to manage projects?") read as ad copy and get scrolled. The question has to name a pain precisely enough that the right viewer feels caught.
The anti-patterns: how launch videos die in second one
- The logo animation. Your logo is the least interesting frame you own. It means nothing to people who don't know you yet — which, on launch day, is everyone you're trying to reach.
- The slow fade-in. Half a second of black is a sixth of your budget spent on nothing. Feeds punish empty frames immediately and never refund them.
- The "Introducing…" opener. The word announces that the point is coming later. In a feed, later does not exist.
- The narrated cold open. If your first three seconds only make sense with sound, they don't make sense. Muted autoplay is the default, not the edge case.
All four are versions of the same mistake — throat-clearing — which is also the first of the seven reasons launch videos fail. They come from treating the video as a short film with an intro, instead of what it actually is: an interruption that has to pay for itself immediately.
How to test your hook before you post
You don't need analytics to know whether your hook works. Four checks, five minutes, before the video goes anywhere near a feed:
- The mute test. Watch the whole video with the sound off — that is how the majority will see it. If the hook depends on a spoken line, rewrite it as on-screen text.
- The three-second cut. Trim the video to its first three seconds and watch only that. Do you know what the product is, or at least feel a reason to care? If the answer is "not yet", your hook starts too late.
- The frame-one screenshot. Freeze the very first frame. If it shows a logo, a black screen, or a generic gradient, you are opening on a wasted frame — swap in the claim, the number, or the product.
- The stranger scroll. Put the video in a real feed context and hand your phone to someone outside the team. Watch their thumb, not their face. A pause is a pass; polite feedback after a forced full watch means nothing.
Hook first, then everything else
Write the hook before you storyboard anything else. The rest still matters — format, length and captions all compound the hook's work once someone stays. But nothing downstream can rescue a video that lost its viewer before it began. Win the first three seconds and the feed gives you the rest. Lose them and there is no rest.