A good Product Hunt launch video opens on the product already running, works with the sound off, and lands its strongest moment inside the first ten seconds. Sixty to ninety seconds is plenty — the browser deciding whether to click your launch has already judged you in the first few.
That judgment environment is what makes the Product Hunt asset different from a homepage explainer. Same craft, different physics: the viewer is skimming a feed of launches, the video sits in a gallery next to screenshots, and almost nobody arrives with sound on. This guide covers what actually plays there, the length and structure that survive it, and the failures we see in launch galleries every week.
Where the video actually plays
On launch day your video lives in three places at once, and none of them is a cinema:
- The gallery slot. The video sits first in your launch page's media gallery, next to screenshots. A visitor who clicks through from the feed sees a thumbnail and a play button — the video has to earn the click with its first frame, then earn the next ten seconds with motion.
- Social feeds. The same video gets clipped and posted, and every feed autoplays it muted. Whatever your narration says, the feed viewer gets the picture only.
- The scrub bar. Launch-day viewers are triaging dozens of products. They drag the playhead, land somewhere in the middle, and give you two seconds. A video that only makes sense watched start-to-finish makes sense to almost nobody that day.
The consequence: the Product Hunt cut is judged frame by frame, in silence, out of order. Build for that and the full-attention viewer is served for free; build for the full-attention viewer and the skimmer sees a slideshow.
Assume the sound is off
Sound is opt-in nearly everywhere the launch travels, so the picture has to carry the claim alone. Three rules make a video mute-proof without turning it into a text wall:
- The picture shows state; motion is the language. A run firing, a table filling row by row, a status flipping green — these read at a glance with no audio at all. Across sixty-odd produced explainers, the accepted builds all share a surface that visibly fills as the video progresses; that accumulation is legible from any point a scrubber lands on.
- Labels name things; they never explain them. "Table" on a table is fine. A sentence on screen restating narration is the text-wall failure — you've made a slide deck, and slides don't demonstrate. If a muted viewer can read your whole pitch off the frames, the video isn't showing anything.
- Ship captions anyway. For the minority who do want the words, captions beat forcing sound. Every video we deliver includes them as a standard deliverable, and on launch day they're doing real work in the feeds.
The honest test: watch your cut muted, at 2x, starting from the 40% mark. If you can still tell what the product does and what just happened, the gallery will treat you well.
The length that works on launch day
The measured corpus across sixty-odd produced videos puts a single-concept explainer at 54–124 seconds, typically 60–80. For the Product Hunt slot we stay at the short end of that — 60 to 90 seconds — with one structural change: the first ten seconds have to work as a complete argument on their own.
The math behind the front-load. Narration reads at about 1.9 words per second, so the first ten seconds hold roughly one sentence and one picture. That's enough for the strongest single beat you have: the machine running once, the before/after landing, the result arriving. Everything else — the mechanism, the second angle, the closer — plays to the smaller audience that the first ten seconds earned.
In our experience the failure is almost never a launch video that's too short. It's a 2.5-minute video where the product first does something interesting at second 40 — which on launch day is a video of nothing. The length guide covers the general scene-count math; for Product Hunt, the added rule is simply that the scenes are sorted with the payoff first.
Hook-first structure: run the machine
Of the two opens that pass review — the problem-exposition open and the machine-running open — the Product Hunt asset wants the second, almost always. A launch-day viewer clicked your launch because the tagline already named the problem; they're here to see the thing. Opening with a problem setup re-sells someone who's already through the door.
The machine-running open, from an accepted script in our corpus: "Here is a real workflow. A trigger, a file parser, an agent, and a database. Watch how a file moves through it." Three moves in two breaths — orient, enumerate what's on screen, issue one watching instruction — and the run fires in second one. The full anatomy is in the hooks guide; the launch-day version compresses to:
- Second 0–3: the product on screen, already mid-demonstration. Something moves in second one.
- Second 3–10: one run, end to end, with the payoff staged large. Our legibility floor from review: a climax smaller than about 1/40 of the frame didn't happen — and a gallery thumbnail shrinks everything further, so stage the payoff larger still.
- Second 10 onward: the mechanism, macro to micro, for the viewers who stayed. One idea per scene, each scene 8–12 seconds.
- The closer: behavior, never a slogan. What the viewer can do now, which on launch day usually means: this is live, go try it.
The failure modes we see in launch galleries
Every one of these maps to a rejection that actually happened in our review corpus — Product Hunt just punishes them faster.
The throat-clear. Logo animation, mission statement, "we believe…" The highest-attention seconds your launch will ever get, spent on the lowest-value content you have. The feed viewer is gone before the product appears.
The trailer-voice video. Suspense fragments and epic music over a product demo. One rejected draft in our corpus opened "From out here, it's already over" — the note that came back: this is an explainer, not marketing. On Product Hunt it's worse than dead weight, because the audience is product people who have watched a thousand launch videos and can smell selling where showing should be.
Fake UI. Dashboards the product doesn't have, invented values, a cursor clicking controls that don't exist. This is the single most expensive failure on Product Hunt specifically: the audience knows real interfaces on sight, and early adopters who sign up will meet your actual product within the hour. A launch video is a product claim; staged UI is a claim the product can't keep. If a moment can't be shown with your real surfaces, change the moment, never the honesty.
The text-wall. On-screen sentences carrying the pitch. It feels like a solution to the muted-autoplay problem and it's the opposite: you've shipped a slide deck with a soundtrack, and the mute test exposes it — a viewer can read the whole video and watch none of it.
The dead middle. Everything animates in the first three seconds of a scene, then holds frozen while the voice finishes. One measured bad build in our records spent 44% of its runtime fully static. Launch-day scrubbers land in those frozen stretches at random, see a still image, and leave. Cap dead holds at three seconds and keep ambient motion — a drift, a ticking cell, a working dot — in every hold longer than one.
The five-things video. "Five ways X changes your workflow." Counts date, the count is never the lesson, and a listicle structure guarantees no single moment is the strongest one. Name the capability; make one claim; the broader launch guide covers the one-claim discipline in depth.
FAQ
How long should a Product Hunt launch video be? 60–90 seconds, with the strongest beat in the first ten. Our produced corpus runs 54–124 seconds for single-concept videos; the launch slot rewards the short end plus a front-loaded payoff, because most viewers are skimming, not watching.
Does the video need a voiceover if everyone watches muted? Yes — a meaningful minority does turn sound on, and the narration carries the why the picture can't show. But the picture must never depend on it. Write narration to the finished visuals, ship captions, and pass the mute test before you pass anything else.
Should we use the founder talking to camera? A short one works as a frame, not as the video: 25–30 seconds ≈ 70–80 words, state the subject plainly, then hand off to the product running. What kills launches is three minutes of talking head where the product appears as a 30-second afterthought.
Can we reuse our homepage explainer as the Product Hunt video? If it opens cold on the machine running, yes. If it opens with a problem setup or brand build-up, cut a launch version: same scenes, payoff moved to the front. That's a re-sequence, not a new production.
If you're staring down a launch date, send us your product's URL and you'll have twenty candidate directions to judge within 24 hours.