How do you make a launch video?

A launch video makes one claim — this exists now, and here is what it does — and proves it by showing the real product running once, end to end. It sits one altitude above an explainer: the explainer teaches how the machine works; the launch video establishes that the machine exists and is worth a look.

Getting that altitude right, and getting the video done by a date that won't move, are the two problems every launch video actually has. This guide covers both: the launch-vs-explainer split, the one-claim discipline that keeps the message sharp, what deadline production really looks like in the numbers, and the cut order when time runs short.

Launch video vs explainer: the altitude difference

Our production curriculum separates two energy archetypes, and the split maps exactly onto launch vs explainer:

• Concept explainers are calm, diagrammatic, one-structural-idea pieces. They enrich the viewer's causal model — after watching, the viewer can predict what the product will do. They live on docs pages and in onboarding sequences, and they get watched by people who already decided to care.
• Launch videos excite. The machine runs end to end, framed as a real outcome. They get watched by people deciding whether to care, which means the video's job is orientation and appetite, never mastery.

The practical rule that falls out of this is an altitude ceiling. A launch video shows what the product does and when you'd reach for it — never how the algorithm works under the hood. Mechanism appears only at the depth needed to make the behavior believable: one run, one visible cause and effect. The moment your launch script starts explaining internals, you're writing the explainer that comes after the launch. Save it; you'll need it in week two.

Same craft rules govern both. One idea per scene, real product surfaces only, narration written to finished visuals. What changes is the question each video answers: the explainer answers "how does this work?", the launch video answers "what is this and why now?" A video that tries to answer both does neither — mixed-altitude videos read as a demo that keeps interrupting itself with a lecture.

If the launch is happening on Product Hunt specifically, the gallery has its own physics — silent autoplay, skimming, scrubbing — worth reading alongside this.

The one-claim discipline

Every script in our pipeline opens with a mandatory field: the one idea — a single declarative sentence the whole video argues, with everything on screen existing to make it feel obvious. For launch messaging this discipline is the whole game, because a launch is the one moment you get to define what the product is in the viewer's head, and a viewer keeps exactly one sentence.

Three tests for a launch claim, straight from how we test thesis sentences:

1. Is it falsifiable? "Our platform is powerful" fails — nothing on screen can prove or disprove it. "You connect a source once and your agent's knowledge updates itself instead of going stale" passes: a video can show that, and a viewer can check it.
2. Does every scene serve it? The claim is a deletion criterion. Any scene that doesn't push the sentence toward "obvious" gets cut, no matter how good it looks. Launch videos bloat precisely because the team wants to show everything they built; the claim is what gives you permission to leave nine features out.
3. Never frame by count. "The five things our new version does" is a list, and a list has no strongest moment. Counts date, and the count is never the lesson. Name the capability and argue it.

One more rule from the corpus that matters double at launch: calibrated claims only. If you can't support "10x faster," the line becomes "faster," or better, the video just shows the timing and says nothing. A launch video is a set of product claims your earliest, most skeptical users will verify within the hour — every uncalibrated adjective is a small trust debt that comes due on day one.

Deadline production: what the numbers say

Launch dates create a specific production trap: the deadline pressures you to skip review, and skipping review is the single most expensive thing you can do. The real numbers from our production records:

• About half of all production work is post-review response. Roughly 27% of mainline production commits are revision-shaped, and counting parallel rejected attempts, revision is half the spend. A first-pass build is about half the work; plan the calendar that way.
• Flagship videos took three staged versions each — a first build, a rebuild after review, a design-fix pass. The launch videos that look effortless were re-made, not touched up.
• The still gate catches most fatal errors at ~5% of the cost. Before any motion, narration, or final render, review two static frames — the money shot and the final frame. One production batch that skipped this gate shipped nine rejects out of ten; the next batch, with the gate, passed five of five on first review. A bad frame costs minutes to catch as a still and hours to catch as a finished render.

The deadline translation: with two weeks to launch, the gates matter more, not less, because you can't afford the full-render-then-reject loop even once. Review the scene list before anything is built (structure failures are free to fix on paper), review stills before motion (framing and fake-UI failures at 5% cost), and only then spend the expensive hours. Under deadline, the cheap gates are how you buy back the revision half of the budget.

One more deadline lever: volume plus selection beats polishing one draft. Our production unit was never a single attempt — it was a parallel batch of complete candidates, reviewed together, winners promoted. Ten directions generated and judged in a day converges faster than one direction iterated for a week, because picking is faster than fixing. This is the mechanism our whole candidate process is built on, and it's why a tight launch date is a reason to start wide, not narrow.

What to cut when time is short

There's a correct cut order. Working from cheapest sacrifice to dearest:

1. Cut scenes, never verbs. A shorter video is fine; amputated narration is not. When the runtime must shrink, delete whole ideas — the worked example goes before the concept, the second angle goes before the first. Compressing prose into fragments to keep all the ideas produces captions, not speech, and reads exactly as rushed as it was.
2. Cut the custom voice. A good house voiceover is a solved line item; custom recorded VO adds a booking dependency you don't control. Recording the founder can wait for the director's cut.
3. Cut the sound design. Silence beats bad sound. A full sound-effects pass we once shipped — 25 cues, carefully mapped — was rejected wholesale because the assets were stock UI-demo clicks and whooshes. Perfect integration of cheap assets still reads cheap. Music bed plus narration is a complete, respectable launch soundtrack.
4. Cut the second video, and say so. The mechanism deep-dive, the feature tour, the onboarding walkthrough — each is a real video with its own claim, shippable in the weeks after launch. Naming what you're not making is a scope decision; running out of time for it is an accident.

And the things that never get cut, at any deadline:

• Real product surfaces. Every rejected take in a four-topic comparison we ran had invented its display surfaces; every accepted take used the real ones. Fake UI is faster to build and it costs you the launch — the audience most likely to watch a launch video is the audience most likely to know your product on sight.
• The still review. It's the cheapest gate you have. Five minutes of looking at two frames is not where the schedule gets saved.
• The mute test and the hook. The first ten seconds and the sound-off legibility are where the entire audience is; a launch video that saved time there saved time on the only part everyone watches. The hook structure is not optional garnish.

FAQ

How long should a launch video be? 60–90 seconds for the main asset. Our produced single-concept videos measure 54–124 seconds, typically 60–80, and a launch claim is by definition a single concept. Anything past two minutes means the video is carrying an explainer's payload; split it and ship the explainer after launch.

How far ahead of the launch date should production start? Work backward from the numbers: a first pass is about half the total work, and good videos take a review cycle or more. In our experience three weeks is comfortable, two is workable with disciplined gates, and inside one week you should be cutting scope by the list above, in order.

Should the launch video and the homepage explainer be the same video? They can start as the same build, but they answer different questions at different altitudes — "what is this and why now" versus "how does it work." The efficient path is one set piece, two edits: launch cut with the payoff front-loaded, explainer cut with the full causal chain. A demo and an explainer split the same way.

What if the product isn't finished by the time the video is due? Show only what's true. A launch video is a set of claims; a staged feature is a claim the product can't keep, and early adopters find out immediately. Ground every on-screen value in the real product, and if a feature isn't real yet, it isn't in the video.

If the date is set and the video isn't started, submit your product's URL — twenty candidate directions come back within 24 hours, and picking is faster than briefing.