A product launch video is not a homepage video

Most founders make one product launch video, put the same file on the homepage, and wonder why launch day went quiet. Two different objects, two different jobs. A homepage video is watched by someone who already clicked. A launch video has to survive a feed: autoplayed muted, in a list beside dozens of other products, on a date that does not move.

Fifteen years of making promo and product videos — and a disproportionate share of that work happened inside the last ten days before someone else's launch date.

On a homepage, context is free

Your homepage viewer arrived on purpose. Your logo is in the corner, the headline above the player, pricing one scroll down. That's why a slow-burn opening works there: you can spend three seconds inside a problem before naming the product, because the page already did.

Your launch viewer gets none of that. They're triaging a list. Your video is a small rectangle between two other small rectangles, and the name under it is a word they've never seen, in grey, at eleven pixels. Whatever they learn in the next second, the video has to teach them.

The launch cut is also disposable, which is the useful part. A homepage video has to age for a year; a launch video has one shift.

The first frame is a thumbnail before it is a frame

For years, frame zero of my work was a coloured background. Nothing else. I never noticed, because I never looked at it as a still.

I was designing openings as motion. Headline easing up from twelve pixels below on a 400ms delay, product panel scaling in from 0.96 behind it, the whole thing assembling over about a second. Lovely on the timeline. Paused at 00:00 it said nothing.

On a homepage that's survivable, because playback starts around the time the viewer looks. In a feed, frame zero is a poster: what sits there before autoplay engages, what shows on a slow connection. Nobody art-directs it.

So art-direct it. Look at it at the size a card renders. Two type sizes in that frame, not three: at that width a third stops reading as hierarchy and starts reading as texture. Put the product name with the biggest element, not alone in a corner where it reads as a watermark. Crop the product to one recognisable piece of UI, because nobody parses a whole window at card size.

The fix in the timeline is small. Let the composition exist at t=0 and animate inside it: a number counting up, a row dropping into a list, a toggle flipping. Or hold the finished frame eight frames before anything moves. You gain the most valuable frame you have and lose almost nothing in feel.

Launch day rewards clarity over cleverness

The withheld reveal is a homepage move. A line about how you've been doing this wrong for years, a beat, a turn, then the product. It works on someone leaning in. In a feed it spends a stranger's second on a promise you haven't paid off, and most don't wait.

On launch day I want the product named and legible almost immediately, on screen and not only in the voiceover, with the category attached in the same breath. Not "introducing" over a logo sting: a name in frame while something useful is already happening behind it.

Then one specific problem, in words a person would actually say. Then the product doing the thing, once, at a size you can read. Then one action.

Cleverness isn't banned; it moves into the phrasing of the problem line and out of the structure, because structural cleverness costs beats and the beat is what you don't have. Same argument as the first three seconds, made in a worse room.

That's agency work made for a Product Hunt launch, and the structure is the lesson. Watch it twice. First time, count the seconds until you could say out loud what the product does; that number is the whole brief for a launch cut. Second time, pause at five random points. Every pause should land on a frame holding one idea, never on a handoff where the last thing is still leaving as the next arrives. Overlapping beats are how a dense video reads as frantic — more on why timing does this.

The end card is also the loop point

A homepage end card is a signature: logo, one line, a destination still there in March. A launch end card has a date on it. Ask for the thing that only matters today, keep the product name beside it, and hold it a second longer than feels right. Then remember the feed loops back to frame zero and plays again. If the card is a full-bleed dark panel and frame zero is white, that join reads as a fault rather than a repeat.

Assume nobody hears it

Feeds autoplay muted. Assume that's what your video is. On a launch cut the on-screen text carries the argument and the voiceover supports it, which is the reverse of how I'd build a homepage video.

Don't caption the voiceover verbatim, though. A transcript in a feed is a wall of small type moving faster than anyone reads. Write the on-screen line as its own object, one idea per beat, arriving with the word rather than trailing it.

And every audio cue you were leaning on is silence now: the click when the action completes, the tone when the result lands. Each needs a fast visual stand-in. A confirmation should land in about 150ms with a touch of overshoot while everything around it eases over half a second or more. That gap in duration tells a silent viewer which moment mattered.

The vertical cut ships the same morning

Launch day is never one surface: the launch page, X, LinkedIn, Shorts. Cropping the wide file to 9:16 throws away most of the width and takes your UI type with it, which I went through in why cropping your 16:9 promo doesn't work.

What changes for a launch is the schedule. The vertical version can't wait for the wide one to be signed off, because they ship the same morning. Author them in parallel. And 9:16 is where the product name has to be biggest, because nothing else on that screen carries it.

That one is ours, designed and rendered from a URL with no editing pass afterwards. Run the two tests on it, on your phone: pause the first frame, then mute it. I'd rather you applied them to my work than took my word.

The honest limit: the system reads your site, so it inherits your site. If your homepage never says plainly what the product does, the script's opening line will be vague in the same way. That's the line I rewrite by hand before rendering. If Product Hunt is your surface, the format details are on our Product Hunt launch video page.

The date does not move for a revision round

The last argument here isn't about craft. It's about the calendar.

Bespoke agency animation commonly runs $5,000 to $25,000 and takes three to six weeks. That buys real things: a creative director with actual taste, a producer whose whole job is pushing back on you, and a motion language invented for your brand and nobody else's. I've worked in that tier and I'd defend it.

But a launch has a date. If round two lands eleven days out and you want a different opening, you're choosing between shipping a cut you don't believe in and moving a day you've announced. Moving isn't an inconvenience, it's unpicking coordination: the emails said Tuesday, the slot was booked for Tuesday, the changelog is queued for Tuesday. More on that in our take on launch strategy.

If your launch is eight weeks out, defines the company, and wants live action or a founder on camera, hire the studio. Genuinely. Nothing I sell replaces that.

In the time-boxed case, what matters is being able to bin the first version on Thursday morning and have a different one by Thursday afternoon. That's the slot PromoHyper was built for: you paste your product URL on Thursday morning and have a different cut by the afternoon, in 16:9, 9:16 and 1:1. From $69 for one video, or $99 for three if you expect to bin a couple, and everything before the final render is free. It won't do actors, live action, or a talking head with your face on it — the rest of the spec, and the tests I'd judge it by, are in how to choose a promo video maker.

Make two cuts. The homepage one can take its time. The launch one names the product in the first breath, works as a still, works in silence, and is built tall as well as wide. If there's only room for one this week, build the launch cut. The homepage will still be there on Wednesday. Launch day won't.