What Makes a Good Launch Video: Six Teardowns and What One Really Costs

Founders have been asking the same question on Hacker News for seventeen years. In January 2009, someone asked how you produce product demo videos. In May 2024, someone asked how people create those sleek looking demos for startups and collected 395 points and 167 comments doing it.

The answers, both times, were tools. Screen recorders, editors, zoom plugins. Which is a reasonable answer to the question that was asked and a poor answer to the question underneath it, because the launch videos everyone actually remembers were not won on tooling. One of them was a screen recording made by a founder with no budget.

So this is the other article: six launch videos taken apart, what they have in common, and what the equivalent costs to commission today.

If you want the writing structure first, our 7-beat blueprint is the underlying skeleton, and the six launch video formats covers which shape to pick. This one is about judgement — how to tell a good cut from a polished one.

The one test a launch video has to pass

A viewer who has never heard of you should be able to finish the video and say, out loud, what your product does and who it is for.

That sounds trivially low as a bar. It is not. Most launch videos fail it, and they fail it in a specific way: they establish a mood, name a category, and never show the thing working. The viewer comes away knowing that a company exists and that it has good taste, which converts approximately nobody.

Everything below is a variation on passing that test under different constraints.

Six launch videos, taken apart

1. Dropbox (2007) — the video that stood in for the product

Drew Houston recorded a screencast walking through file sync — no jargon, no feature list, just a series of ordinary situations resolved by the product. He posted it to Digg and Hacker News with in-jokes seeded for both audiences.

The beta waitlist went from roughly 5,000 to 75,000 overnight, a number retold constantly in startup circles precisely because the production value was so low. There is no music cue, no motion design, no colour grade.

What it does structurally: it demonstrates rather than describes, and it demonstrates in the viewer's context rather than the product's. The video is a sequence of "here is a moment from your week, here is what happens now."

What to steal: the ordering. Situation first, product second. Almost every weak launch video inverts this.

2. Dollar Shave Club (2012) — $4,500 and a voice

Michael Dubin wrote the script, starred in it, and had a friend shoot it in a single day for $4,500, as he told Inc.. It went live on 6 March 2012.

More than 12,000 orders landed in the first 48 hours and the servers went down. Unilever acquired the company in 2016 for a reported $1 billion.

What it does structurally: it makes one claim — good blades, low price, no hassle — and repeats it in four different registers. The warehouse walk is a single continuous spine that lets each gag land without a cut to reset attention.

What to steal: the discipline of one claim. A $4,500 video cannot afford a second idea, and it turns out neither can a $40,000 one.

3. Slack (2014) — the video that isn't about your product

Slack hired Sandwich to make a case study and got a mockumentary instead. The subject is Sandwich's own team: sceptical, then converted, then slightly embarrassed about how converted they are.

What it does structurally: it hands the argument to someone who is not you. The video opens on scepticism — the Sandwich team explicitly says products like this never work — which buys enormous credibility for the second half. Slack's own claims are never spoken by Slack.

What to steal: the objection-first opening. Naming the reason someone would dismiss you, in their words, before you answer it.

4. Apple Vision Pro (2023) — 51 million views, features last

The reveal became the most-watched video on Apple's YouTube channel, passing 51 million views and overtaking the Apple Watch Series 8 introduction. It is not embeddable here, so: watch it on YouTube.

Its order of operations is worth writing down, because it is almost exactly backwards from how most launch videos are cut:

The specifications — the part most companies open with — arrive last, after the viewer already wants the thing.

What it does structurally: it establishes a new mental model before naming a single component. You cannot appreciate a micro-OLED display until you understand what it is for.

What to steal: put the capability before the mechanism. If your video opens on architecture, model names or an integrations grid, you have inverted this.

5. Arc (2024) — motion doing the explaining

The Browser Company has run two video-led launches worth studying. In January 2024 it shipped Arc Search, whose "Browse for Me" button, as TechCrunch described it, reads at least six web pages and assembles a fresh page out of them. In October it followed with the Act II film, announcing a browser that browses for you.

What it does structurally: it solves the hardest problem in software marketing, which is that the product is invisible. A browser that browses for you produces no photographable moment. So the meaning is carried by motion — pages assembling themselves, results collapsing into a single view — rather than by narration explaining what you would be seeing if you could see it.

What to steal: when there is nothing to film, animate the mechanism, not the brand. Abstract shapes drifting behind a headline are decoration. A page visibly building itself is an argument.

6. Google Gemini (2023) — the one to learn from backwards

In December 2023 Google published a demo of Gemini reasoning about drawings and objects in what appeared to be real-time conversation. Within about a day, TechCrunch and others established that the interaction had been assembled from still frames and text prompts, with the spoken exchange added afterwards and latency edited down. Google had disclosed some of this in the video description. A Gemini co-lead said the aim was to inspire.

What went wrong: nothing in the frame. The product could genuinely do most of what was implied. What failed was the gap between the experience depicted and the experience available, and the gap was discovered by the audience rather than declared by the company.

What to steal: the inverse. State the seams. "Sped up for time," "concept," "coming this autumn" cost you nothing in a launch video and cost enormously when someone else supplies them for you.

What the good ones have in common

Five traits survive across all of them, from a $0 screencast to a nine-minute Apple film.

1. Proof beats promise. Every video that worked put the actual behaviour on screen. The one that didn't work put a depiction of behaviour on screen. Viewers are unusually good at telling these apart, and unforgiving when they catch the difference.

2. One idea, held. Dollar Shave Club has one claim. Dropbox has one mechanism. Arc has one metaphor. None of them attempt a feature tour. Feature lists belong on a pricing page, where the reader is already convinced and now wants specifics.

3. The first five seconds do work. Not a logo. Not an establishing shot. Either the product mid-action, or a single line naming the change. Wistia's 2025 State of Video found that in sub-60-second videos, a call to action placed in the first quarter converts close to 40% of viewers — attention is front-loaded and so should be everything else.

4. Motion carries meaning. In the launches with nothing to film, the animation is the explanation — pages assembling, files syncing, results collapsing. Motion used as decoration reads as a screensaver. Motion used as mechanism reads as a demo.

5. The seams are declared. Say what is a mock-up. Say what ships later. The credibility you spend disclosing is a rounding error against the credibility you lose when a journalist discloses for you.

What a launch video actually costs in 2026

Prices below are gathered from studio pricing surveys and freelance rate guides. Treat them as the shape of the market, not quotes.

Two independent reference points support the middle of that table. A 2025 studio survey of motion design pricing put the average 60-second project at $3,130, with a spread of $1,000–$16,000 and a sharp geographic split: $6,460 average at US companies, $4,170 in Europe, $1,880 in Asia. Per finished minute, the same survey puts 2D motion graphics at $4,000–$8,500 and 3D at $5,000–$9,500. Aggregated agency data puts motion graphics broadly at $900–$4,000 per minute, with early-stage companies clustering at $1,500–$3,900.

For freelancers, School of Motion's long-running rate guide prices by day rather than by project: $350/day for someone recently out of school, $500/day with a couple of years and a decent reel, $650–$1,000/day for experienced multidisciplinary designers, and $1,500–$2,000+/day for specialists and directors. A 60-second launch piece is realistically five to fifteen designer-days, which is how you get from a day rate to the project ranges above.

At the top, Sandwich — the shop behind the Slack video — has been reported to start around $50,000, with larger engagements well into six figures and a history of trading part of the fee for equity. That is not a market rate; it is a scarce-brand rate, and it is priced against the outcome rather than the hours.

Why a 30-second video doesn't cost half of a 60-second one

Roughly 70% of the effort is fixed — the concepting, the script, the storyboard, the asset build and the visual system all happen once regardless of runtime. Length is comparatively cheap after that. This is why quotes rarely scale linearly, why cutdowns are the best value in the whole process, and why a fifteen-second teaser can genuinely cost more per second than the film it came from.

The costs quotes leave out

• Revisions. Budget for them explicitly; they can consume around 12% of the total, and structural changes requested late — after animation has started — routinely add 30% or more, because you are paying to rebuild finished work.
• Voiceover. $50 to $1,000+, with most decent professional reads landing around $250–$450.
• Music licensing. Usually a small line item, occasionally the thing that blocks a launch date.
• Cutdowns and reframes. Plan 9:16 and 1:1 versions before animation. Reframing a finished 16:9 composition for vertical usually means rebuilding it, not cropping it.
• Your time. The largest hidden cost in the DIY route by an order of magnitude, and the one nobody puts in the spreadsheet.

The questions founders keep asking

The Hacker News threads are the honest record here, and what is striking is how stable the questions are across fifteen years.

"What tools do people use for these?" The 2024 thread's answers converged fast: Screen Studio for the automatic zoom-on-action look, Arcade for interactive walkthroughs, Rotato for 3D device mock-ups, CleanShot X into Premiere for people who already edit, Camtasia and Loom-plus-DaVinci for everyone else. Notably, almost nobody in the thread recommended After Effects to the person asking — which is roughly the conclusion we reached in motion graphics without After Effects.

"Should I make the video before the product?" A 2023 thread titled "Always 'Show and Tell' Your Product Before You Develop It" drew 167 points arguing yes — the Dropbox pattern, essentially. The caveat from the Gemini teardown applies: showing a concept is fine, implying it ships today is not.

"Is it worth paying for?" The recurring answer is that it depends entirely on whether the video is doing sales work or brand work. A video that has to explain an unfamiliar mechanism repays craft. A video that restates something obvious does not, at any budget.

A checklist for judging your own cut

Watch your video with the sound off, then again from the top, and answer honestly:

1. In the first five seconds, does anything happen — or is it a logo?
2. Could a stranger say what the product does after one viewing?
3. Is there a moment where the actual product does the actual thing?
4. How many distinct ideas does the video contain? If the answer is more than one, cut until it isn't.
5. Does the motion explain something, or is it moving because motion is nice?
6. Is anything shown that doesn't exist yet — and is that labelled?
7. Does it name who it is for, explicitly, at least once?
8. If you removed the voiceover, would the visuals still carry the argument?
9. Is there a single clear next action, and does it appear early rather than only at the end?
10. Would you send it to someone you respect without a caveat in the message?

Question 10 is the one that catches most videos.

Making one in GenMotion

The reason we built GenMotion is the gap this article keeps circling: the videos that work structurally are often simple, and simple is still expensive when it means a studio and a six-week schedule.

You describe the launch, and GenMotion generates a fully editable motion composition — real timeline, real keyframes, real text and colour you can change — rather than a flat render you have to re-commission to fix a typo. That covers the case the teardowns above make hardest: a product with nothing to film, where the motion has to do the explaining. Export in 16:9, 9:16 and 1:1 from the same composition, so the cutdowns cost minutes instead of a change order.

If you want to see the format before writing anything, the showcase has full examples, and the free tools turn a single number — stars, downloads, subscribers — into an animated clip in the browser without an account.

Where to go next

• How to make a product launch video — the 7-beat script blueprint and the four production routes
• Product launch video examples — the six formats and how the beats are allocated in each
• Motion graphics without After Effects — the tooling question the HN threads keep asking
• Glossary — aspect ratio, frame rate, easing and the rest of the vocabulary a quote will use