The Complete Guide to SaaS Launch Videos

Most launch videos fail for a dumb, avoidable reason. Nobody decided what the video was actually supposed to do. It got made because a launch was coming up, it lists a bunch of features nobody asked about, and it dies quietly at 2,000 views.

Here's the whole job of a launch video, in one sentence: make a specific person understand a specific change fast enough that they go do one specific thing. That's it. Everything else is decoration you're paying for.

Not every feature deserves the same video

This is the part people skip, and it's the part that saves you the most money.

• Tier 1: the company-wide launch. Once or twice a year, max. Big, cinematic, a real story arc, the whole company rallies around it. This is Apple-keynote energy, not "we shipped a button."
• Tier 2: the flagship feature. A polished 60-90 second video for something that actually changes how people use your product.
• Tier 3: the solid improvement. A quick 20-40 second screen recording with some clean on-screen labels. No voiceover needed.
• Tier 4: the small fix. Skip the video entirely. A GIF and a line in your release notes is plenty.

Deciding the tier before you write a single word of script is what stops you from spending Tier 1 money on a Tier 3 feature. This happens constantly and it's why marketing budgets quietly evaporate.

Four formats that actually work

1. The "before it sucked, now it doesn't" film. Show the painful old way of doing things, let it sting for a second, then cut to the better way. This is basically how Slack sold itself in its early days: show a team drowning in email chaos, then show a calm room where all the work lives in one place. You're selling a feeling, not a feature list.

2. The guided product tour. One continuous walk through your actual product, with motion guiding the eye so people don't get lost. The golden rule: never show a screen someone can't understand in under two seconds. Zoom in, dim distractions, highlight the one thing that matters.

3. The metaphor explainer. Use this when your product is invisible, like most infrastructure or backend tools. Dropbox's classic explainer barely showed the actual software. It showed a folder magically appearing on a different computer. If your product is a dashboard full of numbers, you probably need this format.

4. The "just me talking to camera" video. Low-budget, high-trust. Good for changes that need context or, honestly, an apology, like a pricing change or killing off a feature people liked. People forgive a human explaining themselves way more than they forgive a press release.

The 90-second structure that keeps people watching

1. 0-5 seconds: name the problem. In their words, not yours. Not "introducing X." More like "tired of doing this by hand?"
2. 5-20 seconds: what's different now. The shift, in plain language.
3. 20-60 seconds: show, don't tell. Two or three real moments in the product. One idea per shot. Resist the urge to cram in more.
4. 60-80 seconds: the payoff. What does their week look like now that this exists?
5. 80-90 seconds: one action. Just one. Not "sign up, read the docs, join our community, and follow us on every platform we exist on."

If you can't explain the whole video in one sentence before you write the script, you're not ready to write the script.

Where the video actually needs to live

A launch video only earns its budget back if it shows up everywhere the launch does:

• Right at the top of the landing page, playing quietly on loop, with captions on (most people watch with sound off, always).
• Sitting at the top of your changelog or release notes.
• Chopped into a punchy 15-30 second vertical clip for social, uploaded natively, not linked out to YouTube where it'll never get watched.
• Popping up inside the product itself as a "here's what's new" moment.
• In your sales deck and in the follow-up email your sales team sends after a demo.

That last one is underrated. A good launch video often lives longest as a sales tool, not a marketing moment. Sales teams love having something better than a bullet-point slide.

The mistakes everyone makes

• Cramming in every feature. Six features in 90 seconds means the viewer remembers none of them. Pick one and go deep.
• Showing UI nobody can read. A real screenshot at real size looks like alphabet soup on a phone screen. Simplify it. Think of it as a diagram, not a screen recording.
• Jargon in the opening line. "Unified observability platform" means nothing to a normal human. "Find out why the checkout broke, in one search" means everything.
• Trying to talk to everyone. A video aimed at both developers and finance people is a video that connects with nobody.
• Launching once and moving on. Treat it like a three-week campaign, not a Tuesday afternoon announcement.

What to actually measure

Forget view counts, they tell you nothing. Instead, track:

• How many people who land on the page actually hit play, and how many watch to the halfway point or the end.
• Whether the page with the video converts better than a version without one.
• Whether existing users who saw the in-app "what's new" video actually start using the feature more than those who didn't.
• Whether your sales team is actually sending it, and whether prospects mention it back.

That third one is the easiest experiment you'll ever run, and almost nobody bothers to run it.

A realistic production timeline

Three weeks is doable for a solid Tier 2 video, assuming someone with actual decision-making power is in the room.

• Days 1-3: figure out the one-sentence promise and who exactly this is for.
• Days 4-6: write the script.
• Days 7-10: storyboard it.
• Days 11-17: animation and sound.
• Days 18-21: revisions, cutdowns for social, captions, done.

The part that always runs long isn't the animation. It's the week everyone spends arguing about what the product actually does. Settle that in writing, before anyone opens a design file.

---

If your last launch video felt like it disappeared into the void, that's usually a script problem, not a talent problem. [Get in touch with Icarus](/contact) and we'll help you figure out what your next one should actually say.