How to Craft a Product Launch Video Script: A Step-by-Step Walkthrough

Templates tell you what to fill in. This walkthrough shows you how to think. We'll craft a launch video script from zero — starting with five hook candidates, picking one, building the story beat by beat, and ending with a finished 45-second script you can read side-by-side with your own product. The worked example is a fictional Slack-to-action tool called Relay. Swap the nouns and the structure holds.

If you want the blank template first, start with how to write a product video script. If you want the timestamp anatomy without the writing process, read the anatomy of a launch video. This post is the middle path: the decisions behind each line.

The product we're launching (our example)

Relay is an AI that reads Slack threads and turns decisions into tracked action items — assignee, due date, linked context — without leaving chat. Target viewer: a team lead at a 20–200 person startup drowning in async decisions. Placement: homepage hero and Product Hunt. Runtime target: 45 seconds. Sound-off first; voiceover is enhancement.

Step 1: Choose the one belief

Before any hook, finish this sentence: “After watching, the viewer should believe ___.” Not “learn about Relay.” One belief. For Relay, we chose: “Decisions in Slack can become tracked work automatically — I don't have to copy-paste into a project tool anymore.” Every line we write either supports that belief or gets cut.

If you can't state the belief in one sentence, you're not ready to script. You're still positioning. Write three belief candidates, pick the one that would make someone click “Try free,” and throw the others away.

Step 2: Write five hooks — then kill four

The hook is not your tagline. It's the first thing a cold viewer sees in a feed with the sound off. You have about three seconds. We drafted five openings for Relay and scored each against three questions: Does it name a pain or promise a transformation? Could a competitor say the same thing? Does it work as on-screen text alone?

We picked: “Your team decided in Slack. Nothing happened.” It wins because it describes a moment the viewer has lived — a thread full of “sounds good” and “I'll do it” that never becomes work. It's not about Relay yet. That's the point. The hook earns three more seconds.

What makes a hook work (four patterns)

Strong launch hooks usually fall into one of four patterns. You don't need to memorize them — but recognizing the pattern helps you diagnose a weak opening.

• Status quo tension — name the painful normal (“Still copying decisions out of Slack?”).
• Broken promise — something should have happened and didn't (“Your team decided in Slack. Nothing happened.”).
• Hype contrast — everyone talks about X; you show Y (“Everyone ships AI demos. This one runs in production.”).
• Transformation glimpse — flash the after-state before explaining how (“Action items, assigned, in 8 seconds.”).

The best hooks in recent launches mix pattern with specificity. Sana's Admin Mode video opens on “Meet Admin mode” — minimal, but the *next* card is “Done in minutes,” which is transformation glimpsed. Pletor Brain's copy names the villain directly: “Then loses it every time someone opens Claude or ChatGPT.” Specificity beats clever.

Step 3: Build the story — one beat at a time

Once the hook lands, the story has four jobs: make the problem concrete, show the product solving it, prove it's real, and ask for one action. We wrote each beat as two layers — voiceover (VO) and on-screen text (OST) — because most viewers will only see OST.

Beat 1 — Stakes (seconds 3–8)

Goal: the viewer thinks “that's me.” Don't explain Relay yet. Make the cost of the status quo tangible.

• VO: “Decisions live in chat. Tasks live somewhere else. The gap is where work dies.”
• OST: “Decisions in Slack. Tasks in Notion. The gap is where work dies.”

Why this works: it names two tools the viewer already uses — Slack and a task tracker — and personifies the failure (“work dies”) instead of listing features. We cut an earlier draft that said “context gets lost,” which was accurate but abstract. “Work dies” is emotional and short enough for kinetic type.

Beat 2 — Reveal (seconds 8–22)

Goal: show outcome before mechanism. The viewer should see tracked action items appear, then understand *how*.

• VO: “Relay reads the thread, pulls out what was decided, and creates the tasks — assignee, due date, link back to the message.”
• OST (staggered): “Reads the thread.” → “Creates the tasks.” → “Assignee. Due date. Linked.”

Visual direction: open on a messy Slack thread with three unresolved “I'll handle it” replies. A subtle highlight sweeps the decisive lines. Cut to a clean task list populating in real time — three rows, each with an avatar and a “View in Slack” chip. No dashboard tour. One interaction, one outcome.

We resisted listing integrations (“works with Asana, Linear, Jira…”). Integration lists are for pricing pages. The story beat is: thread in, tasks out. Save the rest for the landing page below the fold.

Beat 3 — Proof (seconds 22–35)

Goal: convert skeptics. Skeptics are most of your audience. One believable proof point beats three vague ones.

• VO: “Teams using Relay close 40% more thread-born tasks in the first week.”
• OST: “40% more thread-born tasks closed — week one.”

If you don't have a metric yet, proof can be a recognizable logo strip, a one-line quote, or a before/after screen pair. What doesn't work: “Trusted by leading teams” with no names. We chose a number because Relay is early — but even a beta quote from a design partner (“We stopped losing Friday decisions by Monday”) would beat an empty superlative.

Beat 4 — Close (seconds 35–45)

Goal: one action, one URL, no menu. The close is where most launch videos bleed conversions.

• VO: “Stop re-typing decisions. Try Relay free.”
• OST: “Try Relay free → relay.app”

We cut “Book a demo,” “Watch a webinar,” and “Follow us on X.” One CTA. The brand card holds for two full seconds — longer than feels comfortable — because viewers scrub back to catch the URL.

The finished script

Here's the full 45-second script with timing, copy, and visual direction in one table. Read it top to bottom as a viewer would experience it.

Step 4: Pressure-test before you animate

A script that reads well in a doc often fails on screen. Run these four checks before you open a timeline:

• Sound-off test: hide the VO column. Does the OST alone tell a complete story?
• Stranger test: someone outside your company watches once, muted. Can they repeat the one belief back to you?
• Competitor test: swap your product name for a rival's. If the script still works, your hook isn't specific enough.
• Cut test: delete any line that doesn't move the story forward. Most first drafts lose 20–30% here.

Read every line out loud. Spoken lines should land in 6–12 words. If you stumble, rewrite. Motion graphics turn short phrases into kinetic type — long sentences become unreadable clutter.

Step 5: From script to storyboard

Once the script passes the four tests, each row in the table becomes one scene. Don't add scenes the script doesn't earn. A 45-second launch video should have 5–7 scenes, not 15. For frame-by-frame planning, see how to storyboard a video. For production with AI, how to make a launch video with AI picks up where this script ends.

Adapt this walkthrough to your product

• Replace Relay with your product; keep the beat structure.
• Rewrite five hooks — force yourself past the feature-first instinct.
• State the one belief before you write another line.
• Write OST and VO separately; design for OST-only viewers.
• Cut integrations, pricing, and roadmap from the script — they belong on the page.

Launch video script FAQ

A launch video is a story told in seconds. The hook decides whether anyone hears it; the beats decide whether they believe it. Write the belief first, draft five hooks, build one scene per beat, and cut until every line earns its place. Paste the finished script into Maybe Labs and it choreographs the motion around your words — so the thinking you did here becomes a video that converts, not just one that looks on-brand.