# How to start a demo video: the first five seconds

July 18, 2026 · Product Demo Videos · 7 min read · https://aidemo.top/blog/demo-video-hook/

> Viewers decide to leave in the first few seconds. Six opening moves that earn the next fifty, the one intro to retire, and which hook fits muted autoplay.

**Key takeaways**

- The opening is a filter: YouTube's 'intro' metric tracks who is still watching after 30 seconds and advises moving compelling moments earlier, because audiences only shrink over a video's length.
- The throat-clear ('Hi, I'm X, and today I'll show you') is the costliest line: ~22 words at 2.5 words/sec burns ~9 seconds, the window where Facebook's data puts 47% of a video's value.
- Use one of six openings instead: outcome-first cold open, problem-in-one-line, watch-this-happen, number on screen, before/after cut, or named question.
- Match the opening to the surface: autoplay-muted surfaces (hero, feed, README) need silent motion; click-to-play surfaces (Product Hunt, email) hook on the first frame, so pick a populated poster.
- The hook is the most-rewritten line, so define the opening in a script you can edit and re-render, not weld into a take you must re-shoot to test.

## The opening is a filter, not a warm-up

The first few seconds of a demo do one job: decide whether the next fifty happen. They are not throat-clearing before the real video, they are the moment most of your audience makes a keep-or-leave call, and the data on where people leave is lopsided toward the front. YouTube's own analytics name the metric outright: "intro" is the percentage of your audience still watching after the first 30 seconds, and when a compelling moment sits later in a video, the platform's advice is to move it earlier, because "audience sizes typically decrease over the length of the video" ([YouTube Help, 2026](https://support.google.com/youtube/answer/9314415)). Vidyard, which coaches sellers through thousands of demos, puts the window smaller still: "you have, on average, just a few seconds to capture someone's attention," and the best demos "get right to the point ... and end abruptly when the question is answered" ([Vidyard](https://www.vidyard.com/blog/demo-videos/)).

Two structural facts make the opening more load-bearing for a product demo than for a talk or a vlog. First, a demo usually autoplays muted, on a landing hero, in a feed, or looping under a README, so the opening has to land as silent motion and on-screen text rather than a spoken line, which is why [most first views are muted and the caption carries the pitch](/blog/demo-video-captions). Second, a demo competes with a skip reflex, not an audience that already chose to watch. That combination is why the opening is a filter. What follows is a library of openings that pass the filter, the one that reliably fails it, and a rule for which to use where. For how long the whole thing should run and [the retention math behind it](/blog/how-long-should-a-demo-video-be), and for [the end-to-end script and production playbook](/blog/how-to-make-a-product-demo-video), follow the links; here the entire budget goes to the first five seconds.

## The one opening to retire: the throat-clear

Start with the mistake, because it is nearly universal and it is expensive. The throat-clear opening is some version of "Hi, I'm Sarah, and in this video I'm going to show you how our product can help your team work faster." It feels polite and professional. It is the single highest-cost line in the video, and the cost is computable.

English narration runs about 150 words per minute, or 2.5 words a second ([Tools for Clear Speech, Baruch College](https://tfcs.baruch.cuny.edu/speaking-rate/)). That sample greeting is 22 words, so it eats roughly nine seconds. Now price those seconds. Facebook's guidance to advertisers, cited in Vidyard's length research, is that 47% of a video's value is delivered in the first three seconds ([Vidyard](https://www.vidyard.com/blog/video-length/)); YouTube does not even start measuring desertion until 30. The throat-clear spends the most valuable window in the entire runtime, the three-to-nine-second span that decides the view, on three things the viewer did not come for: your name, the fact that a video is about to happen, and a generic benefit claim with no picture behind it.

On a muted autoplay it is worse than wasted, because the words are never heard. The viewer sees a talking head or a title card, a progress bar barely moving, and scrolls. A logo animation, a title card, or a "before we start" preamble is the same mistake wearing a nicer coat. The fix is not a better greeting. It is deleting the greeting and opening on one of the six moves below.

## Six openings that pass the filter

Each of these puts something the viewer wants inside the first frame. Pick by what your product's payoff actually is, and by whether the surface plays with sound off.

| Opening move | The first frame is | Example first beat | Best when |
| --- | --- | --- | --- |
| Outcome-first cold open | the finished result, already on screen | the live dashboard your team fights over every Monday, done, in one link | the payoff is a visible end state |
| Problem-in-one-line | the pain, one sentence over it | "Every deploy, someone re-runs the flaky test and learns nothing" | the pain is sharper than the feature |
| Watch-this-happen | the product mid-action, cursor already moving | a query being typed and a chart redrawing, no setup shown | the core action is fast and legible |
| Number on screen | a single concrete figure | "2 hours to 1 link" held for a beat | you have one honest, specific metric |
| Before/after cut | the old way, hard-cut to the new | a cluttered spreadsheet, then the same data as one clean view | the contrast reads at a glance |
| Named question | the exact question the viewer typed | "Localize a demo without re-recording it?" | your traffic is high-intent and specific |

A few notes on using them. The outcome-first cold open is the safest default for a muted hero: it shows the destination in frame one, then rewinds to how, so a viewer who leaves after four seconds still saw the point. Watch-this-happen is the most demo-native move, because it trusts the interface to be interesting and skips all preamble, but it fails when the action needs context to read, so pair it with a [zoom onto the moment of action](/blog/professional-screen-recordings) so a muted viewer's eye lands where it should. The named question is the highest-variance move: it converts hard when the question is precise and the viewer typed it, and it flops when it is generic ("Tired of manual reports?"), which is the throat-clear in disguise.

Whichever you pick, write it as the first line of the script, not as an afterthought once the recording exists. [The problem-walkthrough-proof template](/blog/demo-video-script-template) budgets the hook its own words at the very top, so the opening is designed rather than salvaged.

## Which opening for which surface

The right move depends less on the product than on where the video plays and whether playback is gated. Two regimes matter, and they invert each other.

On autoplay-muted surfaces, a landing hero, a native social upload, a looping README clip, the first five seconds are five seconds of silent motion. The opening has to work with the sound off, which rules out anything that lives in the narration and rewards motion plus burned-in text. On click-to-play surfaces, playback is gated behind a click or a hover, so the hook is not the first five seconds at all, it is the first frame, the poster. Product Hunt is the clean example: a launch gallery accepts video only as a YouTube link, which is click-to-play, and even animated GIFs "do not autoplay (they animate on hover)," so the visible thumbnail is whatever first frame you chose ([Product Hunt](https://www.producthunt.com/launch/preparing-for-launch)). Choose a frame that shows a populated, mid-action interface, never a login screen, a spinner, or a title card.

| Surface | Playback | What the hook actually is | Openings that fit |
| --- | --- | --- | --- |
| Landing-page hero | autoplay, muted, looping | first 5 s of silent motion plus text | outcome-first, before/after, watch-this-happen |
| Social feed | autoplay, muted | first 3 s of motion | before/after, number on screen |
| README loop | autoplay, muted, no controls | the first loop, no audio ever | watch-this-happen, outcome-first |
| Product Hunt or email | click-to-play | the poster frame you choose | any, if the first frame is populated |
| Sales follow-up | click-to-play, sound on | the first spoken line | named question, problem-in-one-line |

The pattern under the table is one sentence: the more a surface gates playback, the more the still frame carries, and the less it gates playback, the more the opening motion carries. A sales follow-up the prospect clicked, with sound on, is the one place a spoken opening is safe, and even there the first line is a named problem, not a greeting. Match the opening to the surface the same way you [match the length to the placement](/blog/how-long-should-a-demo-video-be), because both are decided by the viewer's attention budget, not by taste.

## Make the opening the cheapest line to change

The hook is the line you will rewrite the most, because it is the one you can only judge against real drop-off once the video is live. That argues for an opening you can change without re-shooting the take. In a hand-recorded demo the first five seconds are welded to everything after them: swap the cold open for a before/after and the cursor timing, the voice track, and the captions all slide out of sync, so testing a second hook means recording the whole thing again. When the demo is defined as a script and an action list, the opening is a few lines at the top of a file, and a new hook is an edit plus a re-render.

Our own engine, aidemo, works this way, with the honest limits that come with it: capture is browser-only, the storyboard is written by a coding agent instead of assembled by hand in an editor, and it ships no visual timeline for tweaking the opening by eye. The principle holds whatever renders the pixels. Treat the first five seconds as a variable you test, not a performance you nail once, and pair a strong opening with [a single, matched closing ask](/blog/demo-video-call-to-action) so the video that hooks well also converts.

## Sources

- [YouTube Help — Audience retention (intro percentage and drop-off)](https://support.google.com/youtube/answer/9314415)
- [Vidyard — How to Make Demo Videos That Win New Business](https://www.vidyard.com/blog/demo-videos/)
- [Vidyard — How Long Should a Video Be? (47% of value in the first 3 seconds)](https://www.vidyard.com/blog/video-length/)
- [Wistia — How to Choose the Right Marketing Video Length](https://wistia.com/learn/marketing/optimal-video-length)
- [Product Hunt — Preparing for launch (gallery, video, and thumbnail guidance)](https://www.producthunt.com/launch/preparing-for-launch)
- [Tools for Clear Speech, Baruch College — Speaking Rate](https://tfcs.baruch.cuny.edu/speaking-rate/)

## FAQ

### What should the first five seconds of a demo video show?

Something the viewer came for, on screen, before any greeting. The strongest openings put the payoff in the first frame: the finished result (an outcome-first cold open), a one-line statement of the pain, the product already mid-action, a concrete number, or a hard before/after cut. Because a landing hero or social clip autoplays muted, the opening has to read as silent motion and on-screen text, so lead with the picture and let the narration catch up.

### How do you start a demo video without saying your name?

Delete the introduction and open on the work. A greeting like "Hi, I'm X, and today I'll show you" spends roughly nine seconds, at 2.5 words a second, on your name and a promise, during the exact window Facebook's data says carries 47% of a video's value. Replace it with a cold open on the outcome or the problem; your name, if it belongs anywhere, goes on the end card next to the call to action, not in the hook.

### Does a demo video need an intro?

No, and most demos are better without one. Logo animations, title cards, and "in this video we'll cover" preambles burn the highest-value seconds on content the viewer did not ask for, and they are invisible on a muted autoplay. YouTube's own guidance is to move compelling moments earlier because audiences only shrink over a video's length. Open on the product, and save any framing for after you have earned the next few seconds.