How to Write a Demo Video Script (Template) | DemoPolish
To write a demo video script: open with the viewer's problem (not your name), then move through 7 sections with strict time budgets. Hook (5–10s), Stakes (10–15s), Promise (5–10s), Setup (10–15s), Demo (45–75s), Result (10–15s), CTA (5–10s). Aim for 90 seconds, 200 to 250 spoken words, one viewer, one problem, one outcome.
The hard part of recording a demo isn't the recording. It's the words.
In 2026, screen recording is a 60-second job. AI voiceover is a 60-second job. Writing what to actually say is the only thing left that takes you four hours, three rewrites, and still feels off in playback.
This post is how to write a demo video script that actually converts. You get a 7-section template with time budgets, 7 rules that separate good scripts from bad, and a copy-pasteable 90-second example you can adapt for your own SaaS.
One rule runs through everything: if your demo opens with "Hi, I'm Sam and today I'm going to show you InvoiceHero," you've lost half your viewers. The hook isn't your name. It's their problem.
(Once your script is ready, DemoPolish turns a rough screen recording into a polished demo with AI voiceover in about 60 seconds. Free for your first video.)
What a demo video script actually does
A demo script has three jobs, in this order:
- Get the right viewer to keep watching past the 10-second mark.
- Connect a problem they have to your product, without selling.
- Leave them with one specific action.
If your script does anything else, cut it.
The most-violated rule of demo scripts: trying to explain everything. The script that lists 12 features explains zero of them. Pick one watcher, one problem, one product, one outcome. Build the whole script around that quartet. If a sentence doesn't serve all four, it goes.
The 7-section demo video script template
A 90-second demo script breaks into 7 sections with strict time budgets. Hit the budgets and the pacing takes care of itself.
1. Hook (5–10 seconds)
Lead with a problem the viewer has right now. No logos, no intro slate, no "thanks for watching."
Example: "Most SaaS founders ship a demo nobody watches past 12 seconds. Here's why."
Trap to avoid: introducing yourself or your company before naming the viewer's problem. Save the brand for the CTA.
2. Stakes (10–15 seconds)
Why does this problem actually hurt? Quantify if you can. Make the viewer feel it.
Example: "Investors who skip your demo at 12 seconds also skip your deck. Same for prospects. Same for hiring candidates."
Trap to avoid: dropping a generic statistic with no specific tie to the viewer's situation.
3. Promise (5–10 seconds)
What this video will show, in one specific sentence. Don't promise the moon.
Example: "In the next 75 seconds, I'll show you a 7-section template that turns a blank doc into a finished script."
Trap to avoid: promising five things. Promise one. Deliver one.
4. Setup (10–15 seconds)
The single use case you're about to demo. Tell the viewer who you're imagining for context, so they can map themselves onto the example.
Example: "Imagine you run a 3-person SaaS that sends invoices. You charge $29 a month. You need a demo for your landing page."
Trap to avoid: generic personas like "imagine you're a marketer." Specific is sticky.
5. Demo (45–75 seconds, around 60% of the script)
The actual product walkthrough. This is where the script earns its keep.
The sub-rule that beats every other rule: narrate intent before action.
- Bad: "Now I'm clicking the button."
- Good: "Now I want to send this to my customer, so I'll click here."
The first one describes the screen. The second one describes the thinking. Your viewer doesn't care what you click. They care why.
Trap to avoid: narrating what they can already see. Your cursor and the UI are doing the visual work. Your script does the cognitive work.
6. Result (10–15 seconds)
Show the after-state. The transformation.
Example: "The invoice is sent, the payment is tracked, and the customer just got a Slack notification, all from one button."
Trap to avoid: listing features instead of outcomes. Features are what the product does. Results are what changes for the viewer.
7. CTA (5–10 seconds)
One action. Not three.
Example: "Try it free at invoicehero.com."
Trap to avoid: "Learn more, watch our pricing video, or book a call." Three CTAs is zero CTAs. Pick the one most-likely to be a yes.
7 rules that separate good demo scripts from bad
- Cut every word a viewer doesn't need to understand the next word. Most demo scripts have 30% slack. Find it.
- Read it out loud. If you hesitate, rewrite that line. The page lies. The voice doesn't.
- Time-cap every section before recording. 90 seconds is the sweet spot, 2 minutes is the ceiling. Go over and you'll be guessing what to cut on the fly.
- Lead with intent, not feature names. "Send this invoice" beats "open the InvoicePro panel." Verbs the viewer wants to do, not nouns from your sitemap.
- Mention your product name 3 times maximum. Once in the promise, once in the result, once in the CTA. Anywhere else and you're selling.
- End with one CTA. Always one.
- Write for the AI voiceover, not for your speaking voice. This is the rule none of the older blog posts cover. TTS engines stumble on contractions, mishandle abbreviations, and run numerals together. Short clauses. Numerals as words ("ten dollars," not "$10"). Periods where you'd put a comma if you were speaking. The script that reads great in your head reads weird out of a TTS engine. Build for the engine.
If your demo currently looks amateur, the script is almost always the cause, not the recording. Here are the common amateur mistakes we see most often.
Got your script ready? Skip the editing rabbit hole. DemoPolish turns the recording into a polished, AI-narrated demo in about 60 seconds.
Worked example: a 90-second script for "InvoiceHero"
Here's the 7-section template, filled in for a hypothetical SaaS called InvoiceHero. Steal the structure, swap the words.
[0:00–0:08] HOOK
"Most solo SaaS founders spend 4 hours a month chasing
unpaid invoices. That's a full work day, every month, on
admin you didn't sign up to do."
[0:08–0:20] STAKES
"And the invoices that bounce around the longest are the
ones that never get paid. Cash flow tightens. The week
gets shorter. You start dreading your own inbox."
[0:20–0:28] PROMISE
"In the next 60 seconds, I'll show you how InvoiceHero
sends, tracks, and chases invoices automatically."
[0:28–0:40] SETUP
"Say you run a 3-person agency. You bill 12 clients a
month at an average of 4,000 dollars. That's 48,000
dollars of invoices going out, with about 7,000 of late
payments to chase every month."
[0:40–1:25] DEMO
"Here's the dashboard. I want to send this month's
invoice to my biggest client, so I'll click 'New Invoice.'
I drop in the line items, and InvoiceHero pulls last
month's terms automatically.
I hit send. Two things happen. The PDF goes to the client,
and InvoiceHero schedules three reminder emails over the
next 30 days, in my voice, on the days when invoices are
most likely to get paid.
If the client opens the invoice but doesn't pay, I get a
Slack ping so I can follow up myself. No automation on
the human relationship. Just the chase."
[1:25–1:35] RESULT
"That 7,000 dollars of late payments? Last month, we
collected 5,800 of it before any human had to send a
reminder."
[1:35–1:42] CTA
"Try it free at invoicehero.com." That's roughly 90 seconds, around 210 spoken words. One viewer (3-person agency). One problem (chasing invoices). One product. One outcome ($5,800 collected automatically). For more on what makes this kind of structure convert, see our demo video best practices.
Free template (copy this)
Empty version. Paste it into a doc and start filling in.
[0:00–0:10] HOOK
[The viewer's problem, named precisely. No intro.]
[0:10–0:25] STAKES
[Why this problem hurts. Quantify if you can.]
[0:25–0:35] PROMISE
[The one specific thing this video will show. No more.]
[0:35–0:50] SETUP
[The specific use case. Numbers, persona, context.]
[0:50–1:35] DEMO
[The walkthrough. Narrate intent before action.
Cut every word the viewer doesn't need.]
[1:35–1:50] RESULT
[The after-state. What changed for the viewer.]
[1:50–2:00] CTA
[One action. One link. Not three.] If your draft runs longer than 2 minutes, the cut isn't in the wording. It's in the sections. Probably setup or result. Read it out loud and feel where you lose interest. That's where to trim.
What to do once the script is written
Three things, in this order.
1. Record without re-reading the script. Read it once aloud. Then perform from memory. Demos read off a teleprompter sound like demos read off a teleprompter. Knowing the script and performing it produces a better take than reading it. (Need the recording side too? Here's our step-by-step screen recording guide for SaaS founders.)
2. Use AI voiceover if narration is your weakest link. Most founders' delivery costs them more than their script does. If you're ESL, if you cringe at your own voice, or if you don't have time to nail eight takes, use AI voiceover. The script is the part that matters. The voice is just delivery.
3. Polish in post. Cut dead air, level the audio, drop a clean intro frame. Or skip all of that and let DemoPolish handle it.
FAQ
How long should a demo video script be?
90 seconds is the sweet spot for marketing demos. 2 minutes is the ceiling. Anything longer and you'll lose the viewer regardless of how good the product is. For tutorials and onboarding videos you can stretch to 5 minutes. But a script destined for a landing page, Product Hunt listing, or cold outreach should run 200 to 250 spoken words and time-cap at 90 seconds.
Should I use AI voiceover or record my own voice?
If your delivery is strong and you're a native speaker, your own voice builds more trust. If it's not, or you're ESL, or you cringe at your own playback, AI voiceover is now hard to tell apart from a decent human reader. It produces a better demo than fumbling through eight live takes. Write the script for the voice you'll use: TTS engines need shorter clauses and numerals as words, while live narration can be more conversational.
What's the difference between a demo video script and a sales script?
A demo script is for video. A sales script is for live conversation. Demo scripts are tight, time-budgeted, one-direction monologues that can't react to the viewer. Sales scripts are talk tracks with branches that respond to objections, questions, and the prospect's body language. If you're confused about which you're writing: if it has to fit under 2 minutes and there's no Q&A at the end, it's a demo script.
How do I know if my demo script is too long?
Read it out loud with a stopwatch. If it runs over 2 minutes, it's too long. The cut is almost always in the setup or result sections, where founders explain too much. The demo section earns its time because it shows the product. The other six sections need to defend every word.
Your script is half the battle. Record your screen, drop the file into DemoPolish, and ship a demo your prospects will actually finish. Try your first video free.