Every guide to video sales letters hands you a framework and stops. AIDA, PAS, the four Ps, a numbered list of beats, and then you are alone with a blank document and a webcam.
So the centre of this page is a complete script, written out, that you can copy and swap your own details into. Around it: how long the video should be and why, what to record it with, where it sits in a funnel, and the four problems a VSL will not solve no matter how well it is written.
How this guide was put together
I read the guides currently ranking for this term and listed what they cover between them, then wrote to fill the gaps they share. Two are worth naming: none of them writes out a finished script, and none of them says plainly when a VSL is the wrong format. Every tool shown below was loaded and screenshotted in August 2026. The two invented case studies that used to sit on this page, a CRM and a nutrition brand with suspiciously precise conversion numbers, have been removed because neither company exists.
What a video sales letter is
A video that does the job a long-form sales letter used to do: take someone from not knowing you to buying, in one sitting, without a salesperson.
It usually sits on a landing page with the buy button underneath it and very little else. That bareness is deliberate. The video is the page.
What separates it from other business video is the job it does. A demo shows a product working. A brand film makes you feel something. A VSL argues a case and asks for money at the end, which is why it is written before it is filmed.
How long it should be, honestly
The stock answer is between 5 and 20 minutes and it is useless, because it averages two different formats.
Cold traffic, higher price. Someone who arrived from an ad and has never heard of you needs the full argument. That runs 12 to 25 minutes and it is normal for the well-known ones to run longer. You are replacing a sales conversation, which is also why the structure of the ad funnel feeding it changes what the video has to do.
Warm traffic, lower price, or a free trial. Someone already on your list, or looking at a 29 dollar product, does not need 20 minutes. Three to six minutes, and past that you are talking people out of it.
The number that matters is not length, it is where people stop watching. Pull the retention graph after a hundred views and look for the cliff. There is almost always one specific moment where a third of the audience leaves, and it is nearly always the point where you started talking about yourself instead of them. Fix that moment and the length question answers itself.
The full script
This is a complete VSL for a hypothetical B2B software product, written to be read aloud in about seven minutes. Replace the bracketed parts. The section labels are for you, not for the viewer, and none of them should be said out loud or appear on screen.
[Hook, 0:00 to 0:20]
"If you are spending more than an hour a week on [the task], this is going to be uncomfortable. Because I am going to show you where that hour actually goes, and it is not where you think. Stay with me for six minutes and you will get that hour back, starting this week."
[Credibility, 0:20 to 0:50]
"I am [name], and I have spent [number] years doing [the task] for [type of company]. Not theorising about it. Doing it, badly at first, and eventually well enough that people started asking how. That is the whole reason this exists."
[The problem, restated better than they can, 0:50 to 2:00]
"Here is what your week looks like. [Specific description of their Monday morning.] Then [the second thing]. And by the time you have finished, [the consequence]. You are not disorganised and you are not slow. The process itself is the problem, and almost everyone doing this job is stuck in the same one."
[Why the usual fixes fail, 2:00 to 3:00]
"So you have probably tried [the obvious fix]. It works for about three weeks. Then [why it breaks down]. You might have tried [the second obvious fix], which fails for a different reason: [reason]. Neither of those is a discipline problem. They fail because they treat [the symptom] instead of [the cause]."
[The mechanism, 3:00 to 4:30]
"What actually works is [the mechanism, named as a thing]. Here is how it works. [Explain in three steps, plainly, in a way that would still be useful if they never bought anything.] That is it. That is the whole idea, and you could build it yourself this weekend if you wanted to."
[The offer, 4:30 to 5:30]
"[Product] is that mechanism, built. [What it does, in one sentence.] You [first thing they do], it [what happens], and you [outcome]. It costs [price]. What you are buying is the [number] hours you would spend building it yourself, plus the [specific thing] you would not have thought of."
[Proof, 5:30 to 6:15]
"[Named customer] runs [context]. Before, [specific before state]. Now, [specific after state]. [Second named customer, one sentence.] Both of them are on the site with their full stories if you want to read them."
[Objections, 6:15 to 6:45]
"Two things people ask. First, [the price objection], and the honest answer is [answer, which concedes something real]. Second, [the fit objection]. If you are [specific situation], this is not for you and I would rather say so now."
[The ask, 6:45 to 7:00]
"Click the button below, [what happens next in concrete terms]. [Guarantee, stated plainly.] If you have read this far you already know whether this is your problem. Go."
Four things about that script that matter more than the wording.
It states the problem before it names the product, and it spends twice as long on the problem as on the offer. It gives away the mechanism for free, which sounds like a mistake and is the single strongest trust move available to you. It raises the two objections instead of waiting for the viewer to raise them silently and leave. And it disqualifies people on purpose, which raises conversion among the people who remain.
Writing yours
Work in this order. Most bad VSLs are bad because they were written in a different one. Then read the draft aloud and check it lands at the level your buyer reads at, which our free reading level checker will tell you in a few seconds.
- Write the offer first. One sentence: what they get, what it costs, what happens after they click. If this sentence is hard to write, the video is not your problem.
- Write the objections next. List every reason a reasonable person would not buy. The three biggest go in the script, answered honestly. An objection you dodge does not go away, it just stops being something you can address.
- Then write the problem section, and make it more specific than feels comfortable. Naming their Monday morning precisely is what makes a stranger think you have met them.
- Write the hook last. You cannot hook someone into an argument you have not made yet. Write it after you know what the argument is.
- Read it aloud and cut everything you stumble on. Written sentences and spoken sentences are different shapes. Anything you cannot say in one breath gets cut or split.
- Time it. Read at your real speaking pace, not your reading pace. Roughly 140 words a minute.
If drafting from a blank page is the part that stalls you, an AI video script generator will produce a serviceable first pass to react against, which is usually easier than starting cold. The objections and the proof still have to be yours, because those are the parts that require knowing your actual customers.
What to record it with
The equipment question absorbs far more attention than it deserves. In order of what actually affects conversion:
Audio, which is most of it. Bad audio reads as amateur within three seconds and viewers leave without knowing why. Any 60 to 100 dollar USB microphone in a room with soft furnishings beats an expensive microphone in an empty one. This is the only piece of gear worth thinking about.
Your script, which you have already written.
A recent phone or a webcam. Genuinely fine. Put a window in front of you, not behind you.
Editing software. For a talking head with a few captions, Camtasia and Descript both handle it, and Descript's approach of editing the video by editing its transcript suits VSLs particularly well because you are cutting words rather than frames.
A host that reports retention. This matters more than the camera. Vidyard and Loom both give you a per-second drop-off graph, which is the single most useful piece of feedback you will get.
If you are not going on camera at all. Pictory and InVideo assemble a video from a script and stock footage, and Sendspark records once and personalises the opening for each recipient, which is aimed at outbound rather than at a landing page. These produce a watchable video from a script quickly. They will not produce a persuasive one from a weak script, which is the same constraint as everything else on this page.
| Tool | What it is for | Free tier | Watch out for |
|---|---|---|---|
| Descript | Editing the video by editing its transcript | Yes, limited hours | Transcript editing tempts you into over-cutting |
| Camtasia | Screen recording plus a full timeline editor | No, one-off licence | More editor than a talking head needs |
| Loom | Fast recording with viewer analytics | Yes | Built for messages, not landing pages |
| Vidyard | Hosting with per-second retention data | Yes | The analytics are the reason to use it |
| Pictory | Script to video with stock footage | Trial only | Generic footage looks generic |
| InVideo | Template-driven video assembly | Yes, watermarked | Templates read as templates |
| Sendspark | One recording personalised per recipient | Trial only | Aimed at outbound, not at a page |
Pick the host before the editor. Retention data changes what you make next, and an editor only changes how the current one looks.
The page the video sits on
The video is not the whole job. The page around it decides a surprising amount.
Keep it close to bare: headline, video, buy button, and little else. Every additional link is an exit. Put the button under the video and again at the bottom, and if the video runs long, reveal the button at the point in the video where the offer is made rather than at the start.
Do not autoplay with sound. Do autoplay muted with captions burned in, because a large share of viewers start watching without sound and decide whether to turn it on within a few seconds.
Give the page a text version of the script underneath, collapsed. It costs nothing, it serves people who will not watch a video, and it gives search engines something to read on a page that is otherwise a video file. If you are building that page from scratch, our comparison of AI landing page builders covers the tools that will get you there without a developer.
Measuring it
Three numbers, in this order.
- Retention curve. Where do they leave? Fix the cliff, re-record that section only, and re-test. This is the highest-return loop in the whole exercise.
- Click rate on the button among people who reached the offer. This isolates the offer from the video. If they watch to the end and do not click, the problem is the offer or the price, and no amount of re-recording will help.
- Conversion from click to purchase. If this is low, the video promised something the checkout does not deliver.
Test one thing at a time and start with the first 30 seconds, because that is where the largest share of the audience is lost and it is the cheapest part to re-record.
Four things a VSL will not fix
Worth knowing before you spend a week on this.
An offer nobody wants. A VSL makes a good offer legible. It does not make a bad one appealing, and a well-produced video for a weak offer just gets you rejected faster.
No traffic. A VSL converts visitors. It does not produce them. This is the most common way the effort is wasted: a good video on a page nobody reaches.
A price and product mismatch. If the price is wrong for the market, the video will surface that fact rather than fix it.
A broken checkout. Obvious, routinely true, and worth testing on your own phone before you write a word.
That second one is worth being blunt about, because it is where most of these projects actually stall. Distribb is on the traffic side of that line: it handles the SEO content and publishing that gets people to the page in the first place. It has nothing to do with video, scripts or hosting, and it will not help you make the thing described on this page. If your VSL is fine and nobody is seeing it, that is the half it addresses.
VSLs worth studying
Three that are real, documented, and still instructive.
Dollar Shave Club. The reference case for humour carrying an argument rather than replacing one. Watch how quickly it gets to the price, and how much of the runtime is spent on objections disguised as jokes.
Squatty Potty. Proof that a product people are embarrassed to discuss can be sold by meeting the embarrassment head on rather than talking around it.
Slack's early explainer work. Stacked short user testimonials instead of a single narrator. Useful if your product is hard to describe but easy to demonstrate through other people's relief.
Watch each with the script structure above open and mark where the sections change. The beats are all there.
Frequently asked questions
Do I have to be on camera?
No. Screen recordings with voiceover and script-to-video tools both work, and for software products a screen recording often works better than a face. What you cannot skip is the script.
How much should one cost to make?
Your first one should cost the price of a microphone and a weekend. Agencies charge from a few thousand up, and that is worth considering only after a self-made version has proved the offer converts. Paying for production before you know the argument works is the expensive mistake.
How long before I know if it works?
A few hundred views gives you a readable retention curve. A meaningful conversion read needs a few thousand, depending on price.
Should the video autoplay?
Muted with captions, yes. With sound, no.
Can I use the same VSL for cold ads and my email list?
You can, and you should not. The cold version needs the problem section in full. The warm version can skip most of it and often converts better at a third of the length.
What if my retention drops in the first ten seconds?
Then your hook is describing you rather than them. Rewrite the first two sentences so they name the viewer's problem before they name you, and re-record only that part.
Where to start
Write the offer sentence. Write the three objections. Write the problem section longer than feels right, then the hook. Read it aloud, time it, record it on whatever you have, host it somewhere that shows retention, and look at the graph on Friday.
That is a week of work and it will teach you more about your own offer than another month of tweaking the page.