A VSL, or video sales letter, is a long-form sales video that does the job a written sales page used to do. It presents a problem, tells a story, explains the mechanism behind the solution, makes an offer, and asks for the sale, usually with a buy button that appears at a set moment in the video.
VSLs are the standard format for direct response offers sold through paid traffic: supplements, information products, software, coaching. The format works because the viewer cannot skim it. They hear the pitch in the order you wrote it, or they leave. A good analytics setup tells you which second they left at.
What makes a video a VSL
Not every video that sells something is a VSL. A brand film, a product demo, and an explainer all show up on sales pages, and none of them are video sales letters. Four things separate a VSL from the rest.
- One offer. The whole video exists to sell one thing at one price.
- One call to action. There is a single next step, usually a buy button or an order form, and the video tells the viewer exactly when to take it.
- A scripted sequence. Hook, problem, mechanism, proof, offer, close. The order is deliberate, and the viewer cannot reorder it.
- Cold traffic. The viewer typically arrives from an ad and has never heard of the product. The video has to do the whole job.
The name is literal. Direct response marketers spent decades writing long-form sales letters. In 2007 the copywriter Jon Benson, who is widely credited with the first VSL, put the text of his sales letter for The Every Other Day Diet on plain slides, read it aloud, and found the video outsold the page. The format has since grown talking-head, animated, whiteboard, and hybrid variants, but the underlying structure is still a letter.
| Format | Who watches it | Goal | How you know it worked |
|---|---|---|---|
| VSL | Cold or warm traffic from ads and affiliates | Make the sale, in this session | Conversions and revenue attributed to the play |
| Explainer | Website visitors, trial users | Explain what the product does | Sign-ups, support tickets avoided |
| Webinar / replay | Registered leads | Teach, then pitch at the end | Attendance, pitch-time drop-off, sales |
| Ad creative | Feed scrollers | Earn the click | Click-through rate, cost per click |
| Product demo | Evaluators, buyers late in the decision | Prove it does the thing | Demo requests, close rate |
The anatomy of a VSL
Most VSLs follow the same skeleton because it is the sales-letter skeleton. The exact labels vary by copywriter. The jobs do not.
| Section | Its job | What the retention curve shows when it fails |
|---|---|---|
| Hook | Earn the next 30 seconds from someone who was about to leave | A cliff inside the first few seconds |
| Problem and story | Make the viewer feel the cost of not solving this | A steady slide before the mechanism arrives |
| Mechanism | Explain why this solution works when others did not | A drop right after the claim, if belief broke |
| Proof | Evidence, demonstrations, credentials | Little drop-off, or a long flat stretch nobody needed |
| Offer and price | Reveal what you get and what it costs. This is the pitch. | Buyers hold, non-buyers leave in a cluster |
| Guarantee and close | Remove the last reason not to act, then ask | Viewers stay but do not click |
The moment the offer and price are revealed is the pitch, and it is the most important timestamp in the video. Everything before it is spent getting the viewer there. Everything after it is spent converting the ones who stayed. In TrackPlay the pitch second is a first-class setting: it drives when the timed CTA appears, and the retention graph marks the median pitch moment so you can see how many viewers actually reach it.
The numbers that describe a VSL
A play count describes your ad. The video needs its own numbers. These are the ones TrackPlay prints on every video card, with the formula on the card, and they map one to one onto the anatomy above.
| Metric | Formula | What it tells you about the VSL |
|---|---|---|
| Play Rate | plays ÷ visitors | Whether the page and player get people watching at all |
| Unmute Rate | plays with sound on ÷ plays | Whether the hook is heard or plays silently to nobody |
| Hook Rate | plays still watching past 5 seconds ÷ plays | Whether the opening earns the next minute |
| Hold Rate | plays that reached the midpoint ÷ plays | Whether the story and mechanism carry viewers toward the pitch |
| Completion Rate | plays that reached 90% ÷ plays | Whether the close is being heard |
| Conversion Rate | conversions as a share of plays | Whether the pitch converts the people who reach it |
| Revenue per Play | revenue ÷ plays | What one play of this video is worth, in money |
Two things make these numbers trustworthy. First, the definition of a play has to be published and honest: one session, not one event per reload, and a muted autoplay that nobody watched does not count. TrackPlay publishes its definition in the Honest Play Standard. Second, the conversion has to be the cart's sale, posted back onto the session that produced it, not a click on the buy button. A click is intent. A sale is money.
A worked example
The figures below are illustrative, not a benchmark. Use them to see how the metrics chain together, then read your own.
Say a 16-minute VSL gets 10,000 plays from paid traffic in a week, with the pitch at 11:40 and a $97 offer.
- Hook Rate of 72% means 7,200 viewers are still there past five seconds. 2,800 left inside the hook.
- Hold Rate of 30% means 3,000 reached the eight-minute midpoint.
- 1,400 viewers are still watching at the pitch marker, 11:40.
- 140 orders come back from the cart, attributed to those plays. Conversion Rate is 1.4% of plays, but 10% of the people who reached the pitch.
- Revenue is $13,580. Revenue per Play is $1.36.
That last number is the one that talks to your ad account. If a play costs you more than $1.36 in media, the VSL is losing money regardless of how the video "performs". And the chain above tells you where the biggest gain is: the 2,800 people lost in the hook never had a chance to be in the 140.
Why the VSL is the format for paid traffic
A VSL is where ad spend turns into revenue, or does not. That makes it different from every other video a business publishes: every second has a cost attached, because someone paid for the click that started it.
It also makes it the most measurable format there is. The click carries an id from the ad platform. The play starts a session. The session reaches the pitch or it does not. The cart posts the sale back. When those four things are joined, you can say which ad, which video, and which second produced each dollar. That is the reason a VSL owner should care about video analytics more than a brand marketer does: the attribution is not a nice-to-have, it is the whole business model.
This is also why hosting a VSL is a different problem from hosting a video. A VSL host has to keep the hook alive when the browser blocks autoplay, keep suggested videos off the page, time the buy button to the pitch, and carry the session through to the cart.
Your host can tell you how many people pressed play. It cannot tell you which one of them bought. For a VSL, that is the only question.
How to diagnose a VSL that is not converting
Work down the chain in order. Each step rules out one explanation before you touch the script.
- Play Rate and Unmute Rate first. If people are not playing, or the video is playing muted to a page nobody is watching, the script is not the problem yet. Every modern browser blocks autoplay with sound until the visitor interacts with the page (Chrome, Safari), so a VSL that relies on sound-on autoplay can die before the first word is heard.
- Hook Rate. The share still watching past five seconds. If this is the weak link, no edit further down the video will matter, because those viewers never see it.
- The retention curve. A slope is normal. A cliff is a specific second where something went wrong. Open the retention graph, switch to buyers versus non-buyers, and look for the point where the two curves separate. Where buyers stayed and non-buyers left is usually a belief problem. Where both left is a boredom problem.
- The pitch. How many viewers reach the pitch marker, how many see the CTA, how many click it. A high reach and a low click means the offer is the problem. A low reach means the video is the problem.
- The cart. Conversions attributed back to the play, not clicks on the button. If clicks are healthy and sales are not, look at the checkout, not the video.
If you are still working from play counts and an average watch time, start with why play counts are a vanity metric. An average cannot show you a cliff.
How to improve a VSL
- Fix the hook before anything else. Every metric downstream is multiplied by it. Make sure it is heard: smart autoplay attempts sound first, falls back to muted, and turns the sound on with the visitor's first touch anywhere on the page, without restarting the video.
- Cut what the curve tells you to cut. Find the cliff, watch that exact section, and remove or rewrite it. Then check the new version's curve against the old one on the same traffic.
- Line the button up with the pitch. A CTA card timed to the pitch second shows the order button when the viewer has heard the offer, and hides it from the viewer who has not.
- Test one change at a time, on revenue. A split test decided on Revenue per Play, with the same traffic sent to each arm, is the only way to know an edit helped. A test decided on clicks can pick the wrong winner.
- Feed the ad platform watch time, not page loads. A pixel that fires on page load teaches the algorithm to find people who load pages. A pixel that fires at a watch-time milestone teaches it to find people who listen.
None of this fixes a bad offer. It tells you the offer is the problem, which is a cheaper thing to learn from a retention graph than from a month of media spend.
If you want to see these numbers on your own VSL, the free plan gives you 3,000 plays a month, free forever, with no card.