A VSL should be exactly as long as it takes to sell the offer to someone who has never heard of you, and not one second longer. There is no universal number, and any "ideal length" quoted without data behind it is a guess. The right length for your video is written in its retention curve: it is too long at the precise second the people who would have bought stop watching, and it is too short if they reach the pitch without enough reason to click.
So this article will not tell you twelve minutes or forty. It will tell you what sets the length, how to find yours in your own data, how to cut without cutting the sale, and how to test a shorter version on revenue instead of completion.
Why the number everyone wants does not exist
Ask ten VSL owners and you will get ten confident answers, all different, all drawn from the last video that worked for them. That is not evidence. It is one script, one offer, one traffic source, remembered fondly.
The length of a video sales letter is an output, not an input. It falls out of how much a cold viewer has to be told before the price sounds fair. A $37 ebook with a familiar promise needs very little. A $997 program built on a mechanism the viewer has never heard of needs the mechanism explained, the objections handled, and the proof shown, and all of that takes minutes. Your 42-minute VSL has one job, which is to get the viewer to the offer. Whether it needs 42 minutes to do that is a question only your viewers can answer, and they answer it every day by leaving or staying.
What actually sets the length
Before you look at the curve, it helps to know which way each part of the offer pushes. None of these is a rule. Together they explain why two good VSLs can differ by half an hour.
| Factor | Pushes shorter | Pushes longer |
|---|---|---|
| Price | Low ticket, impulse range | High ticket, a decision the viewer will justify to someone |
| Mechanism | A familiar promise with a familiar method | A new mechanism that has to be taught before it can be believed |
| Audience temperature | Retargeting, email list, people who know you | Cold traffic from an ad, no prior trust |
| Proof required | Claims the viewer already accepts | Claims that need demonstrations, credentials, or evidence |
| Objections | One obvious objection | Several, each needing its own answer before the close |
| Where the pitch sits | Early pitch, short close | Late pitch, long guarantee and bonus stack |
Notice that the traffic source is on the list. The same script is too long for a warm audience and just right for a cold one, which is why a single answer to "how long" was never going to work.
Your curve already knows
The retention curve is the length question answered per second. For every five seconds of the video it shows the share of viewers still there, and three points on it decide whether the length is right.
- Hook Rate, the share still watching past five seconds. Nothing about length matters until this holds. A hook problem looks like a length problem because the total watch time is low, but no cut further in will fix it.
- Hold Rate, the share that reached the midpoint. This is the length checkpoint. If the curve slides steadily from the hook to the midpoint, the middle is longer than the story it is carrying.
- Pitch reach, the share still watching at the Pitch marker. A VSL is the right length when the people who would buy are still there at this second. Everything after it is for the ones who have not decided yet.
The decisive view is buyers against non-buyers. Switch the curve to show them as two lines. Where the buyers' line holds and the non-buyers' line falls, the video is doing its job and the leavers were never going to buy. Where the buyers' line falls before the pitch, the video is too long for its own customers, and that is the cut to make first.
Completion Rate is the wrong goal for a sales video. The viewer who clicks the buy button at the pitch and never sees the last ten minutes is your best viewer.
Cutting without cutting the sale
Look for two shapes on the curve. A cliff is a specific second where a large share of viewers leave at once. Watch that section. It is usually a claim that broke belief, a sale asked for too early, or a stretch where nothing happens. A dead minute is different: a flat, slow slide where the story stopped moving and viewers drifted off one at a time. Both are edits, and both are located to the second, which is more than a play count or an average watch time will ever give you.
Once a video has passed roughly 300 plays, AI Optimize reads the transcript against the real retention, clicks, and conversions and flags the dead zones worth trimming, with the CTA placements to go with them. It does not replace watching the section yourself. It tells you which section to watch.
After the cut, publish the new version to the same traffic and read the new curve against the old one. The measure of a good cut is not a higher Completion Rate. It is the same or better pitch reach with the same or better Revenue per Play.
When longer is the right answer
Sometimes the curve says the opposite. Three patterns argue for adding, not cutting.
- Buyers rewind. If the buyers' sessions show viewers scrubbing back to rewatch a claim, that claim is carrying weight and probably deserves more proof, not less time.
- High pitch reach, low click. Plenty of viewers hear the offer and do not act. That is a belief gap or an objection left standing, and the answer is usually more of the right minute, not fewer minutes overall.
- High click, low sales. If the CTA is clicked and the cart does not confirm, the problem left the video at the click. Look at the checkout before touching the length.
Test the length on revenue, not on completion
The honest way to settle a length argument is a split test. Clone the video, cut one arm, and run both under the same embed so the ad link never changes and each visitor sticks to one version. Set the goal to conversion rate, the sales per play that the significance check runs on, and read Revenue per Play beside it for each arm. The short arm will always win on Completion Rate, because it has less to complete, and that proves nothing about money. Wait for the significance flag: the winner ships on its own once it clears, or you can declare it yourself. A coin flip that ended early is not a result.
Making a long VSL feel shorter
If the offer genuinely needs the time, the player can stop the length from being the first thing the viewer notices. Three settings do most of the work.
- A smart progress bar fills quickly in the opening minutes and slows later, so the video feels shorter at the moment the viewer decides whether to stay. Most VSL owners hide the timestamp for the same reason.
- A timed CTA means the pitch is not hidden behind the whole runtime. The viewer who is ready at the pitch can act at the pitch.
- Continue Watching means a viewer who left at minute 18 is offered Continue or Restart when they come back, for 90 days, on any device. For a long video, the second session is often where the sale happens.
Playback speed is a fourth lever. Viewers keep their own speed control, and TrackPlay's Turbo engine can run 1.0x against a slightly faster speed as a split test decided on real conversions, which is the only way to know whether faster delivery helps your offer or hurts it.
The one-paragraph version
Ignore the number. Fix the hook first, because length is invisible until it holds. Then read Hold Rate and pitch reach on the buyers' curve: too long where buyers leave before the pitch, too short where they reach it and do not click. Cut cliffs and dead minutes to the second. Add proof where buyers rewind. Settle any argument with a split test on Revenue per Play. If you want to read that curve on your own video, the free plan gives you 3,000 plays a month, free forever, with no card.