VSL hosting is video hosting built around a sales video that paid traffic is sent to. A general video host stores the file, plays it, and counts views. A VSL host has a longer list: keep the hook alive when the browser blocks autoplay, keep the viewer on the page, show the buy button at the pitch, measure retention per second, and connect each play to the sale it produced in your cart.
The difference matters because every play of a VSL was paid for. A hosting problem on a brand video costs you a view. The same problem on a VSL costs you the click you bought and the sale it was supposed to become.
What makes hosting "VSL hosting"
The file is the same. The player is not. A video sales letter is a linear pitch to a cold viewer with one call to action, and the host's job is to protect that pitch from the first frame to the order confirmation. The table below is the honest version of the comparison, without a competitor's name in it.
| Question | YouTube embed | General video host | VSL host |
|---|---|---|---|
| Does the hook get heard when autoplay with sound is blocked? | No. Muted start. | Usually no. Muted start. | Sound where allowed, muted where not, sound on at the first touch anywhere on the page without restarting. |
| Can the viewer leave through the player? | Yes. Suggested videos and the channel link. | No. | No, and the video plays only on your domains. |
| When does the buy button appear? | Whenever the page shows it. | Whenever the page shows it. | At the pitch second, or after a watch gate. |
| What do the analytics tell you? | Views and average duration. | Plays, watch time, an engagement graph. | Retention per second, split by buyers and non-buyers. |
| Does it know which viewer bought? | No. | No. | Yes. The cart posts the sale back onto the session. |
| Can you test a new hook without changing the ad link? | No. | Sometimes, on click-through. | Yes, under one embed, decided on sales. |
The last two rows are the ones that separate the category. Plenty of hosts have a nice player. Very few can tell you which play made money.
The seven jobs a VSL host has to do
Each of these maps to a moment in the funnel where a general host quietly loses a sale. The order is the order the viewer meets them.
- Get the hook heard. Every modern browser blocks autoplay with sound until the visitor interacts with the page (Chrome, Safari). A VSL host has to start with sound where it can, start muted where it must, and turn sound on at the first touch anywhere on the page without restarting the video. That is what Smart Autoplay does, and Unmute Rate is the metric that tells you whether it worked.
- Start fast. A slow first frame is paid-for silence. The space the video will occupy should be reserved before anything loads, the poster should hold until real pixels replace it, and the host should show you its own startup times per video rather than asking you to trust it.
- Keep the pitch on your page. No suggested videos, no channel link, and a stream that plays only on your domains. Content protection goes further: bots, proxies, and download tools get a decoy video instead of your pitch, so a ripped embed on someone else's lander plays nothing useful.
- Show the buy button at the pitch. A timed CTA appears at the second the offer lands, not on page load. A watch gate holds the order button, or the whole offer section, until the viewer has heard enough to buy.
- Measure per second, and know who bought. A retention curve shows the second viewers leave. The version that matters for a VSL also draws buyers as their own curve, so you can see where buyers stayed and non-buyers left.
- Connect the play to the sale. The player writes a session key into the checkout link. The cart confirms the order and posts it back with that key. The sale lands on the session, the video, the watch depth, the split-test arm, and the ad click that produced it. Cart postbacks are live for ClickBank, BuyGoods, Digistore24, WarriorPlus, JVZoo, and Kiwify, and affiliates who do not own the cart can use the network's postback instead.
- Feed the ad platform and test without breaking the link. Pixels should fire on watch time, not page load, from the browser and from the server with deduplication so one sale is not counted twice. And a split test should run under one embed, so the ad link never changes and the winner is decided on revenue per play.
The numbers that describe the hosting itself
Most VSL analytics describe the video. A smaller set describes the host, and these are the ones to read before you blame the script. They all come from the Metrics Catalogue, where the formula is printed on the card.
| Metric | What it measures | What a bad number means |
|---|---|---|
| Play Rate | Plays as a share of visitors | The page or the player is not getting people to watch |
| Time to Play | How long from load to the first frame | Viewers are paying for silence, and some leave before it ends |
| Unmute Rate | Plays where sound came on | The hook is playing to nobody |
| Video Ready Time | How fast the stream was ready to play | Delivery is slow, usually on mobile networks |
| Buffer Rate | Plays that stalled mid-pitch | The stream is not keeping up with the viewer |
| Error Rate | Plays that failed | The player itself is losing sales |
None of these show up in a play count. That is the point. If the player is the thing losing sales, the Quality tab is the only place that will ever tell you.
What "a play" means, and why the definition is a hosting feature
VSL hosts bill by the play and report by the play, so the definition sets both your invoice and your metrics. Most do not publish one. TrackPlay does, in the Honest Play Standard: a play is one session in which the viewer genuinely played the video. A muted autoplay nobody watched is not a play. A bot is not a play. A reload mid-session is not another play. Billing and every analytics surface report the same number.
Ask any host you are evaluating the same question. If the answer is vague, assume the number is inflated in both places: you pay for more plays than you had, and your Conversion Rate looks worse than it is.
You do not have a hosting problem. You have an attribution problem. VSL hosting exists because the two are the same problem once the traffic is paid for.
Why it matters more for direct response than for anyone else
A brand marketer measures a video library. A direct response marketer measures one video against one ad account. Every play was bought, so a hosting loss is multiplied by the cost per click, and every uncertainty about attribution is multiplied by the ad budget it misdirects.
Consider the pixel. On most pages it fires on load, so the ad platform is taught to find people who load pages, including the ones who bounced at second three. Fire it at a watch-time milestone instead and the platform learns from people who listened. Nothing about the media spend changed. The host just stopped lying to the algorithm.
Now consider the cart. When the sale posts back onto the session, a split test can be decided on revenue per play instead of clicks, an affiliate can see EPC per video from the network's real payout, and the retention curve can show where buyers stayed. Without the postback, every one of those is a guess dressed as a report. If you are still working from play counts, this is why they are a vanity metric.
How to tell whether your host is costing you sales
Run this before you touch the script. Each step rules out one hosting cause.
- Check Unmute Rate and Time to Play. If sound rarely comes on, or the first frame is slow, the hook is not being heard. That is a host problem, not a copy problem.
- Check where the pixel fires. If it fires on page load, your retargeting audiences and your conversion optimisation are built on bounces.
- Check the buy button timing. If it is visible from the first second, viewers can skip the argument. If it never lines up with the pitch, people who are ready cannot act.
- Check the attribution chain. Pick one order from last week and ask your host which play produced it. If the answer is a shrug, your split tests and your retention analysis have no ground truth.
- Check the play definition. Compare the plays you were billed for with the plays your analytics show. If they differ, ask why.
What to look for in a VSL host
The list below is the seven jobs turned into questions. A host that answers all of them with a feature, not a promise, is a VSL host. A host that answers three is a video host with a nice player.
- What happens to the sound when the browser blocks autoplay, and does the video restart to fix it?
- Is the stream locked to my domains, and what does a scraper get?
- Can I time the buy button to a second, and gate the offer behind watch depth?
- Is retention per second, and can it draw buyers as their own curve?
- Which carts post the sale back, and what happens to a postback with no session key?
- Do pixels fire on watch time, server-side, with deduplication?
- Does a split test run under one embed, and is the winner a statistical claim or a suggestion?
- What is the published definition of a play?
If you are weighing a specific host against these questions, the roundups of Vimeo alternatives, Wistia alternatives, Vidalytics alternatives and Voomly alternatives compare the common choices on price, meter and sale tracking.
TrackPlay was built in this order: the conversion layer first, the player around it. The VSL hosting page walks through each job with the actual product, and the free plan gives you 3,000 plays a month, free forever, with no card so you can read your own numbers before deciding anything.