To increase VSL conversions, fix the parts of the funnel where most people leave: the first few seconds of the video, the headline and player around it, the button under it and the order page after it. Make the first seconds repeat the promise from the ad, add captions for people watching on mute, show proof under the video, and test when the button appears. Then stop guessing: write 5 to 10 different hooks, run them against each other, and keep the one that makes the most revenue per visitor (or books the most calls per visitor).
I've sold info products with video sales letters for years. My publishing business did over $20M in revenue on more than $10M in ad spend before I sold it for seven figures, and the VSLs, order pages and upsells in it were split tested. This post is the list I work through when a VSL funnel isn't converting, in the order I'd do it, with real tests from my funnels and from a friend who runs eight-figure VSL offers.
If you're still deciding whether a video is right for your offer at all, start with my comparison of a VSL vs a long-form sales letter. This post assumes you already have a VSL and want it to sell more.
Key takeaways
The first seconds decide most of it. Repeat the ad's hook in the first line of the video, and get to the promise in the first 30 seconds.
Most people watch on a phone with the sound off. Captions and a big tap-to-unmute prompt aren't extras, they're how the video gets watched at all.
Test the button timing. A button that shows up too early lets people skip the pitch. Too late, and ready buyers leave.
The order page is part of the VSL funnel. One change on my order page took earnings per click from $3.03 to $4.68.
Find the drop-off before you fix anything. Your video host's retention graph plus Microsoft Clarity (free) show you where people leave.
Judge on money, not plays or clicks: revenue per visitor, or calls booked per visitor for book-a-call funnels.
1. Match the ad hook to the first seconds of the video
Whatever made someone click the ad is the only reason they're on your page. If the ad says "The 3-step piano method for adults who quit as kids" and the video opens with "Hi, I'm Sarah, and I've been teaching piano for 20 years", the visitor has to wonder if they're in the right place. Many of them won't wait to find out.
Why it matters: the first seconds are where the biggest single drop happens. In Wistia's audience retention analysis, the average engagement drop in the first 2% of a video is 4.9% for videos one to two minutes long, and 17.3% for videos five to ten minutes long. A VSL is usually longer than that, so the opening has to earn the next minute, every time.
What to do: write the video's first line as a direct continuation of the ad. Same promise, same words where you can. If you run three ad angles (say, "quit as a kid", "learn in 15 minutes a day" and "play songs you love by ear"), you'll often get more from three openings of the same VSL, each matched to its ad, than from one generic opening.

2. Get to the promise in the first 30 seconds
Once they've pressed play, the next job is to tell them what they'll get by watching. Not your life story and not a long "before we begin". The promise, who it's for and why they should keep watching.
Why: people decide quickly whether a video is for them. A long intro costs you the impatient viewers, and on paid traffic most viewers are impatient.
A simple structure for the first 30 seconds of a course VSL:
The hook (matches the ad): "If you quit piano as a kid, this is for you."
The promise: "In the next few minutes I'll show you the 3-step method that has adults playing their first full song in 30 days."
The reason to stay: "Stay until the end, because I'll show you the one thing every adult beginner gets wrong."
And if a video doesn't work at first, rewrite before you give up on video. After my upsell video test (number 6 below), my takeaway was that when a video flops, it's most likely the script, what you're saying, not the fact that it's a video. My high-converting VSL script format post goes section by section through the whole thing.
3. Put a headline above the video
The headline above the player is the first thing a visitor reads, before they decide whether to press play. On a phone, it's often the only text on the first screen.
Why: the video can't sell anyone who never starts it. A headline that repeats the ad's promise and tells them what the video shows ("Watch: the 3-step method adults use to play their first song in 30 days") gives them a reason to tap play.
Keep it short enough that the headline and the whole player fit on one phone screen. If the player gets pushed below the fold, you've added a scroll before the first second of the video. I go through how to write and test headline variants in how to test landing page headlines.

4. Make autoplay and unmute obvious
On most VSL pages the video autoplays muted, because browsers block sound until the visitor taps. That means the first thing the visitor sees is a silent person talking.
Why it matters: if they don't realise they need to tap for sound, they scroll past it or leave. A big, clear "Tap to unmute" (or "Click for sound") over the video gets far more people into the actual pitch than the small speaker icon in the corner of the player.
What to do: put the unmute prompt over the middle of the video, make it big enough to hit with a thumb, and when they tap it, restart the video from the beginning with sound. Otherwise they unmute halfway through the hook and miss the line that made them stay.
5. Add captions for muted mobile viewing
Even with a good unmute prompt, many people won't turn the sound on. They're on a bus, in bed next to someone, or at work.
Meta's own advice for video ads says it plainly: “most video ads in mobile feed are viewed without sound”, and its internal tests found that captioned video ads increased view time by an average of 12%. The same page says that in one study, 41% of Facebook video ads were basically meaningless without sound. That's about ads in the feed, not sales pages, but your VSL gets the same visitor, on the same phone, a few seconds later.
What to do: burn the captions into the video (not just the player's CC button, which people don't turn on). Big words, high contrast, two lines at most, placed where the player controls won't cover them. In my paid-ad funnels, 90% or more of the traffic is on phones, so I always watch the VSL on mute on my own phone before it goes live.
6. Test when the button appears
This is one of the most argued-about parts of a VSL, and the honest answer is that you have to test it.
A delayed button (it appears after the offer is revealed) stops people from clicking through to a price before they've heard why it's worth it.
A visible button from the start lets people who are already sold, like returning visitors or your email list, buy without sitting through the whole video again.
Here's a real test from my funnels. On my upsell pages I tested the plain page against the same page with a short video on top. On the first upsell, the video version also hid the rest of the page and the button until a 50-second timer ran out. I ran it from September 25 to October 29, across three upsells, with over $44,000 in revenue going through them:
Upsell 1: 38% converted on the page without the video, 43% with the video. That gap alone added over $2,000.
Upsell 2: 26% vs 29%.
Upsell 3: 20% vs 23%, almost $700 extra.
I ran the same test in another funnel and the video version won there too. One honest caveat: on that first upsell the test changed two things at once, the video and the delayed button, so it tells me the combination won, not which part did more. If you want to know whether the delay is helping your VSL, test the same video with a delayed button against a visible one.
Why Video Upsells Beat Text (Real Funnel Data)

For book-a-call funnels, the same question applies to the "Book your call" button. If calls come in from people who clearly didn't watch (they don't know the price range or who the program is for), a later button may give you fewer but better calls. Judge that on calls booked per visitor and on show-up rate, not on clicks.
7. Pick the video length for the offer, then cut
There's no perfect VSL length. A $7 ebook doesn't need 40 minutes of selling. A $997 course or a $5,000 coaching program may need every minute to handle the objections.
Why it matters: every extra minute is another place for people to leave. Wistia's video length guide found an 11% drop in engagement once videos cross 30 minutes, and recommends keeping videos under that line.
What to do: decide the length from the price and how much explaining the offer needs, then cut anything that doesn't move the viewer closer to buying. Long stories, a second recap and three versions of the same testimonial are usually the first to go. Your retention graph (see number 12) shows you which minutes people skip. I compare the conversion rates people actually see by video and offer type in my video sales letter conversion rate post.
8. Put proof under the video
Many visitors don't watch to the end. They scroll down to see if this is real. What they find under the video decides whether they go back up and keep watching, or leave.
Why: proof under the video catches the skeptics the video lost. It's the same reason proof near the button works on any sales page.
What to put there, in this order:
Results with specifics: screenshots of student wins, short video testimonials, "played my first song in 3 weeks" with a name and photo.
Who you are: one short paragraph and a photo, with real numbers if you have them.
What's inside: a mockup of the course or book, with the price only if the video has already revealed it.
On one of my book funnels, adding a product mockup with a money-back seal and the review count lifted checkout from 44% to 53%. Proof you can see works.
9. Make the player work on a phone
Watch your VSL page on a real phone, on mobile data, not on your laptop with the window made narrow.
Check these:
It starts fast. A video that spins for five seconds on 4G loses people before the hook.
It fits the screen. No black bars eating half the player, no controls covering the captions.
It doesn't jump to full screen and lose the page around it, unless you want that.
The button is reachable with a thumb once it appears, and it doesn't hide under a sticky chat widget.
Why: if 90% of your paid traffic is on phones (it is in my funnels), a player that only works well on desktop is broken for most of your buyers. My landing page CRO checklist covers the rest of the mobile checks.
10. Keep the page around the video simple
A VSL page has one job: get the video watched and the button clicked. Everything else on it competes with that.
What to take off: the top menu, links to your blog, social icons, a second offer. What to keep: the headline, the player, the unmute prompt, the button and the proof under it.
Why: every link is a way out. Someone halfway through your video who taps "About" probably isn't coming back to minute seven. If you're deciding how much should sit under the video, my post on short vs long landing pages goes into when more copy helps.
11. Fix the offer, the price and the order page
A lot of "my VSL isn't converting" problems are actually order page problems. People watch the whole video, click the button, see the order page and leave.
My friend Mike has funnels that have done over $10 million. On a call I posted on my channel, he said his order page after a 30-minute VSL normally converts around 20% to 30%, and he keeps it to no order bump on purpose, because of decision fatigue. I've seen why. When we tested two order bumps plus three payment options on one of my order pages, it went from converting at 40% down to 8%. Too many choices at the moment of paying is expensive. Mike also said they tried a $497 price against $297 and got about the same earnings per click, just fewer buyers, so they stayed at $297.
Here's one from my own funnels that went the other way. On another order page I made the order bump pre-checked, showed the discount on it, and put reviews right under the buy button. Earnings per click went from $3.03 to $4.68, and average order value from $35 to $39.
Why This Tiny Order Page Change Almost Doubled My EPC

And not every "best practice" wins. On a $5 book funnel I added a 30-day money-back guarantee to the order form, sure it would help. It lost in two different funnels. The opt-in step on the order form dropped from 69% to 44%, and checkouts from 47% to 31%.
Why this belongs in a VSL post: the video's job ends at the click, but your money arrives on the order page. Price matters too. In my funnels, a $97 product usually converts at 1% to 3% on a sales page, and above $97 it's around 1%, which is why most people selling higher-priced offers move to a webinar or a call. My anatomy of an order form post breaks down each part of the page.
12. Find exactly where people drop off
Before you rewrite anything, look at where people actually leave. There are two places to look:
Your video host's retention graph. Wistia, Vidalytics and most VSL players show the percent of viewers still watching at each second. Look for the steep drops: the first seconds, the price reveal, the point where the story drags.
Microsoft Clarity on the page. It's free, and its recordings and heatmaps show you what people do around the video: whether they tap the unmute prompt, scroll down to the proof, rage-click a button that isn't there yet, or leave the second the price shows.
Then look at the whole funnel, step by step: ad click to page, page to video play, video to button click, button to order, order to upsell. Mike described the same idea on our call: retention and drop-off at each part of the VSL (the open, the body, the close) tells you which part to rewrite. Fix the biggest drop first, because a test aimed at a part of the video nobody reaches can't move your sales. My follow-the-money sales funnel post shows how to follow revenue through every step.

13. Test 5 to 10 hooks at once, and judge on revenue per visitor
This is where the big wins come from. Not from changing one sentence, but from testing completely different openings.
Mike put it simply on our call: the smaller the change, the more visitors you need to see if it worked. In his words, changing one sentence in a VSL might need something like 50,000 visitors to tell, while changing the whole body or close might need a couple of hundred. Bigger swings need less traffic to read.
So instead of one A/B test of two similar openings, I write 5 to 10 hooks that come at the offer from completely different angles. For a piano course that could be:
"If you quit piano as a kid, this is for you."
"Play a full song in 30 days, 15 minutes a day."
"Why adult beginners learn faster than kids."
"The 3 chords behind most pop songs."
"What my piano teacher never told me."
Then I split traffic between them with a multi-armed bandit. It sends more visitors to the hooks that are earning and fewer to the clear losers, so you don't pay full price for a bad hook for two weeks. The test itself is still an A/B test (or a multivariate one, if you test the hook, the headline and the button together); the bandit is just how the traffic is split. My post on multi-armed bandit vs A/B testing explains how the two fit together.
So when you launch a new VSL, launch four or five very different versions at once instead of polishing one. Different versions are easy to tell apart, so they need much less data, and the bandit moves traffic to the winner while the test runs.
How to Test and Scale Funnels (Book a Call, VSL, Low Ticket)

How to judge the winner:
VSLs that sell on the page: revenue per visitor, counting the order bump and upsells. A hook that gets more clicks but attracts cheaper buyers can lose on revenue. I go through why in revenue per visitor vs conversion rate.
Book-a-call VSLs: calls booked per visitor while the deals are still closing, then revenue per visitor once they do. Keep an eye on show-up rate, so a hook that books more calls from people who never show doesn't win.
How long to run it depends on traffic. With high traffic a week can be enough, two weeks is typical, and with low traffic plan on a month or more. Run full weeks either way, because people buy differently on a Saturday than on a Tuesday.
How Splitlab helps
Splitlab runs A/B, multivariate and multi-armed bandit tests on funnel pages, including VSL pages built in ClickFunnels and GoHighLevel. You can test different hooks, headlines and button timings, and every test has revenue per visitor built in.
Its integrations with ClickFunnels 2.0 and Classic, GoHighLevel, Cal.com, Calendly and stores credit each order, order bump, upsell and booked call to the variant the visitor saw. That's what makes "which hook made the most money" answerable. We'll increase your conversion rate by 30% or you don't pay (terms apply). See the features or book a demo.
FAQ
What is a good conversion rate for a VSL?
It depends on the price and the traffic. In my funnels, a $27 product converts around 5% on a sales page, a $97 product 1% to 3%, and a $997 offer after a webinar 10% to 15%. Paid traffic converts lower than an email list. My video sales letter conversion rate post goes deeper.
Should the buy button be hidden until the end of the VSL?
Test it. A delayed button stops people from seeing the price before they hear the pitch, but it can annoy buyers who are already sold. Run the same video with a delayed button against a visible one and judge it on revenue per visitor.
How long should a VSL be?
As long as the offer needs and no longer. Cheaper offers need less explaining. Wistia found an 11% drop in engagement once videos go past 30 minutes, so if yours is longer, check your retention graph for the minutes people skip.
Do captions really matter on a VSL?
Yes, especially on paid traffic. Most of it is on phones, and Meta says most mobile feed video ads are viewed without sound. Burn the captions into the video so they show even if the viewer never taps unmute.
What should I test first on a VSL?
The hook, the first seconds of the video. That's where the biggest drop usually is, and it's the change most likely to move revenue. Write 5 to 10 different hooks and split traffic between them with a multi-armed bandit.
My final thoughts
Most VSLs I look at aren't bad videos. They lose people in the first seconds, they're hard to watch on a muted phone, and the order page after them throws away the buyers the video worked to get.
If you do one thing this week, open your video's retention graph and install Microsoft Clarity. Find the steepest drop and fix that first. Then write five hooks that come at your offer from completely different angles and let a bandit split find the one that pays.
Keep score in revenue per visitor, or calls booked per visitor, not plays or clicks. If you want help setting that up on your own funnel, book a demo.
Your next winning hook is probably one you haven't written yet.
Built for revenue
Test for revenue, not just clicks.
Run A/B, multivariate, and multi-armed bandit experiments — with real revenue tied to every variation.
We’ll increase your conversion rate by 30% or you don’t pay. Terms apply.
✓
Revenue per variation
✓
A/B & MVT
✓
Multi-armed bandit
Test for revenue, not just clicks.
A/B, multivariate and multi-armed bandit experiments, with real revenue tied to every variation.


