Landing Page Videos: Above the Fold or Not at All
When landing page videos lift conversion and when they tank it: placement rules, autoplay and speed traps, and which video type fits which page.
The most common landing page video in the wild is a beautiful 90-second brand film, buried below the fold, that 4% of visitors press play on and 1% finish. It cost more than the page it sits on, it added 2.8 seconds to load time, and in the A/B test nobody ran, it's losing to the version without it. Video on landing pages is one of the highest-variance elements in CRO: done right it's worth a 10–30% conversion lift on complex offers; done as decoration, it's a page-speed tax with a play button.
The variance has a pattern, though, and it's mostly about three decisions: whether the offer needs video at all, where the video sits, and whether it plays without being asked. Get those three right and the production part — which AI has made almost trivially cheap — takes care of itself.
First decision: does this page need video at all?
Video earns its place on a landing page when the offer has a comprehension or trust gap that text and images can't close efficiently:
- New-category or mechanism products ("how does this actually work?") — video compresses ten paragraphs of explanation into fifteen seconds of demonstration.
- High-consideration purchases — courses, SaaS, services, anything over ~$100 — where a face and a voice do trust-building that testimonial text can't.
- Products whose value is motion — apps, tools, anything with a before/after.
Video hurts when the offer is already instant: a discounted product the visitor arrived pre-sold on, a simple lead magnet, a one-field signup. There, video adds load time and a decision ("do I watch this?") between the visitor and a button they were already going to press. The fastest CRO win I know is deleting the vanity video from a simple offer page.
Placement: the above-the-fold rule
If the video carries the pitch, it goes above the fold next to (or under) the headline — or it shouldn't exist. The logic is brutal: scroll depth decays fast, and a persuasion asset that 60% of visitors never reach isn't a persuasion asset, it's set dressing. Hence the title of this post. The exceptions are real but specific:
| Video type | Placement | Job |
|---|---|---|
| VSL (video sales letter) | Above fold, IS the page | Carries the entire pitch; page is video + CTA + FAQ |
| Product demo | Above fold, beside headline | Closes the "what is it?" gap in seconds |
| Explainer | Above fold or first scroll | Mechanism-of-action for new categories |
| Testimonial montage | Mid-page, at the objection point | Proof exactly where doubt peaks |
| Founder story | Lower page | Converts the almost-convinced; skippable by design |
The mid-page testimonial slot is the legitimate below-fold placement: it sits where a reader's skepticism naturally peaks (after the claims, before the price), so the people who reach it are exactly the people who need it. Everything else fights for the top or gets cut.
Autoplay, sound, and length: the behavior traps
- Autoplay muted, always, for demo-style video. A static play button gets 5–15% engagement; a muted autoplaying demo gets watched by nearly everyone whose eyes cross it. Design for silent comprehension: captions on, product action visible without narration. If the video only works with sound, it's a VSL, not a demo — treat it accordingly.
- Click-to-play with sound for VSLs. A VSL viewer is opting into a pitch; that self-selection is the point. Autoplay a 12-minute VSL and you'll torch trust and bandwidth simultaneously.
- Length follows placement. Above-the-fold demos: 15–45 seconds. Explainers: 60–90. VSLs: as long as they keep converting — 4 to 20 minutes is normal and only testing settles it. If you're building one, the structure is covered step-by-step in how to make an AI VSL.
- Never loop testimonials or founder videos. Looping reads fine on a 6-second demo; on a human speaking it reads glitchy.
The page-speed tax, and how to not pay it
Video's silent conversion killer is load time — every 1s of added LCP costs meaningful conversion, and a carelessly embedded video adds 1–3s. The checklist:
- Never
<video>-tag a full-res file from your own origin. Serve through a CDN/streaming host with adaptive bitrate. - Poster frame first. The page paints a lightweight image immediately; the video loads behind or on interaction. Your poster frame is also a conversion asset — treat it like a thumbnail, not a random frame.
- Lazy-load anything below the fold. Mid-page testimonial videos should cost zero bytes until scrolled near.
- Third-party embeds (YouTube et al.) are heavy — a facade/click-to-load pattern saves ~500KB+ of script until the user asks for it. Also: a YouTube embed on a landing page offers your visitor recommended videos as an exit. Self-host or use a clean player for money pages.
Producing the video is now the easy part
The economics that used to justify skipping landing page video are gone. Current production reality:
- Product demos: generate scene-by-scene from product reference images through the AI video generator — a 30-second demo assembled in an afternoon. The full walkthrough is in how to make a 60-second AI product demo.
- Founder/presenter videos without a founder on camera: a digital-twin avatar delivers the script to camera — HeyGen Avatar V5 is at the point where a scripted landing-page welcome reads natural, and it regenerates in minutes when the offer changes (the real killer feature: no reshoot when pricing updates).
- Testimonial montages: review text over b-roll with voiceover, or UGC-style readings — an hour of batch work.
Cheap production changes the testing posture most of all: the correct experiment was always "page with video vs page without," but nobody ran it when the video cost $5,000. At $20, you test it on every significant page — and iterate the video like ad creative, because the version-two demo with a tighter first five seconds routinely beats version one.
Measure like it's an ad, not furniture
Minimum instrumentation: play rate (or view rate for autoplay), 25/50/75/100% watch-through, and conversion rate split by watchers vs non-watchers. Two readings matter most. If watchers convert far better than non-watchers but play rate is low, the video works and the placement or thumbnail is the problem — move it up, fix the poster. If play rate is high but watchers convert no better, the video itself is furniture — rewrite it or cut it. Most teams collect neither number and just debate opinions.
FAQ
Do videos increase landing page conversion rates?
They can, substantially — demo and VSL-style videos on complex or high-consideration offers commonly lift conversion 10–30%. But video hurts simple, pre-sold offers by adding load time and decision friction. The honest answer is per-page: test with video vs without, which AI production now makes cheap enough to do routinely.
Should landing page videos autoplay?
Demo-style videos: yes, muted with captions, designed for silent comprehension. VSLs and founder videos: no — click-to-play with sound, because opting in is part of the persuasion. Never autoplay with sound on; browsers block it and users hate it.
How long should a landing page video be?
Above-the-fold demos: 15–45 seconds. Explainers: 60–90 seconds. VSLs run 4–20 minutes and are governed entirely by testing. If the core value isn't communicated in the first ten seconds, the video's length is irrelevant because most viewers won't reach the rest.
Does video slow down my landing page?
Only if embedded carelessly. Serve via CDN with adaptive bitrate, paint a poster frame first, lazy-load below-fold videos, and use a click-to-load facade for third-party embeds. Done right, video adds near-zero LCP; done wrong it adds 1–3 seconds, which quietly erases the conversion lift it was supposed to provide.
What's better on a landing page: a demo video or a founder video?
Different jobs. The demo closes the "what is it?" gap and belongs above the fold on product pages. The founder video builds trust and belongs where skepticism peaks — mid or lower page — or carries service-style offers entirely. High-consideration pages often convert best with both, in those positions.
If a page matters enough to drive paid traffic to, it matters enough to test with video. Build the demo in an afternoon with the AI video generator, put it above the fold, and let the split test settle it — free credits daily.