By Zahid Nawaz, Founder of Motion Graphix ·
Quick answer: Most business video is watched with the sound off, around 80 percent on LinkedIn. So design the video to make sense silently first: put the key point in on-screen text, burn captions into social cuts, keep one idea per scene, and let the animation show the action so the voice-over is a bonus, not a requirement. Then add sound for the people who turn it on.
How many people really watch on mute?
A lot. Around 80 percent of LinkedIn videos play without sound (source), and 87 percent of Americans say they use captions at least some of the time.
Feeds autoplay muted by default, and people watch in offices, on trains and in meetings. If your video only works with the voice-over, most viewers never get the message.
What does “silent-first” mean in practice?
- The headline is on screen. The one sentence you want remembered appears as text, big, in the first 3 seconds.
- Every scene has a label. Three to six words that say what is happening: “Invoices matched automatically”.
- The action tells the story. Show the click, the result, the number. Do not rely on the narrator to explain it.
- Captions on every social cut. Burned in, so they show on every platform without settings.
- Sound as a layer. Music and voice make it richer for the 20 percent who listen.
Burned-in captions or a caption file?
Use both, for different places:
- Burned-in (open) captions for LinkedIn, Instagram, TikTok and ads. They always show and you control the style.
- Caption files (SRT) for YouTube and your website. Viewers can turn them off, search engines can read them, and they help accessibility rules.
Ask your studio for both. It is a small job when done with the edit.
Does silent-first hurt the video for people with sound on?
No. Research on short-form video found captions lifted recall by about 58 percent even for people who listened. Text plus voice beats voice alone. The only rule is not to write a different sentence on screen from the one being spoken; keep them matched or shorter.
How does this change the script?
Write the on-screen text first, as a list of six to eight short lines. If those lines tell the story on their own, you have a silent-first video. Then write the voice-over to add warmth and detail. Our script guide shows the four-beat structure this fits into.
Caption style that reads at feed speed
Captions that work on mute follow a few mechanical rules. Two lines maximum, no more than about 32 characters a line, so a phrase is readable in one glance.
Sentence case, not all caps, because capitals slow reading by roughly a fifth. A high-contrast block or a soft shadow behind the text so it survives a busy frame, and a position in the lower third that does not collide with platform UI; on vertical video that means keeping captions above the bottom 20 percent where the caption bar and buttons sit.
Keyword highlighting (one word in the brand colour per line) helps on TikTok and Reels and looks dated on LinkedIn, so match the platform. Time captions to the voice, not to the scene cut, and never let a caption outlast the shot it belongs to.
Designing the visuals to carry the message
Captions are the fallback; the frame should communicate without them. Every scene needs one legible on-screen element that says what the voice says: a UI label, a number that changes, a short title card.
Test this by watching the cut with the sound off and no captions; if you cannot follow it, the problem is the storyboard, not the subtitles. In our production process this check happens at the storyboard stage, before animation, so fixing it costs nothing. The storyboard guide shows what we look for.
Platform specifics in 2026
LinkedIn autoplays muted in the feed and most viewers never unmute, so burned-in captions are essential and a hook caption in the first frame is what stops the scroll. Meta (Facebook and Instagram) supports caption files but renders them inconsistently across placements; burned-in is safer for ads.
YouTube supports uploaded caption files well and indexes them, so upload an SRT there in addition to any burned-in version. TikTok auto-generates captions that are often wrong on product names; upload your own or burn them in.
For a website hero video that autoplays muted, a “click for sound” affordance plus on-screen text is the pattern that converts, which is how the hero reel on our homepage is built.
Quick questions
What size should on-screen text be?
At least 40px on a 1080p frame for phones. Bigger for the headline.
Do captions count for accessibility?
Yes, captions are the first requirement in WCAG guidelines. SRT files plus a transcript cover most cases.
Should the founder video have captions too?
Definitely. Talking-head clips are the ones most often watched muted.
Every video we make is designed silent-first and ships with burned-in captions and SRT files. Get pricing.
Do burned-in captions hurt accessibility?
They help sighted viewers on mute but cannot be resized or read by screen readers. Where the platform allows, provide both: burned-in for the feed and a caption file for assistive tech. On your own site, a visible transcript under the player covers the gap.
What caption font should I use?
A clean sans-serif at a size that gives roughly 30 characters a line on a phone screen. Your brand font is fine if it is legible at small sizes; decorative fonts are not.
About this guide
Who publishes this
- Motion Graphix, a London-registered studio founded in 2022 by Zahid Nawaz that makes explainer and product demo videos for software companies.
- The guidance comes from producing 85+ software videos; prices quoted for Motion Graphix come from its public pricing page.
- Last reviewed 21 November 2026.
- Corrections are welcome at the contact page.
Sources and further reading
- Wistia: State of Video report, benchmark data on video use and engagement.
- Wyzowl: Video Marketing Statistics, the annual survey on video adoption and results.