Subtitles in short videos — how to do them right

Look around on a train or in a queue — almost everyone scrolling short videos has the sound off. People watch on the go, at work, next to sleeping children. If your video has no subtitles, for a large share of viewers it is just a person silently moving their lips. They will swipe in a second.

Subtitles stopped being an accessibility checkbox long ago. They are part of the format and directly affect retention and views. Here is how to make them help rather than get in the way.

Why subtitles matter

Watching on mute. The obvious reason: viewers understand the content without turning on sound.

Holding attention. Text on screen holds the eye. When words appear in rhythm with speech, it is easier to follow and harder to drift away.

Comprehension. Terms, names and numbers are easier to take in visually, especially when the speaker talks fast or has an accent.

Accessibility. For people with hearing loss, subtitles are the only way to experience the video.

Discovery. Platforms analyse text and speech in videos. Keywords that are both heard and seen help the video find its audience.

Burned-in vs platform captions

Some platforms generate captions automatically. But they often need to be switched on manually, can’t be styled, vary in recognition quality and look different on every platform.

Burned-in subtitles are visible to everyone, look the same everywhere and match your brand. For short video they are the standard.

Size and font

Large. Text must read on a phone at arm’s length. On a 1080×1920 frame that is roughly 60–80 pixels of lowercase letter height.
Bold weight. Thin fonts get lost on busy backgrounds.
A simple sans-serif. Decorative fonts look good in a title but are hard to read in motion.
Outline or shadow. White text without an outline disappears on light backgrounds; a thin dark outline or soft shadow fixes this for any frame.

Position and safe zones

Every platform covers part of the frame with its interface:

Bottom — caption, author name, music, buttons: 15–25% of the frame height depending on the platform.
Right — the column of likes, comments and shares.
Top — menus and tabs in some apps.

Placing subtitles at the very bottom, as in film, puts them under the caption. The safe position is in the lower third but above the bottom interface zone, or closer to the centre, and left of the right-hand button column. If there is a face in the frame, subtitles shouldn’t cover it.

Line breaks

The key rule: little text on screen at a time.

1–2 lines. Three lines read like a paragraph.
5–7 words maximum, 2–4 for fast-paced videos.
Break by meaning. A misplaced break changes the emphasis.
In sync with speech. Text appears when the phrase is spoken and doesn’t linger.

Highlighting the spoken word

Highlighting the word being spoken works for two reasons: the eye follows the highlight, and its rhythm sets the pace.

One accent colour — usually your brand colour or a yellow that shows on any background.
No overkill. If every word jumps, scales and changes colour, it is tiring.
Precise sync. A highlight that runs ahead of or behind speech is worse than none. It needs recognition with exact timing for each word.

Style and brand

Subtitles are a visible part of your brand. When the style is consistent, viewers recognise you in the feed before they see the logo. Define font and weight, text and highlight colours, outline or background box, and position once — then use them in every video.

Proofreading

Automatic speech recognition is good but not perfect. It most often trips on brand, drug and technology names, professional terms, personal names, numbers and mumbled words. Read the subtitles in full before publishing. A wrong term in a medical or financial video undermines trust in the expert more than bad lighting does.

Common mistakes

Mistake	Why it hurts	Fix
Small font	Unreadable on a phone	60–80 px lowercase height
Subtitles under the interface	Hidden	Move above the bottom 20%
Whole sentence on screen	No time to read	1–2 lines, up to 7 words
White text without outline	Vanishes on light backgrounds	Outline or shadow
Out of sync with speech	Distracting	Word-level timing
Typos in terms	Erodes trust	Proofread before publishing
Five colours and animations	Exhausting	One accent colour

Subtitles and warnings

If you add warnings or an ad label to a video, make sure they don’t clash with subtitles — no overlap, both legible. A persistent warning line usually sits at the edge of the safe zone with subtitles above it.

Subtitles in two languages

If you publish for Russian- and English-speaking audiences, don’t put both languages in one frame — it overloads the screen. Make two versions instead: one with subtitles in the original language and one translated. Translate adaptively rather than literally: conversational speech shortened for the screen often needs restructuring in another language, and line breaks and timing must be recalculated for the new text.

A one-minute pre-publish check

Watch on a phone, not a monitor — that’s how viewers will see it.
Toggle the sound. On mute the video must still make sense.
Check the first 3 seconds: text appears immediately, not a second after speech starts.
Find every term and number and compare with what was said.
Check overlaps: subtitles don’t cover the face, logo, warnings or key elements.
Check light and dark scenes: text reads in both.

What it costs in time

Making subtitles by hand for a 40-second video — transcribing, splitting, syncing, styling — takes 15–30 minutes. At 5–7 videos a week that is hours. Automatic recognition with word timing and a brand template reduce the job to proofreading, about a minute per video. Proofreading is the one step not worth automating: it protects your credibility.

FAQ

Should subtitles be a word-for-word transcript? Mostly yes, but remove filler words and false starts. Viewers should read what was meant, not every “um”.

Does the highlight colour have to be yellow? No. Use your brand colour if it contrasts well with both light and dark scenes; otherwise yellow or white with a coloured box are safe choices.

What about emojis in subtitles? Use them sparingly, if at all. One emoji can add emphasis; several per line distract and make the video look less professional, especially for experts.

Do subtitles help search? Platforms analyse speech and on-screen text, so clear subtitles with the terms your audience searches for can help a video reach the right viewers.

Subtitles and accessibility

For deaf and hard-of-hearing viewers, subtitles are the only way to follow a video. To make them genuinely useful, add meaningful sounds in square brackets — [laughter], [phone ringing] — and never rely on colour alone to convey meaning: text and background must contrast enough even without the highlight. The same rules make subtitles easier for everyone else, too.

Summary

Subtitles in short video are not optional. Large bold type with an outline, one or two lines, meaningful breaks, a position that respects safe zones, synchronised word highlighting, a consistent style across all videos and a proofread before publishing. Few things raise retention, widen your audience and make your brand recognisable at the same time — this is one of them.