The Video Caption and Subtitle Style Cheat Sheet

32 characters is the safe zone for vertical video at standard caption size. Beyond that, text starts wrapping unpredictably on smaller phones. Count spaces as characters. When in doubt, break earlier.

Two lines maximum per caption frame. Three or more lines push text into the speaker's face and force the viewer to choose between reading and watching. Two lines is the ceiling, one is usually better.

Captions should land where the speaker naturally breathes. Splitting a phrase mid-thought creates a jarring gap between audio and text. Use commas and sentence ends as your break points.

Fast speakers need shorter line segments so text does not linger after the word has passed. Slow, deliberate speakers can hold a full short sentence. When the caption outruns or lags the voice, viewers notice and drop off.

Bold sans-serif fonts (think heavy weights of Inter, Montserrat, or the platform-native bold option) read clearly at small sizes and over busy backgrounds. Thin fonts, script fonts, and serif fonts are all harder to read in motion.

On a standard 1080×1920 vertical frame, that is roughly 134px minimum for the cap height. Anything smaller becomes unreadable on a phone held at arm's length. Go bigger rather than smaller when in doubt.

ALL CAPS works for one or two emphasis words. For full sentences it slows reading speed by around 10%, which matters when captions change every 2-3 seconds. Use sentence case or Title Case for regular caption lines.

Switching fonts mid-video (even between two clean fonts) signals disorganisation to the viewer's brain. Pick one font, use it throughout, and vary weight or colour for emphasis instead.

A semi-transparent fill, a solid word-by-word highlight, or a subtle drop shadow will make captions readable over any footage. Relying on white text over a light background or dark text over a dark background will lose a significant chunk of viewers.

When you are unsure what will be behind your text, white with a 2-4px black outline reads on almost any background. This is the approach used in broadcast subtitles for decades for exactly this reason.

Most editing is done on a desktop monitor in a dim room. Your viewers are scrolling on a phone in daylight. A caption that looks clear on your screen can become near invisible on a phone with the brightness turned up. Test there before publishing.

Caption lag is one of the most common retention killers in auto-captioned video. If the word has been spoken and the caption has not appeared yet, the viewer is already processing the next word with no text support. Review auto-generated captions at 0.5x speed to catch lag.

A caption that lingers after the next sentence has started confuses the viewer's reading flow. Each caption frame should disappear as the next word begins. This is especially critical in fast-paced educational or hook-driven content.

A very short opacity fade (2-3 frames at 30fps, so under 0.1 seconds) softens the text's entry and feels more professional than a hard cut-in. Longer fades, however, cause the viewer to miss the start of words. Keep it very fast.

Pick one brand-adjacent colour for the word currently being spoken. Everything else stays white or light grey. More than two colours in a word-by-word highlight turns the screen into noise. One highlight colour, consistently applied, looks intentional.

A 10-15% size increase on the active word creates a subtle pulse effect that guides the eye without being distracting. This is the technique used in most high-retention Reels you will see from polished creators. It does not need to be dramatic to work.

Word-by-word highlighting is energy-intensive for the viewer's eye. It suits punchy scripts with short, impactful sentences. For talking-head interview-style content or long explanations, a standard two-line caption block is less fatiguing and often gets better retention.

On Instagram Reels and TikTok, the bottom 15% of the frame is covered by the username, caption, audio name, and engagement buttons. Any subtitle in that zone will be hidden. On a 1920px-tall frame, that means keeping text above 288px from the bottom.

The top 10% is where platform UI overlays appear on some devices and where the viewer's eye is weakest for reading during fast-paced video. Ideal caption placement is between 70% and 85% of the way down the frame, centred horizontally.

Centre-aligned captions suit all frame types and all viewer orientations. Left-aligned captions work well on YouTube horizontal videos but look off-balance on vertical. Centre is the safe default unless your brand style explicitly calls for left alignment.

Safe zones shift between platforms. Instagram Reels, TikTok, YouTube Shorts, and LinkedIn all have slightly different UI overlay positions. If you cross-post the same file, verify that captions are visible on each platform before the video goes live.

Emphasis loses its power when it is everywhere. A colour change, bold switch, or size increase should appear roughly once every 10-15 seconds. That frequency keeps the viewer's eye refreshed without making the screen feel cluttered.

The word worth emphasising is the one a skimming viewer absolutely needs to catch. In a sentence like 'most businesses are wasting 40% of their ad budget,' the emphasis word is '40%'. Not 'most', not 'businesses', not 'wasting'.

Emphasising 'and', 'but', 'so', or 'also' confuses the viewer about what matters. Emphasis signals importance. Apply it only to nouns, numbers, specific actions, or named outcomes.

The first three seconds determine whether someone scrolls on. A single kinetic text entrance on your opening hook word (a pop, a scale-in, or a colour flash) catches the eye in the feed before the viewer has decided to stop scrolling. Use it once, on the hook, not repeatedly throughout.

Platform auto-captions (Instagram, TikTok, YouTube) have an error rate of roughly 10-15% on standard speech and higher on British accents, proper nouns, and technical terms. A five-minute review pass before publishing catches the errors that would otherwise sit on a live video for months.

If your speaker is centred in a vertical frame, placing captions at the vertical midpoint will cover their mouth and eyes. Move captions lower (or use a lower-third layout) so the speaker's face is always fully visible. Face visibility is directly tied to emotional connection and watch time.

Captioning 'um', 'uh', 'like', or long pauses makes the speaker look unprepared in writing even when it sounds natural in speech. Edit filler sounds out of captions entirely. If the audio has significant filler, edit it at the audio stage too.

Standard caption convention: write 'five steps' not '5 steps', and '12 ways' not 'twelve ways'. Numerals read faster on screen and stand out visually for emphasis. Keeping this rule consistent across a video makes captions feel polished and deliberate.