Most videos on social media start playing with the sound off. A viewer who can’t follow the words won’t watch for long, no matter how strong the footage is. Captions fix that, and they do a lot more besides.
Captions vs. Subtitles: The Distinction Worth Knowing
People use these words interchangeably, but they’re not the same thing.
Subtitles assume the viewer can hear the audio but doesn’t understand the language. They translate or transcribe dialogue for foreign-language audiences.
Captions assume the viewer cannot hear the audio at all. They include dialogue plus sound effects, music cues, and speaker identification. Captions are the correct term for accessibility-first text overlays, and they’re what most platform auto-captioning tools produce by default.
In practice, if you’re publishing on social media or a website, you’re adding captions. The distinction matters most when you’re distributing content internationally and need both.
Open vs. Closed: Picking the Right Format
Captions come in two forms, and the choice affects every downstream decision.
Closed captions are a separate file (SRT, VTT, SCC) that viewers toggle on or off. YouTube, Vimeo, and broadcast standards use this model. They keep the video file clean, allow the viewer to customise display, and are the right format for long-form content on a web player.
Open captions (also called burnt-in captions) are rendered directly into the video frame. Viewers cannot turn them off. On Instagram, LinkedIn, and TikTok, where native caption support varies and where captions may not carry when a post is downloaded and reshared, open captions are usually the more reliable choice.
The trade-off is control. Burnt-in captions fix the style permanently; closed captions let viewers adjust size and colour. For accessibility compliance (WCAG, ADA, Section 508), closed captions on a web player that allows customisation are the gold standard. For a social reel you want watched at full retention regardless of sound, open captions win.
Timing Is the Work Nobody Sees
Captions that are even a fraction of a second out of sync with the audio feel broken. The viewer’s eye is trained to match mouth movement to text, and any drift makes the video harder to follow.
Auto-generated captions from tools like Adobe Premiere, DaVinci Resolve, CapCut, or Whisper-based services get timing close. Close isn’t good enough. The most common timing errors to fix manually:
- Caption appears after the speaker has already started the word
- Two speakers’ lines overlap in a single block
- A long block covers multiple sentences spoken at different speeds
- Captions linger on screen after the line ends, pushing the next block late
The rule for block duration: each block should sit on screen for at least one second and no longer than about three. If a speaker talks quickly, break the line into two blocks rather than extending one long block.
Timing is the invisible work of captioning. When it’s right, nobody notices. When it’s off by a beat, it’s all viewers see.
Style Choices That Change Readability
Auto-captions default to all-uppercase or mixed-case text on a semi-transparent bar. That works as a fallback, but for branded content it often isn’t the right call.
A few principles that hold regardless of style:
Contrast is non-negotiable. White or yellow text with a dark shadow or semi-transparent background works. Light text on a light background does not, even momentarily. If your video cuts quickly across varied backgrounds, a solid caption bar is safer than a drop shadow alone.
Design for mobile first. Most video is watched on a phone held vertically. Caption text that reads clearly on a desktop monitor is often too small at 9:16 on a 6-inch screen. Check your caption renders at mobile size before exporting.
Don’t style captions like subtitles. Standard subtitle styling (small, centred at the bottom, no background) was designed for cinematic viewing distances. At social media viewing distances it reads small and disappears against bright lower-thirds.
Keep line length short. Aim for no more than two lines, and no more than about 37 characters per line. Dense text blocks are harder to read at pace.
What Auto-Captions Get Wrong
Auto-captioning tools have genuinely improved. Whisper in particular handles accents, domain vocabulary, and fast speech far better than tools from three years ago. But they still fail in predictable ways.
They struggle with proper nouns, technical terms, brand names, and overlapping speech. They occasionally produce words that sound phonetically close but mean something entirely different. And they have no understanding of spoken emphasis. A sentence where stress changes the meaning entirely (“I didn’t say he stole it”) is transcribed identically to its unstressed counterpart.
The workflow that holds up: use auto-captions for the first pass, then do a line-by-line edit. This takes a fraction of the time writing from scratch would, but produces captions that are actually accurate. Don’t skip this step for content that represents the brand.
For scripted videos, there’s a shortcut. If the video was shot to a tight script, the script becomes the caption source with only timing work needed. This is one reason writing a tight video script pays dividends well beyond production day.
Captions and Search
Search engines can’t watch a video. They read the text around it, the transcript, and the closed caption file if one is present. Providing a timestamped caption file (not just a raw transcript) makes it easier for Google to index specific moments within a video and surface those moments in search results.
For embedded video on a website, a readable transcript below the player also helps. This matters more for educational and how-to content than for brand films, but it’s worth building as a habit. The broader picture of how video fits into content strategy is covered in video content for brands.
How Strynal Approaches Captions
At Strynal, captions are planned at the script stage, not added as a last step in post-production. Knowing a video will carry open captions changes how titles are framed, how lower-thirds are positioned, and how quickly a speaker can deliver a line without the text becoming unreadable.
For social content, we render open captions using brand typography to match the visual identity, rather than defaulting to platform styling. For web video, we supply timestamped VTT files alongside every deliverable as standard.
This is part of how we treat motion and text as connected systems rather than separate outputs. The thinking behind that approach is laid out in purposeful motion design. And if you want to see how captioning fits into a full video production and distribution workflow, our motion and content work covers the whole picture.
If your current captions aren’t doing the job, the problem is usually one of the three things here: timing, style, or accuracy. All three are fixable once you know where to look.