These get treated as two ways of doing the same job. They aren’t. One is a design element inside your frame; the other is a data file about your video. They have different strengths, they fail differently, and on short-form the correct answer for most people is both — which almost nobody does.
The actual difference
| Burned-in (open) | Closed (a caption track) | |
|---|---|---|
| Where it lives | In the pixels of the video | In a separate text track alongside it |
| Can be turned off | No | Yes |
| Looks the same everywhere | Yes — it’s the same file | No — each player styles it its own way |
| Machine-readable | Only via transcription or on-screen text reading | Directly |
| Screen readers / resizable text | No | Yes |
| Word-level styling | Anything you can render | Whatever the player supports — usually very little |
| Fixable after publishing | No — re-export and re-upload | Yes — replace the track |
Why short-form settled on burned-in
Four reasons, and they compound:
- Nobody turns them on. A caption track defaults to off in most contexts. In a feed watched largely on mute, off means the video communicates nothing for the first seconds — which are the seconds that decide everything.
- They’re a design element, not an accessibility afterthought. The whole vocabulary of modern captions — word-timed highlighting, an emphasised hero word, text behind the speaker — is rendering, and a player’s subtitle engine can’t do any of it.
- They travel. One export goes to Reels, TikTok and Shorts and looks identical on all three. A track re-styles itself per player, when it survives the upload at all.
- Placement is yours. A player puts subtitles where it likes, which on a 9:16 phone frame is frequently underneath the interface. Burned-in text can respect the safe band.
What burned-in genuinely costs you
This is where most articles stop, and it’s the half worth reading.
- They can’t be resized. A viewer who needs larger text has no recourse. Your size choice is final for everyone.
- They can’t be turned off. Some people find permanent text distracting, and some footage is ruined by it.
- Screen readers can’t reach them. A caption track is text; pixels aren’t.
- No translation. A track can be swapped for another language. Burned-in text means a separate export per language.
- Typos are permanent. A misspelled name in a track is a one-line fix; in the pixels it’s a re-export, a re-upload, and losing whatever engagement the post had accumulated.
When a track alone is right
- Long-form video. A viewer who has committed twenty minutes can decide whether they want subtitles, and permanent text over a full-screen edit is intrusive.
- Multi-language releases. One video, several tracks.
- Anything embedded on a site you control. Tracks are also text an engine can read — an actual discoverability benefit.
- Content that changes. If the text needs correcting after publication, a track can be corrected.
Doing both without doubling the work
- Transcribe once, with word-level timing.
- Correct the transcript once — names, jargon, numbers. Everything downstream inherits the fix.
- Burn in the styled captions and export the video.
- Where the platform supports it, let it generate its own caption track from the audio, or upload one. Same words, different surface.
The reason this isn’t twice the work: both outputs come from the same corrected transcript. The expensive part — getting the words and their times right — happens once.
Quick answers
What's the difference between open and closed captions?
Open (burned-in) captions are part of the video's pixels — they can't be turned off and they look identical everywhere. Closed captions are a separate text track the player renders, so the viewer can toggle them, restyle them, and a search engine or screen reader can read them.
Which should I use for Reels and TikTok?
Burned-in, for the video itself — it's what everything in short-form is designed around, it survives cross-posting, and it's the only route to word-timed styling. Where the platform also lets you supply or auto-generate a caption track, do both: burned-in for the look, a text track for accessibility and machine reading.
Do burned-in captions hurt accessibility?
They help a lot of people and serve one group less well. Always-on text is exactly what a viewer in a noisy place or with hearing loss needs, but it can't be resized, restyled or read by a screen reader, and it can't be turned off by someone who finds it distracting. That's why a text track alongside is the complete answer rather than a redundant one.
Do captions help discoverability?
A caption track is machine-readable text about your video, which helps. But platforms also transcribe your audio and read text drawn in the frame, so burned-in captions aren't invisible to them either. Saying and showing the same keywords is what makes both signals agree.
One transcript, both outputs
Moonshot times every word, lets you fix the words once, and burns in captions in the style you pick — the same transcript that feeds your hook and your cover.
Open Moonshot free