Best Caption Placement in Video for Retention
Put captions in the lower-middle band of a vertical frame, centered horizontally, with the baseline around 62% of the way down a 1080×1920 clip, not pinned to the bottom and not floating up top. That single band clears every app's buttons and progress bar while staying close to the speaker's face, and in our pipeline it is the position that held viewers longest. Bottom-anchored captions look balanced in your editor and disappear under live UI the moment you post.
This is the placement-only cut of our caption work, vertical position against retention, held apart from caption style and word count on purpose. Those are different decisions. You can pick a perfect karaoke style and the cleanest font on the platform, and still lose viewers because the line sits where the app draws a button on top of it. Where a caption sits is its own lever, and most auto-captioners default it to the worst place: the bottom. This piece maps each platform's UI-obstructed regions, shares a directional finding from our own clip pipeline on how position tracks with completion, names every source and its caveat, and gives you one baseline you can set once.
Methodology, what we measured and what we didn't
This study pairs two kinds of evidence that carry different weight on purpose. The hard numbers, the per-platform UI dead zones and the mute rate that makes captions matter, come from named, public sources you can verify, and they are derived from the rendered app interface rather than an official spec, which is a caveat we state every time. The position-versus-retention finding is a directional cut from QuickReel's own clip pipeline, and it is labeled that way wherever it appears.
For the proprietary cut, we grouped a sample of captioned vertical clips by the vertical position of the caption block, top, dead-center, lower-middle, lower-third, and very-bottom, and compared relative completion within the sample. We are deliberately not publishing a precise headline percentage from that cut. The honest version of any single number here needs to control for hook strength, clip length, speaker framing, and topic, and those confounders are exactly what a separate, controlled experiment is for. So treat the pipeline finding as a direction with a clear mechanism behind it, the mute rate plus the UI map, not a stat to quote out of context. The public safe-zone measurements are where the verifiable rigor lives; the pipeline finding tells you which way they point in practice.
Why does caption position change retention at all?
Because for most viewers the caption is the only version of the audio they get, and a caption they cannot see is a caption that does nothing. Somewhere between 75% and 85% of social video is played on mute, Sharethrough put the mobile figure near 75%, while Digiday found roughly 85% of Facebook video watched silent in 2016 (publisher-reported and directional). If three-quarters or more of your audience never hears a word, the text is doing the work the sound was supposed to, and where that text sits decides whether they ever read it.
Position fails in two ways, and they are different. The first is obstruction: the app stacks its own buttons, caption bar, and progress scrubber on top of your frame, and a line parked underneath any of them gets covered. The second is eye travel, text far from the speaker's face forces the viewer's eye to bounce between the words and the action, which costs attention on a clip that has seconds to earn it. The lower-middle band is the answer to both: it sits above the bottom UI and close enough to a talking head that the eye barely moves. Blitzcut's placement guide makes the same point, putting captions in the "lower-middle third of the usable safe zone" specifically because it clears the bottom dead zone with margin and "reduces eye travel" for talking-head content (Blitzcut).
Where is the retention-safe band, exactly?
There are four UI regions to clear, not two. Most placement advice names only the bottom bar and the right-side button column. The two that quietly cover more captions are the progress scrubber, a thin strip at the very bottom that appears the second a viewer taps the screen, and the top bar carrying the handle, follow button, and on iPhones the Dynamic Island. The retention-safe band is the clear space between them: centered horizontally, in the lower-middle of the frame.
The progress scrubber is the sneaky one. It sits in the bottom sliver of the frame and on TikTok, Reels, and Shorts it appears the moment a viewer scrubs or taps, but it rarely shows in a static editor preview, so captions parked at the very bottom get a bar drawn straight through them in real playback. The good news is that it lives inside the larger bottom dead zone, so clearing that zone clears the scrubber automatically. For the full UI-clearance walkthrough, pixels per platform, the progress-bar trap, and how to set it once, see our companion how-to on caption safe zones for vertical clips. This piece is about the retention angle on top of that map.
What does the data say about caption height and completion?
Lower-middle held viewers best in our pipeline; the very bottom held them worst. Grouping clips by the vertical position of their caption block, relative completion peaked for the lower-middle band, stayed high for the lower-third, dropped for dead-center and top placements, and was lowest for captions pinned to the very bottom, the position auto-captioners pick by default. We are not attaching a precise percentage to those gaps (see the methodology and the limitations), but the direction is consistent and the mechanism is the UI map above plus eye travel: bottom captions get covered, top captions pull the eye off the face.
Read the two ends of that chart together and the story is clean. The default position, very bottom, is the worst, because that is exactly where the caption bar, the scrubber, and the bottom UI live, so a meaningful share of muted viewers literally cannot read it. The top band underperforms for a quieter reason: nothing obscures it, but it drags the eye away from the speaker every time the line changes, and on a talking-head clip that is a steady tax. The lower-middle wins by clearing the UI and staying near the face. Dead-center reads fine on a screen-share or text-on-color clip, but on a face it floats too high and covers the speaker's mouth, which is its own kind of distraction.
One baseline that survives every feed
You do not need a different position per platform. Each app reserves a different slice of the bottom for its UI, but if you set the caption baseline to clear the tightest one, it clears all three. On a 1080×1920 frame, that lands the bottom edge of the caption block around 62% of the way down, roughly 700–730px above the bottom, centered horizontally. Build to Instagram Reels, the most crowded of the three, and TikTok and YouTube Shorts have room to spare.
The per-platform clear zones behind that baseline are measured from the rendered interfaces, then rounded up to a conservative floor, because the live caption bar and a tapped-in progress scrubber routinely push past the static numbers. The underlying figures: Zeely measures TikTok's bottom caption-and-CTA band at ~250–300px and its right column at ~120px, and Kreatli's TikTok guide pegs the text-safe area near 1080×1420 (a ~250px+ bottom clear), so a working floor of ~300–370px at the bottom and ~120–140px on the right is the safe read. Kreatli's Reels guide puts the bottom dead zone at roughly the lower 20–25% (~380–480px) because the caption bar grows when expanded, call it ~320–450px as a floor. Kreatli's Shorts guide does not publish a bottom pixel count (it points to its own checker and recommends placing text centrally or slightly above center); rounded against the other two, ~300–400px is a defensible floor. One pattern holds across all three: the bottom margin is always the largest and the right column always costs more than the left, so a centered, lower-middle caption is the position that is safe everywhere. The honest caveat, stated plainly: no platform publishes an official safe-zone spec, so every pixel above is derived from a live interface and shifts when the app updates. When two guides disagree, take the more generous margin.
| Platform | Keep clear at bottom | Right-side column | Recommended caption baseline |
|---|---|---|---|
| TikTok | ~300–370px | ~120–140px | ~700px up (lower-middle) |
| Instagram Reels | ~320–450px | ~120px | ~760px up (highest, tightest) |
| YouTube Shorts | ~300–400px | ~120–150px | ~680px up (lower-middle) |
A placement check you can run in one pass
Here is the check we use on a clip before it ships. It takes about ten seconds and catches the position failures that quietly cost completion.
- Find the caption block's baseline. It should sit around 62% of the way down the frame, centered. If it is hugging the bottom, it is in the worst position by default.
- Overlay the UI in your head, or in preview. Picture the right-side buttons, the bottom caption bar, and the tap-triggered progress scrubber. If the caption touches any of them, raise it into the lower-middle band.
- Check the eye-travel cost. On a talking head, the caption should sit just below the face, not up near the top. Far text taxes attention every time the line changes.
- Watch it on mute, phone-sized, while tapping. Tapping summons the scrubber. If the caption gets covered or you cannot read it with the sound off, neither can the 75–85% who watch silent.
- Use one baseline across platforms. Set it to clear Reels (the tightest) and you clear TikTok and Shorts for free. Resist a different position per export, it is a maintenance trap with no upside.
That is the actionable core: placement is not a style choice, it is a visibility budget set by the app's UI and the viewer's eye. Where this sits in the larger picture of what makes a clip travel is covered in our analysis of what makes one clip travel across 10,000 clips, and the density-versus-completion cut, a different caption lever entirely, lives in do word-heavy captions help or hurt a clip.
What the data does not tell you (limitations)
Three honest gaps, stated plainly, because the competitor post on this topic usually hands you one clean position and skips all of them.
- The pipeline finding is directional, not a published percentage. Our position-versus-completion cut shows a consistent direction, but completion is confounded by hook strength, clip length, speaker framing, and topic. Until those are controlled in a dedicated experiment, we will not put a precise figure on the gaps between positions, and you should not quote one from us. The verifiable rigor here is the public UI map.
- The safe-zone pixels are derived, not official. No platform publishes a safe-zone spec. The TikTok, Reels, and Shorts numbers above are measured from rendered interfaces by guides like Kreatli and Zeely, and they move with every UI update. The 62% baseline is a conservative floor, not a constant, re-check it when an app redesigns its overlay.
- Lower-middle is the talking-head answer, not the universal one. The eye-travel logic assumes a face on screen. For screen-share clips, text-on-color quote cards, or B-roll where there is no face to anchor to, dead-center or a different layout can read better. The rule is "near the subject, above the UI", and on a talking head that is the lower-middle band.
The messy version is the useful one. Placement is a lever you can set this week; the public UI map tells you which band is safe, and your own completion numbers tell you whether to trust it for your specific clips.
Cite this study
To reference these findings, use: QuickReel, "Where Captions Should Sit for Best Retention" (2026), pairing a directional QuickReel clip-pipeline analysis of caption vertical position against completion with published per-platform UI safe-zone measurements (Kreatli, 2026; Zeely) and watched-on-mute estimates (Sharethrough ~75% mobile; Digiday ~85% of Facebook video, 2016). The summary table below is free to quote with its sources attached. For adjacent cuts from the same pipeline, see how clip duration tracks with views, how long a clip's hook should be, the podcast clipping industry by the numbers, and how the clipping economy actually works.
All pixel ranges are conservative floors derived from rendered interfaces (no platform publishes an official spec), measured on a 1080×1920 frame.
| What it measures | Figure | Source |
|---|---|---|
| Social video watched on mute | ~75–85% (directional) | Sharethrough (~75% mobile, via Digiday); Digiday, 2016 (~85% Facebook) |
| TikTok bottom clear zone | ~300–370px floor | Zeely (~250–300px measured); Kreatli, 2026 (~1080×1420 text-safe) |
| Instagram Reels bottom clear zone | ~320–450px floor | Kreatli, 2026 (lower ~20–25% of frame) |
| YouTube Shorts bottom clear zone | ~300–400px floor | Rounded against Kreatli Shorts/Reels guides (no published px) |
| Recommended caption baseline | ~62% down (lower-middle) | QuickReel; Blitzcut placement guide |
| Caption position vs completion | Lower-middle held longest (directional) | QuickReel clip-pipeline analysis |
FAQ
Where should captions go on a vertical video clip? In the lower-middle band, centered horizontally, with the baseline around 62% of the way down a 1080×1920 frame, roughly 700–730px above the bottom. That clears every app's bottom bar, right-side buttons, and progress scrubber while staying near the speaker's face. It also held viewers longest in our clip-pipeline analysis (directional).
Why not just put captions at the bottom? Because the bottom is where the app draws its own UI, the caption bar, the handle, and the progress scrubber that appears the moment a viewer taps. A line pinned there gets covered in real playback even though it looks fine in your editor, so a chunk of the 75–85% who watch on mute (Sharethrough; Digiday, 2016) never read it.
Should caption placement change per platform? You can use one position for all three. Each app reserves a different bottom margin, Reels is the tightest at roughly 320–450px, TikTok and Shorts a bit less (Kreatli, 2026), so if you set the baseline to clear Reels, it clears TikTok and YouTube Shorts automatically. One baseline near 62% down survives every feed.
Is dead-center better than lower-middle? Usually not for talking-head clips. Dead-center floats above the speaker's face and can cover the mouth, and the eye has farther to travel between the words and the action. Lower-middle clears the UI and sits just below the face. Dead-center can win on screen-share clips or text-on-color cards where there is no face to anchor to.
Does placement matter more than caption style? They are separate levers, and you need both right. Style and word count decide whether a caption is readable; placement decides whether it is visible at all. A perfectly styled line that sits under a button is invisible to muted viewers, see the density cut in do word-heavy captions help or hurt a clip for the style side of the same study.