A viewer gives an overlay about one second. If the words are small, low-contrast or covered by the interface, they may as well not be there. The fix is not always a bigger font; often it is fewer words and a cleaner background.
Good overlay design is mostly subtraction. One idea per frame, one clear hierarchy, and enough contrast to survive a compressed video.
Decide what the text is for
Overlay text usually does one of four jobs: hook the viewer, add context to the visual, label a step, or deliver a quote. Each job needs a different treatment. A hook can be large and brief. A step label can be small and consistent. A quote needs enough size to be read as a sentence.
If a single frame is trying to do all four, the viewer will read none of them.
Build hierarchy in one frame
Pick one primary line and, at most, one supporting line. The primary line should be readable at thumbnail size. The supporting line can be smaller, but still large enough to read without pausing.
Use size and weight for hierarchy before color. A larger, heavier line reads as primary even in a single color. Multiple accent colors on one frame usually create noise rather than emphasis.
Contrast and safe zones
Text needs a clear difference from what is behind it. A solid or semi-opaque background behind the words is more reliable than a drop shadow, especially over moving footage. If you avoid a background, test the frame at the busiest moment, not the calmest.
Keep text out of the areas where platform controls sit: the bottom caption zone, the right-side action column and the top status area. The central band is usually safest. When in doubt, move the text up and keep it away from the edges.
Quick overlay checks
Thumbnail test: shrink the frame to the size of a feed preview. Is the main line still readable?
Mute test: turn the sound off. Does the text still make sense?
Pause test: pause on a frame where the text appears. Is any word hidden by the interface?
Type choices that help
A clean sans-serif with generous spacing is the safest default. Avoid script fonts for anything that needs to be read quickly, and avoid very thin weights on video; compression makes them disappear.
Left-aligned text is easier to read than centered text when there are several lines. Centered text works for a single short hook. Keep line length short: three to five words per line is usually enough for a vertical video.
Do not use all caps for long sentences. It slows reading and removes the shape cues that help people scan.
Timing and reveal
Text should appear when it is needed and leave when it is not. A common mistake is keeping a full paragraph on screen for the entire video, which forces the viewer to read instead of watch.
If you reveal text line by line, keep the rhythm consistent. A new line every beat or every sentence is easier to follow than random timing. For captions, sync to speech rather than to the edit.
A short checklist before export
- One primary idea per frame.
- Readable at thumbnail size.
- Contrast survives the busiest background.
- Clear of platform controls and edges.
- No more than two type sizes per frame.
- Text on screen only as long as it is needed.
If a frame fails two or more of these, simplify it. Removing a line is almost always better than shrinking it.