Make it work with the sound off: four rules for captions

Not everyone watching your video turns the sound on. Between meetings, on a bus, next to someone asleep, people scroll with their eyes and decide within a few seconds whether to listen. If they cannot read what you are saying in those seconds, they move on.

1. Captions should follow the speech word by word

Long lines dropping onto the screen a sentence at a time do not get read: the speech moves on before the eye reaches the end of the line. The right way is to split the speech into two- or three-word chunks and show each one the moment it is said. The viewer does not feel they are reading; they follow along as if listening.

Good captions are not read, they are followed.

2. The bottom of the screen is not yours

Platforms place their own text, the post caption and their buttons along the bottom of the screen. Captions set at the very bottom end up underneath that layer. Lifting them away from the bottom edge, a little above the lower third, keeps them clear of both the interface and your face.

3. A colour nobody can read does not exist

White text on a light background, thin type over moving footage: it is on screen, but not for the viewer. A bold typeface, a dark outline or a semi-transparent backing keeps it legible whatever is behind it. Colouring the one word you want to stress draws attention; colouring every word draws it to none.

4. In the viewer's language

If you speak Turkish but also want to reach an audience abroad, you do not need to film again. Captions in another language open the same recording to a second audience. The one thing to watch is that the translation runs in step with the speech: if a translated line appears well before or long after the spoken sentence, the viewer loses the thread.

• Keep chunks short: two or three words.
• Lift the captions away from the bottom edge.
• Pick a contrast that reads on any background.
• One emphasis, one colour.
• Watch the video once with the sound off before you publish.