Retention Editing: Text Overlays Must Match the Voice

Text overlays timed wrong pull the eye off the story. Timed right, they add a second channel the viewer can read while the voice keeps moving. The difference is almost always a half-second of lead, and most editors leave it out.

On-screen type is not a caption. It is a retention device. When a key phrase appears as the narrator says it, the viewer gets the idea twice, once through the ear and once through the eye, and the point sticks harder.

That double exposure is why overlays work on numbers and names. The ear catches the sound, the eye catches the spelling, and the fact survives past the sentence that carried it.

Lead the voice, do not chase it

Type should land a fraction before the word it names. If it arrives after, the viewer has already moved on and the overlay feels late. If it arrives too early, it spoils the line. A quarter of a beat of lead is the sweet spot.

The same rule as b-roll applies here: audio carries the idea, the visual confirms. Type that confirms is satisfying. Type that explains is work, and work is what makes people leave.

Trim the exit too. Keep the overlay on screen for a beat after the word is spoken, then take it away. A card that lingers while the next sentence begins keeps the eye on the past.

An overlay should feel like emphasis, not like a subtitle. If the viewer could read it later, it is in the wrong place.

Restraint and legibility

The biggest overlay mistake is too many. If every sentence gets a card, the viewer stops reading any of them. Reserve on-screen text for the numbers, the names, and the one line you want quoted. Everything else stays in the voice.

And make it readable. A chart that takes more than three seconds to parse will not be parsed. Nobody pauses your video to study a graph. Simplify until it reads at a glance, or replace it with a single big number.

Match the animation to the sentence. A calm line gets a soft fade. A sharp line gets a fast cut. When the motion of the type fights the tone of the voice, the viewer feels the clash without knowing why.

Key takeaways

• Land type a quarter-beat before the word it names.
• Save overlays for numbers, names, and the line worth quoting.
• Keep every graphic readable in under three seconds.
• Match animation speed to the energy of the sentence.
• Never repeat in text what the voice already makes obvious.

Testing the timing

Scrub through the video with only the overlays visible. If they read as a coherent summary, your timing is working. If they read as disconnected flashes, the pacing is off and the viewer is getting pulled between two tracks.

Then watch with the sound off. The overlays should still advance the argument on their own. If they only make sense with the narration, you have written subtitles, not overlays, and subtitles do not hold attention the way a well-timed phrase does.

When the type and the voice agree, the viewer feels informed rather than lectured. That feeling is why the best channels make even a dry topic feel edited for them.