Where the eye goes first

Somebody told you the edit was too slow, so the next one came back quicker. More cuts, more movement, a zoom on every line, text flying in from the side.

It feels urgent now. Watch again for a different question: can you follow the thought as easily as the movement?

A faster cut can create energy, but it still needs time for the important image or sentence to register. Look for moments where the next thing arrives before the previous one has become clear.

Why does a fast cut lose people?

Because reading and watching use the same attention, and cutting interrupts both.

If the shot removes the text before someone can finish it, they may need to replay or may miss the point. Give the important line enough time to be read in context.

The same applies without text. A new location, a new face, a hand entering frame: each needs a moment to be understood before it can mean anything. Cut before that moment and you have paid for a shot nobody saw.

An ad that moves quickly and is understood is fast. An ad that moves quickly and is not understood is noisy, and the two look identical in an editing timeline.

Where is the viewer looking?

Somewhere specific, and rarely everywhere.

Faces, movement, contrast and words can compete for attention. Arrange them to suggest a useful path, then check whether another viewer follows it.

This is what an eye path means. Work out, shot by shot, where the eye is most likely to go, build the cut around that, then check it rather than trust it, so the next important thing arrives where the eye already is rather than in the corner it has just left.

The consequence is unglamorous. If the eye ends a shot on the left, put the next thing on the left. If a face is talking, do not put your key sentence in the opposite corner and expect both to land.

How long does a line of text need on screen?

Reading aloud is a useful timing check, not a universal minimum. Test the line in the actual edit and viewing size, allowing for the other things the viewer needs to follow.

A viewer in a feed is reading in worse conditions than you are: smaller screen, moving hand, half attention, somebody probably talking to them. And if you are checking the timing by reading the line in your head while already knowing what it says, you are not checking anything.

Do it for every line, out loud, at the speed you would say it to a stranger. If the line disappears before you finish, it is too short, and the fix is either more time or fewer words. Fewer words is usually the better fix, and the less welcome one.

When movement competes with the words

Give the words you need read a clear moment.

If your key sentence appears while the camera is pushing in, while a transition is resolving, or while a hand crosses the frame, it is competing against motion, and motion usually wins that contest. That is not a stylistic claim: sudden movement is one of the strongest things human vision reacts to, enough to beat most on-screen text, even if a viewer's goal and the meaning of the shot can pull the other way.

So the important sentence gets a still moment. Let the movement finish, then show the words, or show them over something calm. It costs a fraction of a second, and it gives the offer time to be read.

Check overlapping elements together. A logo, caption and offer may coexist, but each needs a clear role and enough time. Remove or delay the element that obscures the point.

Three checks before it goes live

None of these need software and all of them work on the version you already have.

Read the on-screen lines aloud as a rough timing check, then watch muted at the intended size. Note where you lose the thread and ask someone unfamiliar with the edit to do the same. These are usability observations, not a retention prediction.

Fix in that order. Timing first, since it is free. Then reordering. Then recutting, and only if it is still not working.

Building the cut around the eye rather than the beat

Most edits are built to the music, because music gives you a grid and a grid is comfortable. The beat does not know where the viewer is looking, and an ad cut purely to a beat will occasionally land a change of shot precisely in the middle of somebody's sentence.

In this studio the eye path is worked out first and the cut is built around it, which means the music is fitted to the edit rather than the edit to the music. It is a slower way to start and it removes an entire category of fault: a change that arrives away from the viewer's eye, or while they are still reading.

What that leaves you with is your current ad. Watch it once more and find the first cut that lands on a sentence. There is usually one, it is usually early, and moving it is a decision you can make today.

Sources and evidence

What here is independent research and what is our own method.

Standards and platform facts. Research on visual attention finds that contrast, faces and motion do pull the eye, but that goals and the meaning of a scene can override that pull, and that attention itself is not the same as comprehension: getting a look is not the same as being understood. A separate strand of that research shows a face draws fixation strongly enough to interfere with reading text placed near it, which is the mechanism behind not putting a key sentence in the corner opposite a talking face. Neither strand is about advertising or editing specifically; both are about where the eye goes, and why, in general.

Our method. Working out, shot by shot, where the eye has just been sent, then checking rather than trusting where the next important thing lands, is a discipline this studio applies to its own cuts, not a formula the research itself hands over.

References.

• Cerf, M., Frady, E. P. & Koch, C. Faces and text attract gaze independent of the task. pubmed.ncbi.nlm.nih.gov. In natural images faces and text attract gaze regardless of task (about 16.6x and 11.1x more than matched regions). Gaze attraction is not click-through or sales; both faces and text are strong attractors, so text is not simply beaten by motion.
• Franconeri, S. L., Hollingworth, A. & Simons, D. J. Do new objects capture attention?. pubmed.ncbi.nlm.nih.gov. Sudden visual transients (motion/onsets) capture attention automatically. A busier/animated frame is not shown to improve click-through, comprehension or retention.