Place Demo Captions Without Covering the Button Being Explained

Give each visual element a job

The example below uses a fictional scheduling app called SlotPath. The video shows a user moving a booking from Friday morning to Friday afternoon. The important evidence is the selected booking, the new time and the saved result.

Before recording, choose a window size and arrangement that leave some unused space. If the app cannot fit legibly with that space, plan a closer view of one action rather than shrinking the whole desktop.

The mobile demo recording guide helps with capture decisions. Do that work before fitting text around a recording that is already too small.

Plan text around the interface's changing states

A caption can fit on the opening screen and collide with a dropdown a second later. Scrub the entire demonstration, including menus and success states.

For SlotPath, the first instruction could read “Move the booking without rebuilding it.” That line sits above the app. Once the time menu opens, a shorter line says “Choose the new time.” When the menu closes, the result line becomes “Friday, 14:00.”

These are authored overlay instructions, not a transcript. They can omit spoken filler because they explain the action independently. If you are captioning a real speaker, preserve the speech's meaning and distinguish the transcript from any added headline.

Keep one dominant instruction at a time. A viewer should not have to decide whether to read a long subtitle, inspect a tooltip or follow a moving cursor simultaneously. Remove a redundant instruction before adding another animation.

Choose a layout by the action, not the template

Three arrangements cover different recording problems:

• A full-width instruction band can work when the active control stays below it throughout the shot.
• A dedicated product panel can keep the interface clear while text sits outside it, provided the UI remains legible.
• A close crop can show one control at useful size, but it must retain enough surrounding context to identify the action.

These are planning options, not claims that Prolifik exposes arbitrary layout or cursor-tracking controls. Prepare the recording and any exact annotations in an editor suited to that work.

If the active control crosses your reserved text area, change the shot or split the step. Do not make the overlay jump around every time a menu opens. That creates a second moving target for the reader.

Keep subtitles distinct from labels

An opening title names the task. An instruction tells the viewer what to do. A subtitle represents speech. They can coexist, but they should not compete for the same place or contradict each other.

YouTube's caption tools allow text and timing edits for its separate caption tracks. Changing such a track does not move an instruction already rendered into the picture. Check which layer needs repair before trying to fix a collision.

For caption timing and line breaks, use our short-form caption guide. Keep that general readability work separate from the interface-specific question of what cannot be covered.

Review the muted version through the result

Watch SlotPath's export with sound off. The viewer should identify the original booking, see the time choice and recognize the saved result. If the text tells the whole story while the interface is hidden, the video has stopped functioning as a demonstration.

Then test with the destination's controls visible. The vertical video safe-zone guide explains that additional boundary. A clean source preview is not proof of a readable social post.

Check at actual phone viewing size. Avoid judging a caption from a large desktop player where the UI appears much bigger than it will in a feed. Pause on the most crowded frame and try to read both the instruction and the control label.

Build the short around prepared footage

Use Prolifik's Video Hook & Demo format when the recording already shows the task clearly. The format combines a hook with your own product clip and overlay copy. It does not remove the need to inspect the final text placement.

Its current output uses library music or silence instead of the recording's original audio. A spoken explanation that disappears at generation cannot rescue an obscured control. Make the action understandable in the actual delivered picture.