Captions people actually read
Why the captions are the only text in a Clipelle ad, what makes a caption readable in a feed, and how to set yours up once.
Clipelle Team6 min read
A viewer scrolling a feed with the sound off meets your ad as a moving picture with words on it. If the words are readable, they get the message. If the words are too small, too many, badly timed or hidden behind the app's own buttons, they get a picture of someone talking. Captions are the part of a vertical ad that decides whether the silent viewer is reached at all, and they deserve more care than they usually get.
Why the captions are the only text
In a Clipelle ad, the captions of what the presenter says are the only text on screen. There is no hook card at the start, no benefit labels floating next to the product, no end card with a button. The one exception is a small optional logo in the top right. This is a design decision, and it is worth explaining.
A video has one audio track and one picture. When the presenter says one thing and a text card says another, the viewer has to choose what to follow. Most people read faster than the presenter talks, so they finish the card, wait, and lose the thread of the voice. When the only text is what is being said, reading and listening move at the same pace. The eye follows the highlighted word, the ear follows the voice, and the two reinforce each other instead of competing.
It also keeps the message honest. If the call to action has to be spoken, it has to be short and clear enough to say out loud. Calls to action that are spoken, not shown goes into what that does to the writing.
What makes a caption readable
Few words at a time
A caption is not a subtitle for a film. In a feed the viewer gives each frame a fraction of a second. Short groups of words, shown as they are spoken, are read; a full sentence dropped on screen at once is skimmed or skipped. Clipelle captions follow the voice word by word, with the word being spoken highlighted, so the amount of text on screen stays tied to the pace of speech.
Timed to the voice
A caption that appears late feels like a mistake; one that appears early gives away the line. Captions generated from the presenter's own speech are timed to it by construction, which is one reason they read better than captions typed in afterwards and nudged into place.
Large, with contrast
The video is watched on a phone at arm's length, often outdoors. Text that is comfortable on a laptop screen is small on a phone. The caption look you choose in your brand style sets the size and the treatment; whichever you choose, check it on a phone, not on a monitor.
In the safe part of the frame
The apps draw their own interface over the bottom of the screen and the right edge: the description, the sound name, the buttons. A caption placed there is partly hidden. Vertical video, explained once covers the layout; the short version is that the middle of the frame is yours and the bottom quarter is not reliably.
One highlight, one colour
Highlighting the word being spoken does two jobs: it tells the eye where to be, and it shows the pace. One colour is enough. In Clipelle the accent colour in your brand style is used for exactly this, the spoken word, and nothing else. Pick a colour that stands out against the presenter's background; if the background is a light colour, a pale accent disappears.
A plain font
Decorative fonts are hard to read at speed and at phone size. A clear sans-serif with even weights reads at a glance. The brand style lets you choose a font; the test is whether you can read a line in a quarter of a second, not whether it matches your packaging.
Setting captions up once
All the caption choices in Clipelle live in the brand style: the accent colour for the spoken word, the caption look, the font, and the optional logo. Set them in Settings and they are prefilled on every ad form. The same choices are also on the form itself, under the brand style section, when one ad needs to differ.
The point of setting them once is consistency. An account whose ads all use the same caption look and colour is recognisable in the feed before the brand is named, and consistency costs nothing when the defaults do the work.
Writing with captions in mind
The script is written for you, but you can edit it when you tick "Review script first", and you can supply your own. Either way, a few habits make the captioned version read better.
- Short sentences. A sentence that takes six seconds to say produces six seconds of caption flow. Two sentences of three seconds each give the eye a rest.
- Concrete nouns. "A ceramic dripper" is read and pictured. "A premium brewing solution" is read and forgotten.
- Numbers that mean something. If the product has a number that matters, the description should carry it so the script can use it: how many hours, how many cups, how many millimetres. A number is one of the few things a viewer reads faster than they hear.
- The call to action in words a person would say. "The link is below" is read and understood. A spoken call to action is also a captioned one, so write it to be read.
The word limit helps here. At about 2.5 words per second, a 30-second ad holds 75 words, which is five or six short sentences. There is no room for a sentence that does not earn its place, and that discipline shows in the captions.
Captions in translated videos
When you translate an existing video with Translate & lipsync, the captions are burned in to the result in the new language, placed low or in the middle. Two things in that process are worth knowing for readability.
First, every translated line is checked against the original before the final render: missing or added words, meaning, numbers, names and natural phrasing. Corrections are made within the timing of the original line, so a caption does not grow into a paragraph to fit a translation.
Second, the brand glossary lets you list terms that must not be translated: the product name, a model number, your brand. Those come through the captions as you wrote them, in every language, which is exactly what a viewer looking for your product in the feed needs to see.
A quick readability test
Before you publish, do this with the finished file:
- Put it on your phone, not your computer.
- Mute it.
- Hold the phone at arm's length and press play once, without pausing.
- Afterwards, say out loud what the ad was about and what it asked you to do.
If you can do step four, the captions work. If you cannot, the fix is almost always one of the items above: too many words, too small, a weak accent colour against the background, or a caption position the app would cover. All of them are a setting away.
Related articles
Captions and formats
Vertical video (9:16) for TikTok, Reels and Shorts, explained once
What 9:16 means, where the apps draw their interface over your video, and how to make or convert an ad for the vertical feeds without redoing it.
29 Sept 2026 · 5 min read
Scripts and hooks
Calls to action that are spoken, not shown
In a presenter-led vertical ad the call to action is a sentence someone says, not a button. Here is how to write that sentence so it gets acted on.
24 Sept 2026 · 6 min read
Translation and localization
Which languages to localize first
A way to choose your first target languages from your own demand, the fit of the video and what each market costs you after the render.
19 Oct 2026 · 6 min read
Translate your first video today
Start with 30 free credits after you verify your e-mail: half a minute of translation. No credit card.
- 30 free credits, no card
- Own voice and lip sync
- Credits never expire
