Captions in vertical video: write for people who read on the move

A caption can repeat every word and still be hard to follow. The problem shows up when the lines change too fast, an expression gets split in two or the text competes with the product for space. On a small screen, each of these decisions weighs on comprehension.

The goal of a caption is to let people follow the message, including when the audio is off or isn't enough for the viewer. To get there, review the meaning before you choose the look. An eye-catching style doesn't fix a wrong word or a sentence that vanishes before you finish reading it.

Start with a checked transcript

Use the final speech as your reference. If the script changed during production, update the caption from what is actually heard. Brand names, measurements, prices and acronyms deserve a separate check, because a small error can change the information.

Don't publish the first automatic version without watching it. Even when the text comes from your own script, the in and out timings may need adjusting. Keep a copy of the approved speech so you can compare the video, the caption and the post description.

Make the break follow an idea

Compare these two splits:

Check the measurements before choosing.
See the options on the product page.

And:

Check the measurements before
choosing. See the options on the product page.

The first split keeps two complete actions. The second leaves a sentence hanging and attaches its ending to the next invitation. Space won't always allow the ideal split, but keeping phrases together usually makes reading feel more natural.

Pay special attention to negations. "Not suitable for hot liquids" must keep its meaning for the whole time it's on screen. If the word "not" disappears before the rest, a viewer glancing quickly may pick up the opposite instruction.

Reduce visual competition

Choose a readable typeface, consistent contrast and a position that doesn't hide the important element of the scene. A background or outline can help when the image changes in brightness, as long as the result doesn't form a heavy block over the face.

Avoid using size, color, animation and position to emphasize every word. If everything gets emphasis, nothing guides the reading. For a product series, it may be enough to highlight the option's name or the information that answers the main question.

The test is functional: can the person follow along without losing what is being shown? Watch on your phone and hold the screen at a normal viewing distance, without bringing it closer to make up for a small font.

Adjust the pace from real reading

There is no number of words per block that works for every sentence, audience and screen. A run of short words can be easier to read than an unfamiliar acronym followed by a measurement. Use numbers as a working reference when you need to, and validate the result by watching.

Read each segment silently while the video plays. If you need to look ahead to the next block, simplify the script or adjust the timings in the editor you have. Speeding up caption changes to fit a long line of speech passes the problem on to the viewer.

When you have to shorten, preserve the meaning. Don't remove an important condition to make it fit on a line. It is better to rewrite the speech and generate a clearer version than to leave an incomplete promise on screen.

Check the composition on the destination channel

The 9:16 file fills a vertical area, but buttons, descriptions and other app elements can cover parts of it. The position of those elements changes with the channel and the viewing experience. That's why you should check the publishing preview with the final material.

Look at the first and last blocks. The opening must come in on time with the speech, and the last sentence must stay on screen long enough to be understood. If the video ends by cutting off the reading of the invitation, adjust the edit or shorten the ending.

Review with and without sound

On the first pass, turn off the audio and check whether the message is still understandable. On the second, turn the sound on and compare sync, words and pauses. On the third, look only at the image to see whether the captions hide any relevant information.

If the editor you use offers only limited controls, adapt the speech to what it can present clearly. Don't count on sync or export adjustments you haven't yet verified in your chosen tool.

When you produce in AvataReel, review the captions in the final file before you publish. To organize the full check, use the AI video checklist. Then prepare your video with sentences that work both for the ear and for reading. A good caption follows the message; it doesn't make the viewer work to rebuild it.