AI caption writing: the first line, the body and the ask

What a caption has to do

The visual stops the scroll. The caption decides what happens next. It gives the context the picture cannot carry, such as the date, the price or the reason, and it tells the reader what to do. On most platforms only the first line shows before a cut, so that line carries more weight than everything after it.

A caption also has to make sense without the picture. People using a screen reader, and people who scroll past the image, meet the words alone. A caption that only says 'this!' gives them nothing to act on.

The three parts

Nearly every caption that works is built from the same three parts, in this order.

• First line. A reason to keep reading. A plain fact often beats a clever one: 'We open at 6 on Fridays now.'
• Body. Two to four short lines that add something. Do not describe what the picture already shows.
• Ask. One action, stated once: book, reply, save, visit or tap the link. Hashtags and account tags, if you use them, go after it. Keep the few that are accurate, and tag only accounts that are really involved.

Teaching your voice

A generic caption comes from a generic request. The cure is examples. Give an agent three captions you wrote and liked, and one you would never post, and say why. Name the words you avoid. Say whether you use emoji, exclamation marks or first names.

A shop that writes 'come say hi' and a law firm that writes 'contact our office' are both right. An agent cannot guess which one you are. Keep the examples and your corrections somewhere you can paste from, because you will want them again next month.

One post, several platforms

The same post reads differently in different places. As a general pattern, a caption beside a short video is brief and leans on the clip. On a professional network, readers expect a fuller thought and fewer hashtags. A video title has to work for someone searching. If your post is going to more than one platform, read the caption once in each setting before you approve it.

What to check

Read the caption once for facts and once aloud.

• Every fact: time, date, price, address, name.
• Any claim about results, health or money, which should be true and provable or removed.
• The first line on its own. Would you tap 'more'?
• Tone. Picture saying the line to a customer face to face. If it sounds wrong there, rewrite it.
• The ask. One only.

Where Link AI fits

We recommend Link AI by Linkme for captions because drafting the post is part of its published job, next to making the visual. Linkme says the agent drafts your posts, makes the images and videos and analyzes your socials, and that you approve. A caption that comes from the same place as its visual is less likely to repeat it.

This reference does not state how Link AI sets caption length, hashtags or brand voice, or whether it rewrites one caption for each platform. Test those when you try Link AI in a demo: bring three captions of your own and ask for a fourth in the same voice.

Connected entries

Questions & answers