AI Caption Optimizer

What a good caption actually does

A caption has one job before it has any other job: survive the first line. Every feed truncates. Instagram shows roughly 125 characters, LinkedIn about 140 on mobile, Facebook three lines. Whatever sits after the fold only exists if the opening line made someone tap "more". Optimising a caption is mostly optimising that opening line and then making sure nothing after it wastes the attention you just bought.

The second job is to give the reader something to do. Not "double tap if you agree" — platforms have been dampening engagement bait for years — but a genuine prompt: a question you actually want answered, a disagreement you are inviting, a specific next step. Replies and saves are weighted far more heavily than likes on every major feed, and they are the two things a caption can directly cause.

The third job is to be about something searchable. Instagram, TikTok, YouTube, and Pinterest all run keyword search over caption text now. Writing the plain words a person would type — "sourdough starter troubleshooting", not "the loaf chronicles pt 4" — is worth more long-tail reach than a block of hashtags.

How to use the output

Treat the score as a diagnostic, not a grade. A caption that scores 62 because it has no call to action is a one-line fix; a caption that scores 62 because the hook is buried in sentence three needs the paragraph reordered. Read which dimension is dragging before you rewrite anything.

Take the suggested rewrite as a draft, then put your voice back in. AI rewrites tend to flatten idiom, cut the specific number, and replace the odd word that made the sentence yours. The version worth posting is usually the rewrite's structure with your original's texture.

Then run it once more. Optimising the hook usually changes the length, and changing the length usually changes whether the CTA still lands above the fold. Two passes is normal; five means you are polishing something that was not worth posting.

Mistakes this tool will flag most often

Front-loading context. "So I've been thinking about this for a while and I wanted to share…" spends the only line that matters on throat-clearing. Cut until the first sentence is the claim.

One caption pasted to five platforms. A 280-character post reads thin on LinkedIn; a 2,000-character LinkedIn essay is unreadable on Threads. Same idea, different shape — that is what the per-platform pages below are for.

Hashtag stuffing. Thirty tags on Instagram no longer buys reach and reads as spam; hashtags actively suppress nothing on X but do nothing either; on Mastodon and Pinterest they are load-bearing. The right number is platform-specific and this tool checks it against the platform you selected.

No point of view. The most common reason a technically clean caption gets no engagement is that nobody could disagree with it. If the caption could have been posted by any account in your category, it will perform like a caption from any account in your category.