Pre-publish QA
About to post. Run the caption through the analyzer first. Catch a weak hook or a missing CTA before it ships.
TikTok caption analyzer.
Paste a caption, get scores for hook strength, clarity, CTA, length, readability, and engagement intent. Plus three rewrites.
Paste a caption, set the goal
Six dimensions scored, three rewrites returned. Methodology disclosed below.
Methodology
We score on six dimensions. Hook strength: first-line specificity, curiosity gap presence, reader-relevance. Clarity: is the central idea obvious in 2 seconds. CTA: is there a clear action requested. Length: appropriateness for TikTok (short captions outperform). Readability: punctuation, breaks, plain language. Engagement intent: does the caption invite a reply, save, or share.
Rewrites target the lowest-scoring dimension first. We don't predict virality — scoring measures the caption against best practices, not against the future.
How it works
Paste the caption
Up to 2,200 characters (TikTok's caption limit). Optional fields: niche, target audience, and what action you want viewers to take.
We score six dimensions
Hook strength, clarity, CTA, length appropriateness, readability, and engagement intent. Each on a 0-100 scale with explanation.
Read the rewrites
Three concrete rewrites: tighter, punchier, and a variant optimized for the strongest signal we found. Methodology fully disclosed.
What people use this for
Reviewing a creator's work
Brand manager checking copy from an agency or creator. Score it, share the report, request rewrites with specifics not vibes.
Teaching
Coaching a creator on caption structure. The analyzer shows them what 'a strong hook' actually means in their own copy.
Frequently asked
Will high scores predict virality?
No. Captions are one input among many (video quality, posting time, FYP signal, niche saturation, audio choice). The analyzer scores the caption against best practices, not against the future.
How is hook strength measured?
Three-factor scoring: first-line specificity, curiosity gap presence, and reader-relevance. Methodology is expanded inside the result panel.
Are rewrites guaranteed to score higher?
We tell Claude to optimize for the lowest-scoring dimension first. Usually yes; sometimes the original was already near ceiling and the rewrite is just a stylistic variant.
Privacy: do you store my captions?
Captions go to our server, then to Anthropic. We don't log caption content in a way that identifies an author. See our privacy policy.
The same engine, scaled up.
Run this same scraper across thousands of accounts, route through audited supplier networks, drip-feed inside the FYP probe window. That’s the paid product. The free tool is the engine, scoped down to one query.