Skip to content
All free tools
Hook Lab

Hook A/B Predictor

Pick the winner before you post.

~1 min Free · no sign-in ⌘/Ctrl + Enter to run
Input readiness0 words
Nothing stored · results are yours
Go deeper
Run a real video through the full ViralHookAnalyzer.

Paste any YouTube, TikTok or Shorts link and get benchmarks, retention curves, and rewrites — grounded in your actual footage.

Analyze a video free
Before vs after
Before

Today I want to show you something interesting.

After

This one setting doubled my revenue overnight — nobody told me about it.

Why it works: Removes throat-clearing, adds outcome + specificity + curiosity gap.

Real examples by niche
Learn why this matters
+

A/B testing hooks before posting eliminates the #1 cause of dead videos: guessing.

The wrong hook can cost you 60% of reach — the right one compounds distribution for weeks.

Common mistakes
  • ×Writing for yourself instead of the scrolling viewer.
  • ×Being clever when you should be clear.
  • ×Burying the promise past the 3-second mark.
  • ×Copying a format without understanding why it worked.
  • ×Optimizing only one variable (title OR thumbnail OR hook) instead of the whole triangle.
Best practices
  • Iterate — don't ship the first output.
  • Combine numeric specificity with a curiosity gap.
  • Match the title, thumbnail, and hook promise.
Professional tips
  • Read the output out loud. If you stumble, the viewer will too.
  • Save the best 3 outputs and rotate them across the next 3 posts to A/B in the wild.
  • Pair every improvement with a retention benchmark — improvement without measurement is a guess.
  • Run the same input through 2 different tools (title studio + hook rewriter) and compare — the overlaps are gold.
Try this on a real YouTube or TikTok URL

Paste any video link into the full analyzer and get scoring grounded in your actual footage — hooks, retention curves, and thumbnail CTR.

Open live analyzer
How this score is calculated
+
  • ·Scores blend viral-pattern models trained on 2.4M+ analyzed videos with live 2026 platform benchmarks.
  • ·Every metric maps to a real signal: retention curve, click-through, curiosity tension, or pattern-interrupt strength.
  • ·No output is generic filler — the model is prompted to justify every number with a concrete psychological driver.
  • ·Nothing is stored. Your input is never used to train models or shown to anyone else.
Estimated time to complete: ~1 min
Keyboard shortcuts
+
Run the tool⌘/Ctrl + Enter
Focus the input⌘/Ctrl + K
Copy output when focused⌘/Ctrl + C
Show shortcuts?
Recommended stack
Keep exploring