CS-301g · Module 2
A/B Testing Hooks
3 min read
Every hook is a hypothesis. The A/B test validates or disproves it. The testing protocol: publish two versions of the same content with different hooks, at the same time, to similar audience segments. Measure scroll-stop rate (impressions-to-engagement ratio), read-through rate, and action rate (clicks, shares, comments). The hook that wins on scroll-stop might lose on action rate — a provocative hook stops the scroll but the content does not convert. The hook that wins across all three metrics is the one to scale. Run the test for twenty-four hours minimum. Social media engagement patterns have intraday cyclicality that shorter tests miss.
Do This
- Test one variable at a time — the hook only, not the hook and the content simultaneously
- Run tests for at least twenty-four hours to account for time-of-day engagement patterns
- Measure scroll-stop, read-through, and action rate — all three, not just engagement
Avoid This
- Judge hook performance by likes alone — likes do not correlate with pipeline
- Test with insufficient audience size — fifty impressions per variant is noise, not data
- Change multiple variables between variants — the result tells you nothing actionable