What is A/B Testing?

A/B testing (split testing) compares two versions to determine which performs better. It's the gold standard for data-driven decisions.

  • Control (A): Current version
  • Treatment (B): Modified version
  • Random assignment: Users see A or B

The A/B Testing Process

  1. Form Hypothesis: "Changing X will increase Y because Z"
  2. Determine Metrics: Primary, secondary, guardrail
  3. Calculate Sample Size: Before starting
  4. Run Experiment: Random assignment, don't peek!
  5. Analyze Results: Statistical significance

Key Statistical Concepts

ConceptDefinitionTypical
Significance (α)False positive probability5%
Power (1-β)Detect true effect80%
p-valueChance of result< 0.05

Common Pitfalls

  • Peeking: Stopping early when you see desired result
  • Multiple comparisons: Testing many variants
  • Novelty effect: Users react to newness
  • Insufficient sample: Underpowered tests

Conclusion

Key Takeaways

  • A/B testing: Control vs. Treatment
  • Hypothesis first, then test
  • Don't peek - run full duration
  • p < 0.05 = statistically significant
  • Consider practical significance too