The biggest mistake in A/B testing is ending the test too early once you see a 'winner'. You must account for the 'Novelty Effect' (people clicking things just because they are new) and ensure you have a large enough sample size for statistical significance.

The Testing Stack

• Hypothesis creation: What are we actually trying to prove?
• Randomization: Ensuring cohorts are truly split
• Post-test analysis: Segmenting the winner by user type (did it only win for free users?)