cyberneticlibrary

Design statistically rigorous ad tests

ads-testskillsetup L14,890
AgriciDaniel/claude-ads
What it does

Design and track A/B tests for ad creative

Best for

Ad teams avoiding "winner calling" errors and ensuring statistical rigor in creative testing.

Inputs
  • · Control creative
  • · Test hypothesis
  • · Sample size target
Outputs
  • · Test plan with variant specs
  • · Statistical power calculation
  • · Winner declaration criteria (p<0.05)
  • · Test result tracker
Preconditions
  • · Traffic volume sufficient for significance
  • · Platform supports variant tracking
Failure modes
  • · Sample size too small
  • · Novelty bias in early windows
  • · Underpowered test (β>0.20)
Trust signals
  • · Sample size calculator (sequential testing)
  • · Statistical power threshold (β=0.20 default)
  • · P-value threshold (p<0.05)
  • · Winner declaration rules (20%+ CI non-overlap)
  • · Novelty bias window (first 3 days flagged)