All tags
Tag · 3 articles

#Experiment design

Articles about Experiment design: practical guides, analysis and case studies with their market context clearly identified.

What to do after an inconclusive A/B test
PerformanceGlobal frameworkOctober 18, 20263 min

What to do after an inconclusive A/B test

Choose whether to stop, repair, simplify or continue an A/B test when its evidence still allows both useful and unhelpful business outcomes.

Marginal vs. average CPA when increasing ad spend
PerformanceGlobal frameworkOctober 18, 20264 min

Marginal vs. average CPA when increasing ad spend

Compare additional spend with additional observed actions, separate marginal from average CPA and avoid a causal verdict from two unmatched periods.

Testing several variants: control false discoveries
PerformanceGlobal frameworkOctober 18, 20263 min

Testing several variants: control false discoveries

Declare the comparison family before testing several variants, show how multiplicity changes a decision and separate exploratory rankings from confirmed evidence.