Principle
High confidence
Test Quality Peaks at 1–10 Annual Tests Per Engineer; Drops 87% After 30
Impact per test peaks when teams run 1–10 tests per engineer annually. Beyond 30 tests per engineer, expected impact drops 87%. This suggests testing more without a prioritization framework destroys value.
Hypothesis
High test velocity without prioritization dilutes quality and reduces average lift per experiment.
Why This Works
Key Learnings
Viewed 12+ times
Related Entries
Research Finding
Principle
Commonly Used Metrics Are Poor Predictors of Business Impact
measurementany
High confidence
Research Finding
Principle
[company] Meta: Top Homepage CRO Patterns — Ranked by Average Lift
testing_strategyhomepage
High confidence
Research Finding
Principle
Personalized Experiences Generate 41% Higher Impact Than Generic
personalizationany
Expected lift: 20%-50%
High confidence
Best Practice
Loser
Ghost / Outline Buttons Lose to Filled Buttons — Consistently
ctaany
Expected lift: -25%-0%
High confidence
Research Finding
Principle
4+ Variation Tests Win 2.4x More Often With 27% Higher Lift
testing_strategyany
Expected lift: 10%-30%
High confidence