What you are looking at
What
Controlled experiments that isolate the effect of a single change on a key metric. Each user is randomly assigned to Control (A) or Variant (B) — everything else stays identical so any difference in conversion is attributable to the change being tested.
Why
Recommendations and intuition tell you what might work. A/B tests tell you whether it worked, with a measurable confidence level. Statistical significance (p < 0.05) means there is less than a 5% probability the observed difference is due to chance alone.
How to read results
Uplift % = relative improvement of B over A. p-value = probability the result is random noise (lower = more confident). Winner declared only when p < 0.05 AND the uplift direction aligns with the business goal.
3
Active Experiments
3
Significant Results
72,000
Total Observations
Tests Won by Variant B
95%
Confidence Threshold

AB-001: Conversion by Device

Price-first layout (B) outperforms image-first (A) on every device type. Mobile shows the biggest relative uplift — users on small screens benefit most from seeing price before scrolling through images.

AB-001: Weekly Conversion Trend

Variant B consistently leads throughout the test period. A stable, consistent gap (not just a spike in one week) confirms the result is not driven by a single seasonal event.

AB-003: App Retention Cohort

Price-anchor creative (B) acquires users with higher intent — they know the price before installing, so they are less likely to churn after seeing it in-app. Day-30 retention gap of ~5pp compounds significantly over the user lifecycle.

Uplift Summary Across All Tests

Each bar shows the relative uplift of Variant B over Control A on the primary metric. All three are statistically significant (p < 0.05). Revenue uplift for AB-002 is highest because it affects a high-value step (checkout completion).