Experimental Design

A/B Testing

End the debate about which version performs better. We give you a statistically rigorous answer — not just which variant had better numbers, but whether the difference is real enough to act on and what it means for revenue.

Dataset

800 Records

Customer acquisition data

1. Analysis Name

A/B Testing

We tell you whether the challenger actually beats the control — and whether that difference is statistically real — so the decision to switch or stay is backed by evidence, not opinions.

2. Problem Context

What you'll be able to decide

Should you roll out the new variant, stick with the control, or run a longer test? This page separates the observed performance gap from statistical significance and revenue impact, so you can make a call you can defend — and understand exactly how confident to be in it.

3. Observed Data

Observed cohort outcomes and descriptive summary

The first chart shows the actual converted and non-converted counts for each cohort. The table summarizes sample size, conversion, revenue, and cost before any inference is applied.

Variant Summary

4. Workflow

How the experiment answer is built

The workflow checks observed lift, quantifies statistical uncertainty, and then reviews revenue behavior to confirm whether the apparent winner holds up commercially.

01

Build cohorts

Keep control and challenger distinct and compare like-for-like channel groups.

02

Test conversion lift

Apply a two-proportion z-test at the selected confidence level.

03

Validate business impact

Compare revenue distributions to avoid picking a statistically better but commercially weaker option.

Experiment Verdict

-

Winning Variant

-

Test Statistics

P-value: -
Z-score: -
Relative uplift: -
CI: -
Highest CLV medium: -
Highest avg CLV: -

5. Conclusion

Recommended experiment answer

Why this is the best answer