Key Takeaways
- This tool is built for scenario planning, not one-time guessing.
- Use real baseline inputs before testing optimization scenarios.
- Interpret outputs together to make stronger decisions.
- Recalculate after meaningful context changes.
- Consistency and execution quality usually beat aggressive one-off plans.
What This Calculator Measures
Estimate absolute lift, relative lift, z-score, and confidence index for A/B tests with sample size inputs.
By combining practical inputs into a structured model, this calculator helps you move from vague estimation to clear planning actions you can execute consistently.
This model turns conversion rate differences into a confidence signal so you can make informed test decisions.
How the Calculator Works
Z = (p₂ − p₁) ÷ √(2p(1 − p)/n)Worked Example
- A 4.2% to 4.8% jump is a 0.6% absolute lift.
- Relative lift helps compare across different baselines.
- Higher sample sizes increase confidence.
How to Interpret Your Results
| Result Band | Typical Meaning | Recommended Action |
|---|---|---|
| Z ≥ threshold | Significant lift. | Consider rolling out the variant. |
| Z slightly below | Borderline evidence. | Increase sample size or wait longer. |
| Z far below | Weak evidence. | Collect more data or refine hypothesis. |
| Negative lift | Performance drop. | Stop or iterate on the variant. |
How to Use This Well
- Enter control and variant conversion rates.
- Input the sample size per group.
- Select confidence level.
- Review z-score and confidence index.
- Compare lift to target threshold.
Optimization Playbook
- Grow sample size: reduce variance.
- Increase traffic share: reach significance faster.
- Refine hypothesis: improve lift potential.
- Watch negative lift: stop underperforming tests early.
Scenario Planning Playbook
- Baseline: use current conversion rates.
- Higher traffic: increase sample size by 25%.
- Target lift: compare to your minimum detectable lift.
- Decision rule: require z-score above threshold.
Common Mistakes to Avoid
- Stopping tests too early.
- Ignoring sample size requirements.
- Overreacting to small lifts.
- Not checking negative lift risks.
Implementation Checklist
- Confirm conversion tracking accuracy.
- Set confidence threshold.
- Run test to target sample size.
- Review lift and z-score.
Measurement Notes
Treat this calculator as a directional planning instrument. Output quality improves when your inputs are anchored to recent real data instead of one-off assumptions.
Run multiple scenarios, document what changed, and keep the decision tied to trends, not a single result snapshot.
FAQ
Does a high lift always mean significance?
No. Small samples can inflate apparent lift.
Can I use 90% confidence?
Yes, for exploratory tests or early signals.
What if lift is negative?
Consider ending the test or iterating on the variant.