Measure variant combinations with dependable testing metrics. See conversion lift, confidence, revenue, and winning combinations. Turn campaign experiments into smarter decisions with stronger evidence.
Enter the control first, then add each tested marketing combination.
| Combination | Visitors | Conversions | Average order value | Conversion rate |
|---|---|---|---|---|
| Control | 12,000 | 480 | $68.00 | 4.00% |
| Headline A + CTA A | 11,800 | 522 | $70.00 | 4.42% |
| Headline B + CTA A | 11,740 | 548 | $73.00 | 4.67% |
| Headline B + CTA B | 11,690 | 534 | $71.00 | 4.57% |
Conversion rate = Conversions ÷ Visitors
Lift vs control = (Variant rate − Control rate) ÷ Control rate
Revenue = Conversions × Average order value
Revenue per visitor = Revenue ÷ Visitors
Z-score = (Variant rate − Control rate) ÷ Standard error
Standard error = √[ p̄(1 − p̄)(1/ncontrol + 1/nvariant) ]
Pooled rate p̄ = (Control conversions + Variant conversions) ÷ (Control visitors + Variant visitors)
Two-tailed p-value evaluates whether a variant differs from control beyond random noise.
Chi-square test checks whether all tested combinations share the same conversion rate.
1. Enter the control combination in the first card.
2. Add each tested marketing combination in the remaining cards.
3. Fill visitors, conversions, and average order value for every active combination.
4. Set the significance level and projected monthly visitors.
5. Submit the form to generate the report above the form.
6. Review the overall significance test first.
7. Compare each variant’s lift, confidence, and revenue per visitor.
8. Download the report as CSV or PDF for sharing.
It compares several marketing combinations using visitors, conversions, and order value. It estimates conversion lift, revenue lift, pairwise significance, and an overall chi-square test across all combinations.
The first row becomes the benchmark for lift, z-score, and pairwise p-value calculations. Every tested combination is evaluated against that baseline to simplify interpretation.
Many teams use 0.05. Lower values are stricter and reduce false positives. Higher values can surface winners faster but increase the risk of acting on random variation.
A higher observed conversion rate does not guarantee a real improvement. Small samples or noisy data can create temporary leaders that disappear as traffic grows.
It checks whether the full set of combinations behaves differently overall. This helps confirm that at least one combination likely differs before you focus on pairwise winners.
Yes. Some combinations boost conversion rate but reduce order value. Revenue per visitor helps you judge business impact, not just conversion count.
Yes. Leave the extra cards empty or zeroed. The calculator only analyzes combinations with usable traffic and conversion data, while keeping the control mandatory.
Stop when your traffic is sufficient, tracking is stable, and the leading combination shows practical business value with acceptable statistical confidence. Avoid stopping after one short-term spike.
Important Note: All the Calculators listed in this site are for educational purpose only and we do not guarentee the accuracy of results. Please do consult with other sources as well.