Average order value calculator
Evaluate order-size changes without losing the visitor-level view needed for experimentation.
Use AOV carefully inside A/B tests
AOV is useful for understanding basket size, but experiments should keep the visitor as the unit of analysis so non-buyers remain represented.
1. Unit
- Keep visitor-level revenue: Upload or enter revenue per visitor rather than only averaging orders. Otherwise, conversion-rate changes can distort the result.
2. Guardrail
- Interpret AOV as context: A higher AOV can still be bad if fewer visitors buy. Pair order value with RPV and conversion rate.
3. Variance
- Expect more sample: Revenue and order values vary more than conversion counts, so they usually need more traffic.
Use visitor-level revenue for the statistical decision and AOV to explain how order size contributed to the result.
How to read revenue and AOV output
AOV is most useful when it explains why visitor-level revenue moved.
- Revenue per visitor: RPV keeps every visitor in the denominator, including visitors who did not order.
- Order-size context: AOV helps explain basket changes, but it should not replace the visitor-level decision metric.
- Sample size: Higher revenue variance usually increases the sample needed for a trustworthy result.
Avoid denominator drift
Order-only averages can hide what happened to visitors who did not buy.
- Use RPV for the main statistical decision when conversion rate can also move.
- Use AOV to explain order composition, not to ignore non-converting visitors.
- If transaction grouping differs from visitor grouping, document that before acting.
AOV and revenue methodology
The calculator uses the revenue path for AOV-related decisions because visitor-level revenue preserves the experiment denominator.
- Visitor-level revenue: Revenue per visitor includes both buyers and non-buyers, which makes it better aligned with A/B test exposure. Use this when a variant may change either purchase rate or basket size.
- AOV guardrail: Average order value helps explain whether winning revenue came from larger orders, but it is not enough by itself when conversion changes. Use this when order composition matters to the launch decision.
- Continuous-metric variance: Revenue metrics rely on observed variance, so outliers and skewed order values can materially increase the required sample. Use this before promising a short revenue test.
Inputs for order-value tests
Use consistent visitor and revenue definitions before interpreting order-value movement.
- Average order value: AOV is revenue divided by orders. It excludes visitors who did not order, so it is best used as context.
- Revenue per visitor: RPV is revenue divided by exposed visitors. It preserves the test denominator.
- Revenue variance: The spread in visitor-level revenue values. Higher variance increases the sample needed.
Related calculators
Common questions about AOV tests
AOV is useful, but only when the denominator is clear.
- Is AOV the same as revenue per visitor? No. AOV divides revenue by orders. RPV divides revenue by exposed visitors. RPV is usually the better primary experiment metric.
- Why does this page use the revenue calculator path? Because the statistical decision should keep visitor exposure in the denominator when purchase rate and order value can both change.
- Can AOV be a guardrail? Yes. AOV is useful for explaining basket-size movement as long as it is interpreted alongside conversion rate and RPV.
- Why do AOV tests need more traffic? Revenue and order values are more variable than binary conversion outcomes, so the detectable effect usually needs a larger sample.