Revenue per Visitor Significance Calculator

Test whether a difference in revenue per visitor between two A/B test variants is real, using actual order-level data, not just conversion rate.

Drop a CSV with one row per visitor: which variant they saw, and the revenue from that visitor (0 if they didn't buy) here or browse files

CSV up to 25 MB. It's read in your browser and never uploaded.

Why revenue per visitor needs a different test

Most free A/B test calculators only handle conversion rate: did they buy, yes or no. But a variant can win on conversion rate and still lose on revenue, if it drives more low-value orders. This tool tests revenue per visitor directly, using a two-sample test built for revenue's uneven distribution (mostly zeros, with occasional large values), rather than treating it like a simple proportion.

Upload a CSV with one row per visitor: which variant they saw, and the revenue from that visitor (0 for anyone who didn't buy).

Example: the sample file (20 visitors per variant)

Control: mean revenue/visitor: $12.05
Test: mean revenue/visitor: $20.15

Test is 67.2% higher, but with only 20 visitors per group the p-value is 0.316, far from significant. That's the point of the test: a gap that looks big in percentage terms can still easily be chance with a small sample, especially for a metric as uneven as revenue.

Getting good results from this tool

Revenue per visitor needs a bigger sample than conversion rate to reach significance, because a handful of large orders can swing the average a lot. Don't expect the same visitor counts that work for a simple conversion test.
Include every visitor, not just buyers. A row with 0 revenue for someone who didn't purchase is required: leaving out non-buyers would bias the average upward.
Above about 30 visitors per group, the test's normal approximation becomes more reliable. Below that, treat the result as indicative rather than final.

Questions people ask

Why not just use the regular A/B test significance calculator for this?

That tool tests conversion rate (did they buy). This one tests revenue per visitor, which behaves very differently statistically: most visitors contribute $0, and a small number of large orders can dominate the average. A variant can win on one metric and lose on the other.

How do I export order-level data with visitor revenue?

Pull it from your A/B testing tool's raw export (most testing platforms can export per-visitor or per-session revenue by variant), or build it from your analytics tool's ecommerce event data.

Why does this need more data than a conversion rate test?

Revenue per visitor has much higher variance than a simple yes/no conversion, since most visitors are $0 and a few are large. That variance means you need more visitors to tell a real difference from noise.

What if I have more than two variants?

The tool compares the two largest variants in your file and notes how many others it ignored. Run it again after filtering the file to a specific pair if you need every comparison.

Last reviewed 30 September 2026.