Revenue per Session Beats Conversion Rate: The Metric That Stops Bad Test Calls

A test can lift conversion rate and still lose money. Conversion rate counts orders, not dollars, so a variant that pushes shoppers toward your cheapest product wins the metric and shrinks the basket. Revenue per session catches that. Conversion rate hides it.

The mistake is treating conversion rate as the scoreboard. It is one number, and it answers one question: how many visitors bought something. The question that matters is how much money each visitor brought, and that is a different number with a different answer.

The cheap-product trap, worked through

Variant A converts at 3 percent with a 40 dollar average order. Variant B converts at 2.8 percent with a 55 dollar average order. A wins the conversion rate by a clear margin. B earns more per session: 1.54 dollars against 1.20. The test that calls A the winner ships the variant that makes less money, and the team celebrates a loss.

The mechanism is common. A variant that promotes the cheapest product, or a discount, or a smaller pack, lifts the order count while shrinking the basket. The conversion rate goes up and the revenue goes down, and the dashboard shows a green number next to a smaller bank balance.

Why revenue per session is the primary metric

Revenue per session folds conversion and order value into one number, so it cannot be gamed by pushing cheap products. It is noisier than conversion rate, because order values vary more than order counts, which means it needs more traffic to reach the same confidence. That is the honest cost of a real answer, and it is worth paying.

Use it as the primary metric for any test that can move order value: product page changes, recommendations, bundles, pricing, free shipping thresholds. For tests that cannot move order value, like button copy or form fields, conversion rate is still the right read, because revenue per session just adds noise there.

When conversion rate is still the right read

Button copy, form fields, checkout steps, and anything where the basket cannot change: conversion rate is the cleaner metric, because the order value is fixed and the only question is whether more people complete. The rule is to pick the metric that matches what the test can change, and to report the other one alongside it so the trap stays visible.

Frequently asked questions

What is revenue per session?

Total revenue divided by total sessions. It answers how much money each visitor brought on average, folding conversion rate and average order value into one number. A variant can win conversion rate and lose revenue per session, which is the trap it exists to catch.

Why is revenue per session better than conversion rate?

Because it cannot be gamed by pushing cheap products. Conversion rate counts orders, so a variant that sells more cheap items wins it while shrinking the basket. Revenue per session counts dollars, so the variant that earns more wins.

How much traffic do I need to measure revenue per session?

More than for conversion rate, because order values vary more than order counts. Small stores may need a month or more to reach confidence on revenue per session, and should run fewer tests for longer rather than many tests that never reach a verdict.

Can revenue per session mislead too?

Yes, when a single large order distorts the average. A one-off wholesale order in the test period can make a losing variant look like a winner. Check the median order value alongside the average, and exclude outliers that have nothing to do with the test.