What is revenue per visitor?

Why is RPV better than conversion rate?

Because conversion rate can improve while revenue falls, and it reports that as a success.

A test promoting a cheaper entry plan lifts conversion rate from 2.0% to 2.6%, a 30% gain. It also shifts buyers from the mid tier to the entry tier, dropping average order value from $180 to $120. Conversion rate declares a large win. Revenue per visitor moves from $3.60 to $3.12, a 13% loss.

That pattern is common in any test touching pricing presentation, plan ordering, discounting, or bundling. Conversion rate is blind to it by construction, because it treats every conversion as identical.

Key takeaway: conversion rate measures whether people bought. Revenue per visitor measures whether the business is better off.

How do you calculate it?

revenue per visitor = total revenue / total visitors



Equivalently, conversion rate multiplied by average order value, which is why it captures both.

Decisions to make explicit:

Visitors or sessions. Divide by whichever unit your test platform assigns on, since a denominator that disagrees with the assignment unit biases the comparison. Per-visitor suits longer horizons and repeat-purchase businesses.

Which revenue. Gross, net of refunds, or contribution margin. Net of refunds is the honest default, since a variant that lifts purchases and refunds equally has achieved nothing.

For B2B SaaS, substitute expected value per conversion, using average contract value or pipeline value weighted by stage.

What makes RPV hard to test on?

Its variance, which is the tradeoff for its honesty.

Conversion rate is a binary outcome per visitor, so its variance is well behaved. Revenue is continuous and heavily skewed, since a small number of large orders dominate the total. A single unusually large order can swing a test result.

Practical consequences:

RPV tests need larger samples than conversion rate tests to reach the same confidence.

Outliers need a policy set in advance. Winsorizing at a percentile, or capping order value, decided before the test rather than after seeing which variant the whale landed in.

Report both metrics. Conversion rate and average order value alongside RPV explain the mechanism. RPV alone tells you the direction and not the reason.

For low-volume B2B sites, RPV is often untestable directly. Use it as a guardrail metric while testing on a validated micro conversion.

Related terms

Average order value · Macro conversion · Guardrail metrics · Landing page conversion rate · Cohort analysis

Service: Conversion Rate Optimization.

FAQ

Is revenue per visitor better than conversion rate?

As a primary test metric for anything that could change purchase composition, yes. Conversion rate remains useful for diagnosis and for tests where order value cannot plausibly change.

How do you use RPV for B2B SaaS?

Weight each conversion by what it is worth instead of counting conversions equally. The trap is stale weights. A value per demo request set a year ago and never revisited turns RPV back into conversion rate with extra arithmetic, and it will agree with conversion rate on every test, which is the signal that the weighting has stopped doing any work.

Why is my RPV test inconclusive when conversion rate was significant?

Revenue is far more variable than a binary conversion, so the same sample supports less certainty. Report both: treat the conversion rate result as the finding about behavior and the RPV result as the finding about money, rather than promoting whichever one cleared the threshold.