Skip to content
Faisal Mahmud TusherDigital marketing, SEO & analytics
with Faisal Tusher

Metric

Revenue per Visitor

Revenue divided by the number of visitors in the same period.
AnalyticsmetricUpdated

Formula

Revenue ÷ Visitors

Unit: currency · higher is better

Definition

Revenue per visitor divides the revenue earned in a period by the number of visitors in that period. Both parts come from the same scope: the same site or test, the same dates, and the same definition of who counts as a visitor.

Revenue ÷ Visitors(currency)

  • Revenue is the order value recorded in the period, with a stated rule on tax, shipping, discounts and refunds. In a test, it is the revenue from the visitors assigned to each variant, including the visitors who bought nothing.
  • Visitors is the denominator, and the main choice to make. Users (people, as far as the tool can tell) and sessions (visits) give different figures, which is why the metric also goes by revenue per user and revenue per session. Choose one, write it down and keep it; in a test, use the unit the experiment randomises on.

The metric is the product of two familiar ones. Orders divided by visitors is conversion rate, and revenue divided by orders is average order value, so:

Revenue per visitor = Conversion rate × Average order value

That identity is the reason to use it. A change can lift conversion rate and still lose money if order value falls further, and revenue per visitor shows the net effect where either part alone would mislead.

How to interpret it

In experiments, revenue per visitor plays one of two roles. As a primary metric it decides tests whose purpose is commercial: pricing displays, offers, delivery thresholds, product recommendations. As a guardrail it watches tests that are judged on something else but could harm revenue. The A/B testing guide names it as the guardrail for a discount prompt, and the experiment hypothesis template asks for it whenever a test is expected to lift order count. The opposite error matters too: the guide to writing hypotheses warns that using revenue per visitor as the primary metric for a small, local change, such as a form field, buries the effect in noise.

Outside tests, read it by traffic source and landing page. Two channels with the same conversion rate can be worth very different amounts per visit if one brings larger baskets. Joining revenue from the order system with visits from analytics at the same grain is a common reporting task, and the guide to blending data in Looker Studio covers how to aggregate both to one date and dimension before dividing.

The noise comes from the spread of order values. Most visitors spend nothing, and a handful of large orders can shift the average for a variant on their own. Revenue per visitor therefore needs a larger sample than conversion rate to detect the same relative change, and a test planned on conversion rate alone may end inconclusive on revenue. The sample size and significance guide explains how the sample is planned before launch.

Illustrative example — not a real result

A free-delivery banner that changed nothing

Gorsefield Wine Cellar tests a banner offering free delivery on every order against its usual threshold of £60. Over four weeks each variant receives about 18,000 visitors. The control converts 2.4% at an average order of £74, giving revenue per visitor of about £1.78. The banner converts 2.8%, which the team first reports as a win, but its average order falls to £63 because buyers stop adding a bottle to reach the threshold, giving about £1.76. The difference in revenue per visitor is small and within the noise of the test, and the free delivery itself costs around £4 on each extra order. The banner is not rolled out.

What moves it

  • Conversion rate. Anything that changes the share of visitors who buy moves the metric directly.
  • Order value. Bundles, thresholds, recommendations and price changes shift the second factor; the average order value levers guide covers them.
  • Traffic mix. More visits from low-intent sources lower revenue per visitor without anything on the site changing.
  • The denominator rule. Switching between users and sessions, or changing how bots and internal visits are excluded, changes the figure overnight.
  • Revenue rules. Including or excluding tax, shipping and refunds changes the numerator; the rule must be the same in every period compared.

What good looks like

There is no reliable benchmark for revenue per visitor, and the site does not cite one. The figure depends on price points, category, purchase frequency and the mix of traffic a site attracts, so a comparison with another business says almost nothing. Judge it against the site's own history under the same denominator and revenue rules, by source and by landing page, and against economics: what a visit is worth set beside what a visit costs to buy. A channel whose cost per visit is above its revenue per visit, after margin, is losing money on every visit unless later purchases make up the difference.

Revenue per visitor is CVR multiplied by AOV, so each of those explains a share of any change. In an A/B Test, it is the metric most likely to need a larger sample before a difference reaches Statistical Significance. Revenue per email is the same idea applied to sends rather than visits. The conversion rate calculator works out the first factor from visits and conversions.

Common mistakes

  • Dividing by buyers instead of visitors. That is average order value, not revenue per visitor.
  • Mixing users and sessions between periods or between a test and its report.
  • Declaring a winner on conversion rate when revenue per visitor, the guardrail, has not moved or has fallen.
  • Leaving out visitors who bought nothing. In a test, every assigned visitor belongs in the denominator.
  • Ignoring outliers. One very large order can decide a small test; read the result with and without it, under a rule set before launch.
  • Using it as the primary metric for small changes whose effect on revenue cannot be detected with the available traffic.

More terms to look up

Calculators

Where this term is used

Related