What is the revenue-per-visitor framework?
The revenue-per-visitor framework is a way of judging conversion work by the revenue each visitor produces rather than by the share of visitors who buy. Revenue per visitor is total revenue over a period divided by the number of visitors in that period. Because it equals conversion rate multiplied by average order value, a change can be traced to the half it moved: more buyers at the same value, or the same buyers spending more, or the trade-off between them.
The Conversion FTE on ftes.ai reports revenue per visitor as its headline number and is reviewed on it in the Targets tab, because it is the conversion metric that a discount cannot flatter.
What are the framework's dimensions?
The framework has four dimensions that make the number comparable across months and useful for diagnosis. Each is fixed before results are read and printed on the report.
- Period: a fixed window, usually a month for reporting and the test duration for experiments.
- Visitors: sessions or unique visitors, chosen once and held; the choice is stated on every report.
- Segments: by device and by traffic source, because mobile and paid traffic convert differently and a change can move one without the other.
- Decomposition: conversion rate and average order value shown beside the headline, so the cause of a change is visible.
How is it calculated, with a worked example?
Revenue per visitor is revenue divided by visitors; the decomposition is conversion rate multiplied by average order value. A labelled worked example, illustrative only: a store had fifty thousand visitors in a month, one thousand orders, and one hundred thousand in revenue. Conversion rate is one thousand divided by fifty thousand, two percent. Average order value is one hundred thousand divided by one thousand, one hundred. Revenue per visitor is one hundred thousand divided by fifty thousand, two.
Suppose a test raised the rate to two point four percent by adding a discount that dropped the order value to eighty. Revenue per visitor becomes zero point zero two four multiplied by eighty, which is one point nine two. The rate rose, revenue per visitor fell, and the framework reports the change as a loss with both halves visible.
Why judge experiments on revenue per visitor first?
Experiments are judged on revenue per visitor first because it is the metric that reflects what the store actually earns from the change, while conversion rate reflects only whether more people bought. Most changes to a funnel move both halves a little. A shipping threshold raises order value and may lower the rate; a simplified product page raises the rate and may lower the value if it hides the premium option. Only the combined number says whether the store is better off.
Conversion rate and order value stay on the report as diagnostics. They say where the change acted, which is what the next hypothesis needs.
When does margin per visitor replace it?
Margin per visitor replaces revenue per visitor as the judge when a change affects cost, not just price: a free-shipping threshold, a discount, a bundle with a low-margin item, a payment option with a higher fee. In those tests the Conversion FTE asks the owner for the cost inputs, landed cost per SKU, shipping cost per order, payment fees, and reports margin per visitor beside revenue per visitor, with the inputs stated. Where the store cannot supply the inputs, the report says so and judges on revenue with that caveat.
Margin inputs are also what make a shipping-threshold test honest: a threshold that raises order value while giving away more shipping than it earns is visible only in margin.
How should it be reported month to month?
Revenue per visitor should be reported monthly against the baseline, by device and traffic source, with the decomposition beside it and the period and visitor definition printed. A month is not a trend; the Conversion FTE's review says which changes concluded in the month and their measured effect over their test periods, and does not annualize a two-week result into a yearly claim, because that is an extrapolation and not a measurement.
Seasonality and traffic mix move the number without any change to the store, and the report names those when it can see them: a paid campaign that brought lower-intent visitors, a sale month, a holiday. The framework does not make those disappear; it makes them visible.