Revenue Per Visitor (RPV)

RPV · Revenue per Visit

Definition

Revenue per visitor is total revenue divided by unique visitors over a period; a summary metric of e-commerce and experiment performance.

What RPV is and how to compute it

Revenue per visitor is the average revenue from one unique visitor over a period:

RPV = revenue / unique visitors

If a store took 3,000,000 in revenue from 100,000 unique visitors in a week:

RPV = 3,000,000 / 100,000 = 30 per visitor

RPV is not what each visitor pays. It is an average across everyone: most buy nothing, a minority
produce the entire revenue.

RPV as the product of CR and AOV

RPV decomposes into two components:

RPV = CR × AOV

At a 2% conversion rate and an order value of 1,500: RPV = 0.02 × 1,500 = 30.

That matters for reading A/B tests: RPV can grow through conversion, through order value, or
through both. Watching one component while ignoring the other risks the wrong conclusion.

Why RPV is the best A/B test metric

E-commerce tests routinely face the question of which primary metric to use — conversion, order
value or revenue. RPV settles it:

Scenario	CR	AOV	RPV	The right conclusion
Conversion up, order value flat	↑	→	↑	A win
Conversion down, order value up	↓	↑	↑	A win, invisible in CR
Conversion up, order value down	↑	↓	→ or ↓	Ambiguous
Nothing moves	→	→	→	A draw

Tip: use RPV as the primary metric and conversion and order value as secondary ones that
explain the mechanism. If RPV rose only through order value while conversion fell, check whether
upper-funnel behaviour degraded.

RPV by segment

An aggregate RPV hides differences between segments. When RPV rises in a test, it is worth checking:

Did it rise equally on mobile and desktop?
Does the effect differ between new and returning shoppers?
Are there categories where the change did the opposite?

Breaking RPV down by segment is a standard step of post-test analysis.

RPV and personalization

Personalized recommendations move both components of RPV:

Conversion rises because the shopper finds a relevant product faster
Order value rises through cross-sell and upsell on relevant items

The combined effect on RPV is multiplicative, which is why well-designed personalization experiments
tend to show a strong return.

Frequently asked questions

Why is RPV better than conversion rate as a test metric?

Conversion rate shows only the share who bought, not how much they spent. RPV combines both effects: a rise in conversion and a rise in order value both show up in it. A variation can lower conversion and raise order value, producing a net gain that conversion rate alone would hide.

What RPV value is normal?

It depends on the vertical and the order value. Fashion with a high order value and low conversion can land at the same RPV as grocery with the opposite profile. The absolute number carries no information without context; the trend and the test delta do.

How does RPV differ from ARPU?

Average revenue per user is computed over registered or identified users, usually monthly. RPV is computed over all visitors including anonymous ones, per session or per period. RPV is lower, because it includes visitors who bought nothing.

Gravity Field

RPV in Gravity Field

Gravity Field reports revenue per visitor for every test and every recommendation strategy, with attribution to the widget the shopper interacted with.

✦RPV as the primary metric in every A/B test

✦RPV broken down by variation, segment and device

✦Revenue attribution from a recommendation click within a configurable window

✦Comparison of recommendation strategies by RPV

Explore the platform →