Conversion rate is the metric most teams put on the wall. It is easy to understand, easy to chart, and easy to celebrate when it ticks up. The problem is that a rising conversion rate can sit alongside falling revenue, and a falling rate can hide a healthier business. Revenue per visitor (RPV) ties optimisation directly to the number that pays salaries, and it changes which experiments you choose to run.
What revenue per visitor actually measures
Revenue per visitor is total revenue divided by total sessions (or visitors, if you prefer a deduplicated view) over a period. It folds three levers into one figure:
- Conversion rate — the share of visitors who buy.
- Average order value (AOV) — how much each buyer spends.
- Traffic quality — whether the people arriving are the right ones.
Because RPV moves when any of these change, it resists the gaming that single-metric optimisation invites. You cannot push it up by stripping margin out of the basket, and you cannot flatter it by chasing cheap traffic that never converts.
Why conversion rate misleads
Conversion rate treats every order as equal. A £15 order and a £400 order count the same. That assumption breaks in several common situations.
The discount trap
Aggressive promotions almost always lift conversion rate. More visitors buy because the offer is good. But if the discount erodes AOV faster than it lifts conversion, RPV falls and so does gross profit. Teams optimising on conversion rate alone will declare the promotion a win and repeat it.
The traffic-mix problem
Suppose a paid campaign sends a flood of bargain-hunting visitors. Conversion rate may dip simply because the denominator grew with low-intent sessions, even though absolute revenue rose. Conversely, pausing a broad campaign can lift conversion rate while shrinking the business. RPV, read alongside traffic volume, tells the real story.
The high-consideration category
In categories where customers research over several visits — furniture, electronics, anything with a meaningful price tag — per-session conversion rate is structurally low and noisy. RPV smooths this because it captures value across the mix of browsing and buying sessions. We cover the buying-journey angle in guided selling for high-consideration purchases.
When conversion rate still earns its place
RPV is not a replacement for everything. Conversion rate remains the right lens for a single, isolated step — a checkout field, a payment method, a shipping-options screen — where AOV is unlikely to move and you want a clean read on friction. Use conversion rate for micro-optimisation of a funnel stage; use RPV for decisions that touch merchandising, pricing, recommendations, or traffic. A full breakdown of stage-level diagnostics lives in our eCommerce funnel analysis guide.
How to put RPV to work
1. Segment before you average
A single site-wide RPV hides more than it reveals. Break it down by:
- Channel — paid search, organic, email, social, direct.
- Device — mobile RPV is almost always lower; the question is the gap and whether it is closing.
- New vs returning — returning visitors usually carry far higher RPV.
- Landing category — which entry points produce value, not just clicks.
These cuts turn RPV from a scoreboard into a map of where money is won and lost.
2. Pair it with margin
RPV uses revenue, not profit. If your categories vary widely in margin, track a gross-profit-per-visitor variant in parallel. An experiment can lift RPV by pushing low-margin lines; the profit view keeps you honest. This is also where a clean data layer matters — see eCommerce data foundations.
3. Make it the primary experiment metric
When you run an A/B test, declare RPV the primary success metric and conversion rate plus AOV as secondary diagnostics. This single change reframes how you read results:
- Conversion up, AOV down, RPV flat — no real win, investigate the trade-off.
- Conversion flat, AOV up, RPV up — a genuine value gain from better merchandising or recommendations.
- Conversion up and RPV up — the outcome you want; ship it.
4. Watch the statistics
RPV is noisier than conversion rate because order values are skewed — a handful of large baskets can swing the average. Two practical defences:
- Run tests longer. Higher variance needs more data to reach confidence. In our experience, RPV tests typically need to run a week or two beyond what a conversion-rate test of the same traffic would require.
- Consider trimming or modelling outliers. Capping extreme order values, or reporting median order value alongside the mean, stops one B2B-sized basket from declaring a winner.
A worked example
Imagine two checkout variants tested over a month:
- Variant A: conversion rate 3.2%, AOV £58 → RPV £1.86
- Variant B: conversion rate 2.9%, AOV £71 → RPV £2.06
A conversion-rate-led team ships Variant A and quietly loses money. An RPV-led team ships Variant B and grows revenue by roughly 11% per visitor at the same traffic. The variants might differ only in whether the basket page nudges complementary items — exactly the kind of change a conversion-only metric is blind to.
Common pitfalls
- Reporting RPV without traffic volume. RPV per visitor and total sessions must be read together; a rising RPV on collapsing traffic is not growth.
- Ignoring refunds and returns. Use net revenue where you can, especially in apparel. A returns-heavy variant can post strong gross RPV and weak reality. We discuss this trade-off in reducing returns with guided selling.
- Mixing visitor and session definitions. Pick one denominator and apply it consistently across every report.
- Treating RPV as a vanity headline. Its value is in the segmented, margin-aware version, not a single dashboard number.
Where RPV points your roadmap
Once RPV is your North Star, your priorities shift from squeezing the funnel to raising the value of each visit. That tends to push effort toward personalised recommendations, smarter search, and guided selling — interventions that lift AOV and match without sacrificing conversion. Our conversion optimisation service is built around RPV as the decision metric, and our AI search and recommendations service targets the AOV side of the equation directly.
The shift is conceptually small but practically large: stop asking “did more people buy?” and start asking “did each visit become worth more?”
If you want a second pair of eyes on which experiments will move RPV rather than just conversion rate, get in touch and we will talk through your numbers.