Conversions vs. Revenue

The metric you choose changes how quickly an experiment can give you a trustworthy answer.

Lesson 1 of 2

Same experiment, different metric

One change lifts conversion and revenue by the same relative amount. Give both the same visitors and compare — then use Orders to see what a few outsized orders cost you.

Traffic Volume Total visitors

Effect (MDE) Rel. %

Reset

Binary metric

Conversion

Base Conversion Abs. %

Power—

Exaggeration—

Sample @ 80%—

Continuous metric

Average revenue per visitor

Buyers

—

Mean Per visitor ($)

Std Dev 500% of mean

Orders

Typical order mix

Power—

Exaggeration—

Sample @ 80%—

Lesson 1 of 2

The exception: when revenue wins

That comparison assumed one thing — that your change moves both metrics by the same amount. Plenty of changes don't. An upsell, a bundle, or a free-shipping threshold barely touches whether someone buys, while moving how much they spend. For those, conversion is close to a null metric, and revenue is the only one that can see the effect.

Pick a type of change below — or type your own lifts — and watch which metric wins.

Conversion Lift Rel. %

Revenue Lift Rel. %

Basket effect

—

Implied AOV change

A clearer buy button or a trust badge — it changes who buys, and each buyer spends about the same.

At these settings

—

Binary metric

Conversion

Power—

Noise (CV)—Relative spread

Continuous metric

Average revenue per visitor

Power—

Noise (CV)—Relative spread

The break-even ratio

Revenue becomes the better-powered metric once its lift is larger than the conversion lift by a fixed factor — the ratio of the two metrics' noise levels:

revenue lift ÷ conversion lift > CV_revenue ÷ CV_conversion = —

The counterintuitive part: revenue is easiest to justify at the top of the funnel. On a low-converting page the zeros dominate both metrics, so revenue adds little extra noise. Deep in the funnel, conversion becomes nearly noiseless while revenue still carries the full spread of order values — so revenue has to work much harder to compete.

Changes that move basket size, not buy rate

Upsells and cross-sells — offered after the decision to buy is made.
Bundles and volume discounts — same buyers, bigger carts.
Free-shipping thresholds — "add $10 more to qualify" nudges basket size directly.
Price and SKU-mix changes — may even lower conversion while raising revenue.
Subscription or quantity defaults — a large revenue effect with little change in who buys.

One caution. A genuine basket-size win and a single outlier order look identical in the topline. Before trusting a revenue-only result, check whether a handful of large orders are carrying it — and consider capping extreme values before you analyse.

Lesson 2 of 2