How is conversion lift calculated?

Subtract the baseline rate from the variant rate, divide by the baseline, and multiply by 100. From 3.0% to 3.6% the lift is 20%.

What is the difference between relative and absolute lift?

Absolute lift is the point difference (0.6 points); relative lift expresses it against the baseline (20%). Relative lift is comparable across tests, and is what this calculator reports.

Does a big lift mean the variant won?

No. Small samples produce large apparent lifts by chance. Confirm statistical significance and adequate sample size before treating any lift as real.

Why does measured lift shrink after shipping?

Test conditions rarely persist: novelty fades, traffic mix changes, and winning variants are often measured at their most favorable moment. Expect regression toward the baseline.

Calculator · 114

Conversion Lift AB Test Calculator

Measure the relative lift a variant produced — and decide what that lift is actually worth before shipping it.

Baseline conversion rate

Variant conversion rate

Monthly visitors

visitors

Conversion lift

—

Average

Scenario lens Current · Benchmark · Optimized

Value / conversion

Leverage

Formula

Lift = (Variant rate − Baseline rate) / Baseline rate × 100

Understanding conversion lift in A/B tests

Reference material — the calculator above stays the primary tool.

What lift measures

Conversion lift is the relative improvement of a variant over its baseline — the difference between the two rates, expressed as a percentage of the baseline. Relative framing is what makes lift comparable across tests with very different starting rates.

A move from 3.0% to 3.6% is a 20% lift, not a 0.6% one. Confusing relative lift with the absolute point difference is the most common way test results get overstated.

How to read your result

The result is labelled against an orientation benchmark, and the leverage line prices the lift in monthly conversions and revenue at your traffic and conversion value.

Low — a small relative gain, easily noise. Average — a meaningful lift worth confirming. Strong — a large lift, if it holds up statistically.

Lift in perspective

Most real test wins are modest. Treat these as orientation, not targets.

Context	Typical median
Under 5%	Often indistinguishable from noise
5–15%	Typical real win
15–30%	Strong result
Above 50%	Verify before believing

Lift is not proof

An observed lift means nothing without statistical confidence behind it — small samples produce large apparent lifts routinely. Confirm significance and adequate sample before shipping, and expect measured lift to shrink in production.

Lift in context

Read this alongside the A/B significance and sample size tools, which the related tools cover. Lift tells you how big the effect looks; significance tells you whether to believe it.