Skip to content
See guides

Conversion Tools · Measure

A/B Test Calculator

An A/B test compares two variants: conversion rates, lift, and whether you likely stopped too early. Write the rule before you launch. Do not call a winner on day one.

The tool

Try it here

Use this with real visitor and conversion counts per variant. Not a screenshot of a “confidence” badge you cannot define.

Did B beat A, or did you stop too early?

Variant A

Variant B

A/B read

This is a two-proportion check, not a full experiment platform. Do not call a 2-day test “significant.”

Why it helps

Why you need an A/B Test Calculator

Lift = (B rate − A rate) ÷ A rate. A 0.2 point change on 80 visitors is noise.

Peeking every morning and stopping at the first green day inflates false winners.

If traffic is thin, you do not have a test. Wait, or change one thing without pretending it was decided.

LIFT IS NOT A DECISION BY ITSELFLIFT IS NOT A DECISION BY ITSELFA80 / 2,000B104 / 2,000Waitsample sizeDay-one winners reverse. Need conversions, not vibes.This tool is a two-proportion check. It is not a testing platform.
Ship the winner. Then keep measuring. Tests rot.

When

When to use it

  • Before you stop a test - check the counts you actually have.
  • When someone wants to ship B because “it looks nicer.”
  • When two headlines already passed the headline analyzer and you need a sample check.

How

How to use it

  1. 1Enter visitors and conversions for A and for B. Same conversion, same dates.
  2. 2Read both rates and the lift. Note which way it points.
  3. 3If either side has a handful of conversions, you are not done.
  4. 4Keep the pre-written rule: minimum sample, primary metric, what “win” means.

Example

Example: H1 test on Acme /demo

Traffic from google-in-demo-exact-2026q3. A: “Book a 30-minute product demo.” B: “Product demo for marketing teams.” Conversion = demo booked.

You put in

A visitors
1,000
A conversions
40
B visitors
1,000
B conversions
52

You get

A rate = 40 / 1000 = 4.0%
B rate = 52 / 1000 = 5.2%
Lift = (5.2 − 4.0) / 4.0 = 30%

B is ahead on this sample. That is not a license to stop at 200 visitors. Use the rule you wrote before launch. If you peeked daily and stopped at the first green afternoon, run longer.

Result

What this changes for you

You see the lift and the sample. You stop shipping a lucky Tuesday as process.

Common mistakes

  • Stopping when B is ahead by two conversions.
  • Testing five things at once and calling it A/B.
  • Different traffic sources on A and B (brand on A, prospect on B).
  • Using pageviews as the primary metric when the offer is a demo booking.

FAQ

A/B Test Calculator FAQ

What does an A/B test calculator show?

Conversion rates for A and B, the lift, and a reminder that small samples lie. You still need a stopping rule you wrote before launch.

When should I stop using an A/B test calculator?

When you hit the sample and window you wrote down, not when B looks nicer. Peeking every morning is how false winners ship.

Can an A/B test calculator work with uneven traffic?

Enter the real visitors and conversions per variant. Uneven splits are fine if assignment was random. Do not compare brand search to cold social as A vs B.

Does an A/B test calculator prove B is better?

It shows the rates you entered. It does not replace a pre-registered rule or a proper significance method. Thin traffic is still anecdotes.

Next

Related tools

All Conversion Tools tools