CRO

A/B Test Sample Size & Duration Calculator

Know exactly how many visitors you need and how long to run your test — before you start. Built on two-sample proportion testing with unequal allocation support.

Your traffic

Average monthly visitorsUse a 4–8 week average from your analytics tool. Seasonal spikes/dips will affect real duration — add a buffer for high-variance periods.

Total visitors across both variants combined.

What are you measuring?

Current conversion rateThe baseline rate your page, flow, or campaign currently achieves. Use at least 2–4 weeks of data to reduce noise.

Your control variant's conversion rate today.

Target conversion rateSet this to the minimum improvement that would be meaningful for your business — not the maximum you hope for. Smaller targets need more visitors.

The improvement you're aiming to detect as statistically significant.

Test settings

Testing objective'Better' plans your test to detect improvements above the baseline. 'Better or Equivalent' adds a ROPE threshold — effects smaller than ROPE are treated as practically equivalent, not meaningful.

Confidence levelThe probability of not getting a false positive (declaring a winner when there's no real difference). 95% is the industry standard.

Higher = fewer false winners, but needs more visitors.

Statistical powerThe probability of correctly detecting a real winner if one exists. 80% is standard — it means 1 in 5 real winners may go undetected. Higher power requires more visitors.

Higher = fewer missed winners, more visitors required.

Traffic allocation (Control / Variation)How you split traffic between the two variants. 50/50 is the most efficient — unequal splits require more total visitors to maintain the same statistical power.

50/50 is optimal for detecting effects with the fewest visitors.

Enter your details to get started

Fill in your monthly visitors and metric details on the left — your estimated sample size and campaign duration will appear here instantly.

Quick exampleMonthly visitors: 50,000 · Current CVR: 3% · Target CVR: 3.5%→ ~26,000 visitors per variant · ~32 days at 95% confidence, 80% power

How to Use This Calculator

Run sample size calculations before your test starts — not after. Deciding your required visitors in advance prevents the most common A/B testing mistake: stopping a test early when you see a promising result.

STEP 01

Enter your monthly traffic

Use a 4–8 week average from your analytics tool. The calculator converts this to a daily rate to estimate test duration. Include weekends — traffic mix affects results.

STEP 02

Choose your metric type

Use Conversion Rate for binary outcomes (purchased / didn't purchase). Use Revenue per Visitor when the magnitude of the transaction matters — e.g., average order value changes.

STEP 03

Set baseline and target values

Baseline is your current performance. Target is the minimum improvement worth acting on — not the best case. Smaller targets require more visitors to detect reliably.

STEP 04

Pick your testing objective

"Better" is a one-sided test — useful when you only care whether B outperforms A. "Better or Equivalent" uses a ROPE threshold, ideal for conservative product changes where regression risk is high.

STEP 05

Adjust confidence and power

95% confidence, 80% power is the industry standard. Use 99% / 90% for high-stakes changes like pricing. Higher settings require more visitors — the calculator shows the trade-off in real time.

STEP 06

Run for the full planned duration

Even if you see a significant result early, stopping early inflates false positive rates. Always run for the full estimated duration, and for at least one complete business cycle (7+ days).

Ready to build a testing programme that moves the needle?

Sample size is the starting point. Search Indicators helps you prioritise what to test, set up clean measurement, and connect CRO results to your broader growth picture.

Talk to us