Enter your baseline conversion rate and the smallest relative lift worth detecting. Get the visitors each variant needs at 95% significance and 80% power.
Answers as you type · Formula shown, not hidden · Free, no sign-up
Your current rate on the control.
Smallest relative lift worth detecting. 20% on a 5% baseline is 1 point absolute.
Two-sided confidence level.
Chance of detecting the effect if it is real.
The calculator gives you the number instantly. Leave your email and we will look at your actual accounts and send back what we would change first.
For an equal traffic split, visitors per variant ≈ 2 × (zα/2 + zβ)² × p(1 − p) / δ². p is the baseline conversion rate, δ is the absolute minimum detectable effect (baseline × relative MDE), zα/2 is 1.96 at 95% two-sided, and zβ is 0.84 at 80% power.
A 5% baseline with a 20% relative MDE means δ = 1 percentage point (5% → 6%). That plan needs about 7,500 visitors per variant — roughly 15,000 total.
Powering for a 2% relative lift on a 5% baseline demands enormous traffic. Power for the smallest lift that changes a decision: a creative swap, a price test, a checkout change. If that lift is unreachable with your traffic in a reasonable window, the test is not worth running yet.
Significance (1 − α) caps false positives; power (1 − β) caps false negatives. 95%/80% is the default pair in most A/B tools. Raising either one increases the sample you need.
Skip the typing. Connect Meta, Google, TikTok or LinkedIn and Sprites reads your actual spend, impressions, clicks and conversions, then shows where they sit against the ranges above.
Put the Sample Size Calculator on your own site or blog. Paste this HTML where you want it to appear. It runs in the reader's browser and needs no account.
<iframe src="https://www.sprites.ai/embed/sample-size-calculator" title="Sample Size Calculator" width="100%" height="620" style="border:0;max-width:720px" loading="lazy"></iframe>
<p style="font-size:13px"><a href="https://www.sprites.ai/tools/sample-size-calculator">Sample Size Calculator</a> by Sprites</p>What people ask about the sample size calculator.
See it on your own ad accounts in a short demo, or try every part free for 7 days.
It depends on baseline rate and the lift you want to detect. At a 5% baseline, a 20% relative MDE, 95% significance and 80% power, plan on about 7,500 visitors per variant.
The smallest change the test is powered to find. A 20% relative MDE on a 5% baseline is a 1 percentage-point absolute change (5% to 6%).
They are the industry default: a 5% false-positive rate and an 80% chance of catching a real effect of the size you named. Stricter settings need more traffic.
Yes. Unequal splits need more total traffic for the same power. Keep the allocation equal unless you have a strong reason not to.
Research, creatives and media buying, run by one agent.