Free A/B Test Calculator – Know When You Have a Winner
Enter your visitors and conversions. Get the verdict in plain English, with the p-value, uplift and confidence interval behind it. Planning a test? Work out the sample size first.
Two-sided asks "is there any difference?". One-sided only asks "is the variant different in the direction I expected?" and needs less data, so only pick it before you start the test.
Your results appear here
Enter visitors and conversions for A and at least one variant. Results update as you type.
How to Read Your Results
Four numbers decide whether you ship the change. Each one answers a different question.
p-value: Is It Real or Luck?
The p-value is the chance of seeing a gap this big if both versions were identical. Under 0.05 at 95% confidence, you can act on it. Above that, keep collecting data.
Confidence: How Safe Is the Call?
Confidence is 1 minus the p-value. A p-value of 0.04 reads as 96% confident this is not chance. It tells you how safe it is to ship the change.
Uplift: How Much Better?
Relative uplift compares the variant to the control. A move from 5% to 6% is a 20% uplift. It shows how big the win is, not whether it is real.
Interval: What Range to Expect
The confidence interval is the range the true difference likely sits in. If it crosses zero, the variant might not be better at all. A narrow interval means a sharper answer.
Common A/B Testing Mistakes
Bad habits turn a clean test into a coin flip. Avoid these four and your wins will hold up.
Peeking
Checking results every day and stopping when the p-value dips under 0.05 inflates false winners. Decide the sample size first, then look.
Stopping Early
A big early lead often shrinks as more visitors arrive. Let the test reach its planned sample size and at least one full business week.
Too Many Variants
Each extra variant is another chance for luck to look like a win, and it splits your traffic. Test fewer ideas, each one bolder.
Tiny Samples
With only a handful of conversions, one lucky visitor swings the result. Aim for roughly 100 conversions per version before you trust a read.
What to Test on Forms, Popups and Landing Pages
Start with the changes visitors notice first. Change one thing per test so you know what moved the number. Browse ready-made widget templates to build your variants.
Headline
Test a benefit-led headline against a feature-led one on your landing page. The headline is the first thing visitors judge.
CTA Copy
Compare a specific button label such as "Get my quote" with a generic one such as "Submit". Keep the color and position the same.
Form Length
Try a form with only email against one with name, email and company. Fewer fields can mean more leads, but check lead quality too.
Quiz vs Static Form
Put a multi-step quiz up against a plain form for the same offer. See which one captures more leads and which leads convert later.
Popup Timing and Offer
Test an exit-intent popup against a timed one, or a discount against a free guide. One change per test keeps the result clear.
Landing Page Layout
Compare social proof above the fold with social proof lower down, or a short page with a long one.
Run Your Next Test Without a Developer.
Embeddable has A/B testing built in. Test variants of your widgets and landing pages, then see the results in the same place as your analytics and your lead database. No third-party CRM needed.
A/B testing is part of the paid plans, and Free is there to try the builder. See pricing for the limits.
Start freeFAQ
What does statistically significant mean in an A/B test?
It means the gap between your versions is bigger than random noise would usually produce. At 95% confidence, the p-value has to be under 0.05. Significant does not mean important. Check the uplift and the interval too.
How many visitors do I need for an A/B test?
It depends on your baseline conversion rate and the smallest lift you care about. Lower baselines and smaller lifts need more visitors. Open the Sample Size and Duration tab, enter both numbers, and the calculator shows the visitors per variant.
Should I use a one-sided or two-sided test?
Use two-sided unless you have a strong reason not to. It catches both a win and a loss. One-sided needs less data but only looks in one direction, so choose it before the test starts, never after you see the numbers.
Why does the calculator use a Bonferroni correction?
Every extra variant gives chance another shot at looking like a winner. With more than one variant, the calculator divides your significance threshold by the number of variants. A variant has to clear a higher bar, so false winners are less likely.
How long should I run an A/B test?
Run it until you reach the sample size you planned, and for at least one full business week. Visitors behave differently on weekdays and weekends. Use the duration estimate on the Sample Size tab as your starting point.
Is my data sent anywhere?
No. All the maths runs in your browser, and the numbers you type are not sent to a server. If you share a link, the numbers travel inside the link itself, so share it only with people who should see them.