Free tools / no signup needed
Conversion Rate Calculator
Compute a conversion rate, compare two periods or variants, and check whether the difference is statistically meaningful, not just noise.
Enter visitors and conversions for A; add B to compare periods or test variants.
Conversion rate, properly defined
Conversion rate = conversions ÷ opportunities. The subtle part is choosing the denominator: per session answers "how well does a visit convert?" while per visitor answers "how many people eventually convert?". Both are valid; mixing them across reports is how teams end up arguing about numbers that are both correct. Pick one, name it, and stay consistent.
Why significance matters
Going from 40 to 48 conversions looks like +20%, but with a few hundred visitors that difference is well within random variation. The comparison mode here runs a two-proportion z-test: it estimates how likely a gap at least this large would be if nothing had actually changed (the p-value). Below 0.05 is the conventional bar for "probably real". Small samples almost never clear it, which is the honest answer: keep collecting data.
And when a change is real, the next question is which segment caused it: one device, one traffic source, one landing page? That decomposition is what FocusStat's "Why did this change?" analysis automates.
Frequently asked questions
What is a good conversion rate?
It depends entirely on what converts. Newsletter signups from a blog might run 1-3%, e-commerce purchases 2-4%, SaaS trials from a pricing page 5-15%. Compare against your own history rather than industry averages, which mix incomparable funnels.
Should I measure conversion per session or per visitor?
Per session for optimizing a specific flow (each visit is a fresh chance). Per visitor for judging overall business performance (people often need several visits). State which one you use; the two can differ by 2x or more.
How much traffic do I need for a reliable A/B test?
As a rough rule, to detect a 20% relative improvement on a 5% baseline rate you need around 5,000-8,000 visitors per variant. Smaller effects or smaller baselines need much more. This calculator's p-value tells you when you're there.
What does the p-value mean?
The probability of seeing a difference at least this large purely by chance, assuming there is no real difference. p = 0.03 means such a gap would appear in about 3% of experiments where nothing changed. It is not the probability that B beats A.