Conversion rate calculator
Use this conversion rate calculator with a confidence interval. Compare randomized variants, check significance and estimate visits for your target lift.
Free, no sign-up. Runs in your browser, with no AI calls.
Conversion rate divides converted visitors by total visitors and multiplies by 100. A Wilson interval shows sampling uncertainty, including zero or full conversion. Compare two variants with a significance check, then estimate visits needed for a target lift. Count one outcome per visitor and plan the test before collecting data.
Your inputs
Use your own campaign data. Business fields start empty. Inputs can be up to one trillion; use this language’s decimal separators. Currency follows the page language, without conversion.
Compare variant B
Plan visits for a target lift
Statistical defaults: 95% confidence and 80% power, following CXL. These are adjustable planning settings, not business benchmarks. Sources checked on 2026-10-12.
Choose a change above zero and up to 100 percent. The table uses your value below and above the current input, holding other inputs constant. These are scenarios, not forecasts.
Your result
Enter your inputs to calculate. No business assumptions are prefilled.
Method sources checked on 2026-10-12.
This calculator models one binary outcome per visitor. Google Ads can report multiple conversions per interaction; those event totals need a different model.
How to use this tool
- Count independent visitors
- Use whole visitor counts and no more than one conversion per visitor. A visitor belongs to only one randomly assigned variant. Returning sessions, overlapping audiences and changing traffic sources can invalidate the comparison.
- Read uncertainty for one sample
- The 95 percent Wilson interval uses the observed conversion share and visitor count. It remains nonzero in width when no visitors or all visitors convert. It describes random sampling uncertainty, not tracking errors, biased traffic or future business performance.
- Read the difference and interval
- Difference = variant B rate minus variant A rate, in percentage points. The test uses a pooled proportion for its standard error; the 95 percent Wald interval uses separate sample rates. This calculator requires at least five successes and failures in each variant for normal inference.
- Plan visits for the lift that matters
- Choose relative lift or an absolute percentage point increase from your observed baseline. The equal-split, two-sided normal approximation estimates visits per variant and total visits for your confidence and power settings. It has no continuity correction, omits the far tail and can be inaccurate for rare outcomes. It cannot guarantee significance.
- Finish the planned test
- Choose the primary outcome, sample size and duration before starting. Checking repeatedly and stopping when p drops below 0.05 raises false positives. This calculator does not adjust for sequential monitoring, many variants or multiple outcomes.
Questions before you start
Where do your next customers find you?
Run a free AI visibility scan to see how assistants describe your business. Compare monitoring plans if you want to follow those answers over time.