Conversion Rate Calculator

The Conversion Rate Calculator turns a count of visitors and conversions into a rate plus a statistically rigorous confidence interval, margin of error, and reliability assessment. Use it for landing pages, sign-up funnels, ad campaigns, lead-magnet pages, checkout flows, and any A/B test where the metric is "did the user complete the action or not?". The tool offers three confidence-interval methods — Wilson score (the recommended default), Wald (the classical normal approximation), and Agresti-Coull (a conservative middle-ground) — and reports the margin of error, the sample-adequacy verdict, an animated funnel visualisation, an industry benchmark band, and the additional traffic you would need to tighten the estimate. Whether you are evaluating a single funnel step or planning the sample size for a future test, the result you get here is more precise than the typical "conversions ÷ visitors" calculator.

How to Use

1. Enter the total number of visitors — sessions, trials, or impressions that reached the step you are measuring.
2. Enter the number of conversions — visitors that completed the target action. Conversions cannot exceed visitors.
3. Pick a confidence level. 95% is the industry standard; use 99% for high-impact decisions and 90% only for early exploration.
4. Pick a confidence-interval method. Wilson score is recommended for all sample sizes; Wald is the textbook formula; Agresti-Coull is a slightly conservative alternative.
5. Click Calculate Conversion Rate to see the rate, confidence interval, margin of error, sample-adequacy verdict, funnel visualisation, method comparison, and a step-by-step math breakdown.

Formula Used

What Makes This Conversion Rate Calculator Different

• Wilson score by default — most online calculators only offer Wald, which gives impossible intervals (negative lower bound, upper bound > 100%) for small samples or extreme rates. Wilson is correct everywhere and is what professional statisticians recommend.
• Live preview before you submit — type any number and the rate, confidence interval, and adequacy badge update in real time, with no full-page reload.
• Animated funnel visualisation — see the shape of your funnel, not just a single number.
• Sample-adequacy traffic light — green / amber / red badge based on the n·p̂ ≥ 10 rule of thumb, so you immediately know if the estimate is trustworthy.
• Side-by-side method comparison — Wilson, Wald, and Agresti-Coull intervals in the same table. The one you selected is highlighted; the others show how the choice would shift the answer.
• Industry benchmark band — a six-band scale (Very low → Elite) puts your rate in context.
• Planning helpers — exactly how many more visitors you would need to halve the margin of error or hit ±1.00 / ±0.50 percentage point precision.
• Step-by-step math — every calculation broken down line by line so you can verify and learn.

Reading the Sample-Adequacy Verdict

• Green — reliable. Both n·p̂ and n·(1 − p̂) are at least 10. The normal approximation holds and the confidence interval is trustworthy for decision-making.
• Amber — borderline. One of n·p̂ or n·(1 − p̂) is below 10 but at least 5. Prefer the Wilson or Agresti-Coull interval over Wald, and collect more data before committing to a decision.
• Red — small sample. Either n·p̂ or n·(1 − p̂) is below 5. Treat the rate as a rough indicator only and collect substantially more data.

Typical Conversion Rate Benchmarks

Channel / step	Typical range	Notes
E-commerce site-wide	2% – 3%	Mature stores cluster around 2.5%; mobile is often lower than desktop.
E-commerce add-to-cart → checkout	20% – 35%	Funnel-step rate, not site-wide.
Landing page (paid traffic)	3% – 10%	Strong copy and a tight offer lift this above the baseline.
SaaS free-trial signup	5% – 12%	Lower friction signups land higher; credit-card-required is lower.
SaaS free → paid	2% – 5%	Trial-to-paid conversion in the bottom half of the funnel.
B2B lead generation form	1% – 5%	Longer forms drop sharply; gated content lifts the rate.
Display ad click-through	0.05% – 1%	Highly creative- and placement-dependent.
Search ad click-through	2% – 6%	Brand searches can exceed 10%.
Email open → click	2% – 10%	Of opened emails; segmentation lifts the rate.

Why a Confidence Interval Matters

A conversion rate measured on a finite sample of visitors is only an estimate of the true underlying rate. If you flip the same biased coin 100 times you might get 47 heads; flip it again and get 53. The same applies to a funnel — the day-to-day number bounces around the true rate because of pure randomness. The confidence interval tells you the range the true rate plausibly lives in, so you can avoid two classic mistakes: declaring a winner from noise, and concluding "nothing happened" when the test was simply too small to detect the change.

Wilson vs Wald vs Agresti-Coull

The three intervals answer the same question but compute it differently:

• Wald is the textbook formula p̂ ± z·√[p̂(1−p̂)/n]. Simple, fast, but breaks down for small samples or rates near 0% / 100% — it can produce negative lower bounds or upper bounds above 100%.
• Wilson is the score interval. It is the inverse of the score test and is the recommended default because it stays inside [0, 100%] at every sample size, has near-nominal coverage for any rate, and matches statistical research best practice.
• Agresti-Coull adds z² pseudo-observations (half conversions, half non-conversions) and then applies the Wald formula on the adjusted counts. It is a slightly conservative compromise that is easy to explain.

For practical work, default to Wilson. Use Wald only for very large samples where the rate is far from 0 or 100%. Use Agresti-Coull when you want a slightly wider, more conservative interval that is straightforward to derive.

Sample Size Planning for a Target Margin of Error

If your goal is a confidence interval of ±E percentage points around the conversion rate, the sample size you need is:

Common Pitfalls When Measuring Conversion Rates

• Mixing visitor scopes — measuring sessions for one variant and unique users for the other inflates one rate. Pick one scope and apply it consistently.
• Bot traffic — uncleaned bot impressions in the visitor count deflate the rate. Filter known crawlers and headless traffic before computing.
• Stopping early — checking results daily and stopping at the first significant lift inflates false positives. Decide the target sample size in advance.
• Comparing rates across different time windows — weekend vs weekday, peak-season vs off-season, or pre-launch vs post-launch mix shifts the baseline. Compare like with like.
• Ignoring segment heterogeneity — a 4% blended rate can hide a 2% mobile rate and an 8% desktop rate. Segment large funnels before drawing conclusions.
• Counting one user multiple times — if a visitor returns three times before converting, decide whether to count one conversion or three. Inconsistency biases the rate.
• Tracking gaps — a missing pixel on the success page silently lowers the conversion count and the rate. Validate the funnel end-to-end before trusting the number.

Connecting to A/B Testing

A conversion rate is the building block of any A/B test. To test whether two rates differ significantly you compare their confidence intervals (or, more precisely, run a two-proportion z-test on the underlying counts). The A/B Test Significance Calculator handles that comparison directly. The Confidence Interval for Proportion Calculator focuses purely on the interval itself. Together these three tools cover most funnel-analytics needs.

FAQ

What is a conversion rate?

A conversion rate is the percentage of visitors that complete a specific target action — buying, signing up, clicking through, downloading, or generating a lead. It equals conversions divided by visitors, expressed as a percentage. Anything that can be expressed as "did the user do X or not" is a conversion event.

Why does the conversion rate need a confidence interval?

A conversion rate measured on a finite sample of visitors is only an estimate of the true underlying rate. A confidence interval tells you the plausible range of the true rate given the data, which is essential for A/B testing, funnel analysis, and any decision that hinges on whether one rate is reliably better than another.

Which confidence interval method should I use?

Use the Wilson score interval as the default — it is accurate at any sample size and never produces impossible values below 0% or above 100%. Use Wald only when the sample is large and the rate is far from 0% or 100%. Use Agresti-Coull when you want a slightly conservative, easy-to-explain alternative.

How is the margin of error calculated?

The margin of error is half the width of the confidence interval. For the Wald interval it equals z · √[p̂(1 − p̂) / n], where z is the standard normal quantile for the chosen confidence level (1.96 for 95%). For Wilson and Agresti-Coull the formulas are slightly different but the interpretation is the same: how far the upper and lower bounds sit from the point estimate.

How many visitors do I need for a reliable conversion rate?

As a rule of thumb the sample is reliable when both n · p̂ and n · (1 − p̂) are at least 10. For a 1% conversion rate that means roughly 1000 visitors; for a 5% rate, 200 visitors. The calculator shows the exact sample size you need to reach a target margin of error such as ±1%.

How do I halve the margin of error?

Cutting the margin of error in half requires roughly four times the visitors, because the margin shrinks with the square root of the sample size. The "How Many More Visitors" panel reports the exact number based on your current data.

Is the conversion rate the same as click-through rate?

Mathematically yes — both are "events ÷ opportunities". Click-through rate is the fraction of impressions that produced a click; conversion rate is the fraction of visitors that produced a conversion event. The math and statistical treatment are identical, so this calculator works for both.

What if my conversion rate is exactly 0% or 100%?

The Wald interval collapses to zero width at the boundaries, which is misleading — a single coin flip landing heads does not prove the coin always lands heads. The Wilson interval handles boundaries correctly and gives a non-zero range. Always prefer Wilson at the extremes.