Conversion Rate Calculator

Background.

A conversion rate is the share of your traffic that did the thing you care about — bought, subscribed, booked a demo, filled in the form. Divide the number that converted by the number you counted, multiply by 100, and you have it. The arithmetic is trivial. Everything that makes the number hard is hidden in the two words "number you counted", and that is what this calculator makes explicit instead of deciding on your behalf.

There is no single conversion rate. Google Ads defines its conversion-rate column as "the average number of conversions per ad interaction", so its denominator is clicks. Google Analytics 4 reports against sessions, and a session is simply a visit that ends after 30 minutes of inactivity — one determined shopper can generate four of them in an evening. Adobe Analytics points its users at Orders per visit, orders divided by visits, for the same job. Divide by unique people instead and the same week of data yields a third, higher number. So pick the basis first, label it wherever the number is published, and never compare a session rate with a user rate and call the difference a result.

The second thing this page insists on is that your conversion rate is an estimate, not a measurement. It comes from a sample of visits, and small samples wobble hard. Three purchases from 40 sessions is 7.50%, but the 95% confidence interval around it runs from 2.58% to 19.86% — on that evidence you genuinely cannot tell a 3% site from a 19% site. Feed in a month of real traffic and the interval collapses: 1,248 converting sessions out of 48,000 is 2.60%, with an interval of 2.4614% to 2.7462%. Every result here carries its interval beside it, because a conversion rate quoted alone invites you to celebrate noise.

The interval is the Wilson score interval rather than the textbook estimate-plus-or-minus form. Brown, Cai and DasGupta showed in Statistical Science that the familiar Wald interval has erratic coverage exactly where marketing data lives — modest counts and proportions close to zero — and recommended Wilson for small samples. At an observed rate of 0% Wilson still returns a sensible upper bound, while the textbook formula returns an interval of zero width, which is plainly nonsense.

Three modes are available. Solve for the rate when you have both counts. Solve for conversions when you want to know what a planned traffic number should produce at a rate you already trust. Solve for traffic when you are working backwards from a target: 100 sales at a 2.5% conversion rate needs 4,000 sessions, and learning that in the planning meeting is much cheaper than discovering it in week three of the campaign.

One deliberate limit, stated here rather than buried in an accordion. This page models the share of traffic that converted, so conversions can never outnumber traffic and the rate can never exceed 100%. Ad platforms do sometimes report rates above 100% — Google Ads documents why: under "every conversion" counting, a single click can be credited with several conversions. That figure is a conversions-per-click ratio rather than a proportion, and no binomial confidence interval applies to it, so this calculator will not pretend otherwise. It also takes your counts at face value: it cannot audit your attribution window, your bot filtering, or whether the same person converted twice from two devices.

What is conversion rate calculator?

Conversion rate is the proportion of a defined population of traffic that completed a defined action, expressed as a percentage. Three things have to be pinned down before the number means anything: the action that counts as a conversion, the window over which both counts are taken, and — the one people forget — the population in the denominator. Sessions, unique users and ad clicks are three different populations and they produce three different rates from the same underlying behaviour. The session-based rate is always the lower of the first two, because a site cannot have fewer sessions than users and one person tends to visit several times before buying; where a click-based rate lands depends on how many of your clicks became recorded sessions at all, so it is not reliably above or below either. Google Analytics 4 defaults to a session-based rate; Google Ads reports per ad interaction; Adobe Analytics offers Orders per visit as its equivalent. None of them is wrong, and none of them is comparable to the others.

How to use this calculator.

1. 01Choose what you are solving for. "Conversion rate" needs both counts; "Conversions expected" needs traffic and a rate; "Traffic needed" needs a conversion target and a rate.
2. 02Choose the denominator — sessions, users or clicks. This is the single most consequential choice on the page, because it decides which metric you have computed, not just how it is labelled.
3. 03Enter the converting units, not the conversion events. If one visit produced three separate purchases, that is one converting session and three orders; use one here and keep the three for your average order value.
4. 04Enter the total traffic on the same basis and over exactly the same date range. Mixing a 30-day conversion count with a 7-day session count is the most common way these numbers go wrong.
5. 05Read the headline rate together with the 95% interval underneath it. If the interval is wider than the difference you are trying to detect, you do not yet have enough traffic to call it.
6. 06Check the "Which rate this is" line before pasting the result anywhere. It names the denominator so the number cannot travel without it.

The formula.

The core identity is one division: conversion rate = converting units ÷ total units × 100. Solving for conversions gives conversions = traffic × rate ÷ 100, and solving for traffic gives traffic = conversions × 100 ÷ rate. All three modes end up holding the same triple of numbers, so the outputs agree whichever one you asked for.

The confidence interval is the 95% Wilson score interval for a binomial proportion. With p̂ = x ÷ n the observed proportion, n the traffic, and z = 1.959963984540054 (the 97.5th percentile of the standard normal distribution, 1.96 to three significant figures), the interval is centred on (p̂ + z²/2n) ÷ (1 + z²/n) with half-width z × √(p̂(1−p̂)/n + z²/4n²) ÷ (1 + z²/n). Unlike the textbook Wald interval it never runs below 0% or above 100%, and it behaves sensibly at the edges: with zero conversions the lower bound is exactly 0%, and when every unit converts the upper bound is exactly 100%.

Rounding stage matters and is worth stating plainly: nothing is rounded on the way through. Every intermediate value — the proportion, z², the variance term, the square root — stays at full decimal precision, and rounding happens exactly once, at the point the results are returned, to ten decimal places. There are no bands or thresholds in this calculation, so no result sits on a rounding cliff; the only hard edges are structural ones, at 0% and at 100%, and both are tested at, immediately below and immediately above the boundary.

A worked example.

Frequently asked questions.

References& sources.

1. [1]Google Ads Help, "Conversion rate: Definition" (current edition, retrieved 2026-07-29). Defines conversion rate as "the average number of conversions per ad interaction, shown as a percentage", gives the worked example "if you had 50 conversions from 1,000 interactions, your conversion rate would be 5%", and states that the rate "might be over 100%" when more than one conversion action is tracked or the "every conversion" counting option is used. Source of the click-based denominator and of the above-100% scope limit stated in the intro. Publicly accessible.
2. [2]Google Analytics Help, "[GA4] About Analytics sessions" (current edition, retrieved 2026-07-29). Defines a session as "a period of time during which a user interacts with your website or app" that "ends or times out after 30 minutes of user inactivity", and describes the session_start event. Source of the session-based denominator and of the statement that one person can generate several sessions. Publicly accessible.
3. [3]Adobe Analytics documentation (Adobe Experience League), "Orders per visit" (current edition, retrieved 2026-07-29). Defines the metric as "the ratio of orders that visitors made over the total number of visits to your site", calculated as Orders divided by Visits, and states it can be used to "understand conversion rate for your site". Consulted as an independent second authority: it agrees that conversion rate is defined by its denominator, and independently chooses visits rather than clicks. Publicly accessible.
4. [4]Wilson, E. B. (1927). "Probable Inference, the Law of Succession, and Statistical Inference." Journal of the American Statistical Association, 22(158), 209–212. DOI 10.1080/01621459.1927.10502953. The original derivation of the score interval used for every interval on this page. Bibliographic reference only — the publisher's copy returned HTTP 403 to an unauthenticated request on 2026-07-29; the JSTOR record below is the stable identifier and is paywalled.
5. [5]Brown, L. D., Cai, T. T. & DasGupta, A. (2001). "Interval Estimation for a Binomial Proportion." Statistical Science, 16(2), 101–133. DOI 10.1214/ss/1009213286 (verified open access on Project Euclid, retrieved 2026-07-29). States that "the chaotic coverage properties of the Wald interval are far more persistent than is appreciated" and that "common textbook prescriptions regarding its safety are misleading", recommending the Wilson interval for small sample sizes — the reason this page reports Wilson rather than the textbook interval.
6. [6]Kohavi, R., Tang, D. & Xu, Y. (2020). Trustworthy Online Controlled Experiments: A Practical Guide to A/B Testing. Cambridge University Press, ISBN 978-1-108-72426-5. The standard practitioner reference on why observed conversion-rate differences on small samples are usually noise, and on the discipline needed before declaring a winner. Print reference, bibliographic entry only — the linked publisher page is the catalogue record, not the text.