Conversion Rate Calculator
Your conversion rate with the range it could really be, plus revenue per visitor, so you stop comparing two numbers that are inside each other's noise.
Visitors 12,400
Conversions 291
Conversion rate 2.35%
95% range 2.09% to 2.63%
Visitors per conversion 42.6
Average order value $75.00
Revenue $21,825.00
Revenue per visitor $1.76
Target rate 3%
Conversions needed 372
More than you have 81
Revenue that would add $6,075.00
291 out of 12,400 is 2.35%, and on this sample the 95 percent range is
2.09% to 2.63%. Any comparison with another page or another week that
falls inside that range has not told you anything yet.
A margin of half a percentage point at this rate takes about 3,522
visitors, and you have more, which is why the range is this tight.
Halving it again takes four times the traffic, or 49,600 visitors.
Revenue per visitor is $1.76, and it is the number to optimise. A page
that converts at a lower rate with a larger order can beat one that
converts at a higher rate, and a conversion rate on its own cannot see
the difference.
Check the denominator before you trust a rate. Sessions, users and
clicks give three different answers from the same day, and the most
common reason a rate jumps is that somebody changed which one the report
counts.
Output is valid and updates as you type.
Fix the highlighted fields to update the output.
Conversions divided by visitors. The division is not the hard part. The hard part is that the answer is an estimate from a sample, and most of the comparisons people make with it are inside its margin of error.
So this prints the range alongside the rate, and it adds revenue per visitor, because a page that converts less often at a larger order value wins and a rate on its own cannot see that.
How to use
- Put in the visitors, sessions or clicks that had the chance to convert.
- Put in the conversions.
- Optionally add an average order value for revenue per visitor, and a target rate to see what the target needs.
Example
291 conversions from 12,400 visitors at a 75 order value, against a 3 percent target:
Visitors 12,400
Conversions 291
Conversion rate 2.35%
95% range 2.09% to 2.63%
Visitors per conversion 42.6
Average order value $75.00
Revenue $21,825.00
Revenue per visitor $1.76
Target rate 3%
Conversions needed 372
More than you have 81
Revenue that would add $6,075.00The useful sentence is not “we convert at 2.35 percent”. It is “we convert at somewhere between 2.09 and 2.63 percent”. If the other page in the test reads 2.5 percent, you do not yet have a winner.
Pitfalls
A rate without a sample size cannot be compared to anything. Nine conversions from 400 visitors is 2.25 percent and the range runs from about 1.2 to 4.2 percent. Two pages in that state can be ranked in either order by next week’s traffic.
Sessions, users and clicks are three different denominators. The same day gives three different rates. Pick one, write it down, and check it before you conclude that a rate changed: the most common cause of a jump is somebody changing the report.
Rates and revenue move in opposite directions more often than you would think. Adding a discount lifts the conversion rate and can lower revenue per visitor. Optimising the rate on its own is how a shop trains its customers to wait for a coupon.
Traffic mix moves the rate without anything changing on the page. A brand campaign brings in people who were going to buy anyway, and a broad prospecting campaign does not. A falling rate during a growth push is often the campaign working as intended.
Averaging rates across pages is wrong. Add the conversions and add the visitors, then divide. Averaging the per-page rates weights a page with forty visitors the same as one with forty thousand.
A target rate is not a plan. The tool tells you the 81 extra conversions three percent needs. What produces them is a change to the page, the offer or the traffic, and the arithmetic has no opinion about which.
Compatibility
Arithmetic in the browser: nothing is uploaded and nothing is stored. The share link carries the figures, which is the quickest way to end an argument about whether a difference is real.
The interval is a Wilson score interval rather than the p ± z·√(p(1−p)/n) taught in most statistics courses. The textbook version is too narrow at small proportions and will happily hand you a lower bound below zero; at 1 conversion in 200 it gives about -0.5 percent. Wilson’s stays inside 0 and 100 percent and behaves at the sample sizes conversion rates actually arrive in.
It is a confidence interval on a proportion, not a significance test between two pages. For an A/B test with two variants, the test to run is a two-proportion comparison, and the sample size you need is larger than most people expect.