Conversion Rate Calculator — CVR + Relative Lift

Percentage points and percent are different numbers

A conversion rate moving from 2% to 2.5% is two true statements at once: a rise of 0.5 percentage points, and a 25% relative lift. Both describe the identical change. They differ by a factor of fifty, and which one you quote decides what your reader hears.

The convention is worth stating plainly, because it is the whole fix: percentage points for the gap between two rates, percent for the proportional change between them. Writing "a 0.5% increase" for this is the ambiguous case, and it will be read as the relative figure — which is wrong by fifty times.

The arithmetic, on the defaults

10,000 visitors, and a rate moving from 2% to 2.5%:

• 2% of 10,000 = 200 conversions
• 2.5% of 10,000 = 250 conversions
• Absolute change: 2.5 - 2.0 = 0.5 percentage points
• Relative change: (2.5 - 2.0) / 2.0 x 100 = 25%

It goes wrong in both directions

This is usually framed as a warning about exaggeration, and that is only half of it.

Quote 25% to someone thinking in percentage points and you have oversold a modest change. But quote 0.5 points to someone expecting a relative figure and you have undersold work that genuinely produced a quarter more conversions per visitor — and that is how good work gets shelved.

The answer is not to prefer one number. It is to name which you are using, and then to give the figure that cannot be misread at all: 50 extra conversions. Whole outcomes have no convention to get wrong, and they are what connects to revenue.

A rate means nothing until the denominator is named

Before trusting any conversion rate, ask what it is divided by. All sessions? Unique visitors? Only people who reached the step where converting was possible? The same page over the same week produces materially different rates depending on which you pick, and none of them is wrong — they are answers to different questions.

Which is why published benchmark rates are close to useless for judging your own. You almost never know the denominator behind them, and without that the comparison is not a comparison. Your own rate measured consistently over time is far more informative than anyone else's number.

The trap: small samples move on their own

A jump from 2% to 2.5% is only a result if the sample was large enough to distinguish it from chance. Over a short window with few conversions, a rate wanders by itself — and the smaller the sample, the wider it wanders.

This calculator cannot tell you whether yours was large enough; it does the arithmetic, not the statistics. Work out the traffic the comparison needs before reading anything into it, using the A/B test sample size calculator. Doing it in that order is what separates a result from a coincidence you have already announced.

Two independent panels

The top converts visitors and conversions into a rate. The bottom converts two rates into a relative lift, and it does not care where those rates came from — so it works equally for two variants, two months, or two channels. Everything runs in your browser and nothing is uploaded.