From ratio to %: 1:999 = 1/(999+1) = 1/1000 = 0.1%

Otherwise you’d get 1:1000 = 1/1001 = 0.0999001%

It doesn’t make a big difference here but consider 200:800 = 1:4 = 20% vs 200:1000 = 1:5 = 16.7%,
or 500:500 = 50% vs 500:1000 = 33%. That’s enough to perceive a false statistical significance if one of the before-and-after numbers isn’t calculated correctly.

Just nitpicking; very solid article.

Considering that you compare

Day 1: number of users = 1

and

Day 2: number of users = 2

The difference is = 1

and 1 is exactly the total number of user of day 1, i.e. 100% of day 1

On day 2 we have double the users buying, but the increase itself compared to day 1 is 100%.

Or am I suddenly dumb? :)

rgrds

Carlo