How about you actually learn something about Steamspy first before commenting on it?
That's an article from 2017. BEFORE the change to the default Steam profiles (used to gather Steamspy's data) that I was mentioning in my reply and that severely compromised Steamspy's reliability as a source.
It didn't compromise the reliability, it should still have a confidence interval of 98%, its just given the fact that you have less useful data to extrapolate from, the size of the range has become much larger. Where as before they could estimate within say ~20,000 on a game with 10,0000,0000 sales, now they need to provide a much larger band (say 1,000,000) for example. Whilst the 2nd figure is less useful, it will still hit the mark just as often as the previous statistic did.
