I know these are just estimates, but 90m monthly actives on android alone seems unlikely for candy crush. Google play claims 10m-50m installs. Sounds like data is being extrapolated from a biased sample, unless play's data is regional.

I just checked my app and the numbers for both monthly and daily actives appear to be about an order of magnitude too high. They are close if you just divide by 10 though. Maybe this is the case for all the apps? The rankings look like they could be right.

That would make more sense, with candy crush being so closely tied to facebook, and currently having ~68m monthly users there (that is of course iOS + android + canvas), I would have guessed 10m-20m on android. Granted this is just one app.

Cool. Which game do you make?

Your comment immediately brought this to mind: http://xkcd.com/605/ :)

The data is sampled from the Heyzap app. So English focused social games will tend to be overestimated.

We did calibrate on quite a few data points. I will think about normalizing for the above bias too