We blocked a lot of spam from our db at some point, that's why we have less pubkeys and notes counted. Same for total zaps - we ignore self-zaps and some big obviously-fake zaps. We also treat 'users' more strictly (a pubkey with a profile, published post, and at least 1 followed person). These should account for most of the differences. 

 Your stats have typically lined up much more with my anecdotal experience too. I think you apply some pretty smart fillers there. 

 Makes sense and thanks for your service.