I can’t be certain in this case, but the usual suspects are not being connected to as many servers (e.g. not scraping some because of robots.txt settings), delays in scraping the stats, or excluding some servers consciously because their stats are deemed a bit suspicious.