If they're not removing obvious frauds in these numbers, they'll be the target of shareholder class action lawsuits. In general, numbers published by big corporations tend to be reasonably sourced due to this reason - gotchas tend to more subtle.
Privately held company numbers on the other hand are generally much more suspect (Uber before IPO, Snap before IPO, We work, Theranos...)
But there are plenty of proxy measures that show similar growth. For example the AppAnnie "active user" measure actually measures on device usage, and shows that 4 out of the top 5 most used apps world wide (or maybe ex-China? Unclear) are Facebook-owned (WhatsApp, Facebook, FB Messenger and Instagram. WeChat is number 4).
Privately held company numbers on the other hand are generally much more suspect (Uber before IPO, Snap before IPO, We work, Theranos...)