[Retric]

Fake accounts among other things. For example ~0.1% of their users die every month but they often don't get notified etc.

Further, FB's own estimates cover a fairly wide range, but they have little incentive to user lower estimates.

[dudurocha]

By their official numbers, they have 1,18 Bi DAUs and 1,7 Bi MAUs.

If somebody dies, they won't be counted as a DAU anymore and in on month won't be counted as a MAU.

More numbers in their Q3 earnings reports: https://investor.fb.com/investor-news/press-release-details/...

[Retric]

This may seem pedantic. However, it's not uncommon for a relative to log into a dead users account. Thus showing up on DAU and MAU even after death. Users will log in if for no other reason than to notify people that the user passed and mention the funeral date etc. What makes this significant is these accounts where real and FB really can't get accurate numbers for this so it's all internal estimates.

So, yes FB's official numbers take this and many other factors into account. However, if their internal estimate is say 5%-40% of accounts are fake they are likely to use a number close to 5% than 40%.

[the_watcher]

While that may happen, the frequency is so low (and would happen once, not every month), that it does not materially impact numbers in the billions.

[Retric]

Dead users are just one of many types of 'fake' user. On it's own it's a small percentage of total active accounts.

But, I bring it up as the proxy for lot's of other similar estimates. I don't know if what % of 'active' accounts in their logs be that 0.1%, 0.2%, or 0.02% are dead, but I suspect they are going to pick a subjective number that makes them look better.

Further, because actuate numbers take a lot of accurate estimates even 'small' biasing of these sub estimates will compound and heavily bias their reported statistics.

AKA, spam accounts may be easy to find. But, spammers may create other fake accounts to mask their 'spam' accounts.