Skip to main content
Dave on Design
Aspects of the important, overlooked and subtle in UX, design, management, and the world.

Facebook's Fake Followers

Currently I moderate a Facebook group with about twenty thousand members. We get about four thousand fake membership applications a month, This is a bit over a hundred applicants a day, and between three admins, we each screen a few dozen applications each day. and it’s often very hard to tell if a profile is a fake profile, including apparently reasonable answers to screening questions during the join process.   The page is a group for foreigners where I live. It’s a natural target for generic spammers. Another group I moderate with about half the membership, which is for users of a specialised technology, sees only a dozen applicants a day and importantly the screening questions are much easier to write to differentiate genuinely interested accounts.

More, the accounts will often not post at all (though they may perform innocuous actions such as likes) for up to a couple of years after joining, only after which they begin posting spam: they are sleeper accounts.

This seems a remarkably long-term tactic. But also seems to successfully bypass Facebook’s anti-spam measures, which mainly focus on new accounts and new group members as potentially fake. There are currently no tools for Facebook group moderators to counter this: for example, you cannot perform member analysis to find longterm accounts without comment or post actions in order to bulk remove.

Two conclusions:

1 It is quite reasonable that significant Facebook group growth and possibly non-obvious-spam interaction such as likes can be driven entirely by simply accepting all applicants, See Metrics and Mistakes for a related discussion of Facebook group follower counts. but worse, by accepting apparently genuine applicants.

2 Facebook’s tools to counter spam in groups focus on new accounts; a tactic of sleeper accounts could devalue groups that even cautiously accept members, on a timescale of years.