We get this question a lot.
There’s no way to accurately identify fake accounts or likes on Facebook or twitter, though some software companies pretend to.
Often, the fake users have more robust profiles and activity than real users, as the spammers know what signals to design for.
We used to be able to filter by demographics, lack of a profile photo, and lack of friends– not anymore with the sophisticated guys.
We do have some manual ways to check for this by looking at funnel behavior– but that’s all that’s really possible now.
You can’t block fake likes, but you can do things like analyze tenure, traffic from Facebook to the web, and conversion values.
We’ve written extensively over the years on why not to buy likes, as well as how people do it.
So what is fake? You have to look at other behavior to see which “fans” don’t seem real– doesn’t have to be Facebook-only behavior.
The percentage of fake fans in general should be a couple percent, tops, unless the page has explicitly bought fans or used an agency that has.
Someone said 7% as a figure, but it’s misreported because of people who create profiles for pages.