TwitterAudit — “exposing Twitter fraud since 2012” — is designed to weed out the real followers from the fake ones and determine how genuine some accounts are about how they get their followers.
To get a better understanding of how this all works and what we can make of these results, we caught up with TwitterAudit co-creator David Caplan.
TwitterAudit debuted in beta just last week on August 2, after Caplan and his business partner David Gross had been working on it since mid-July. “It’s a side project for us and is still a bit rough around the edges,” Caplain said.
The idea for the service came about when the two were looking for ways to get more followers on Twitter. “It turns out it’s pretty easy to buy fake followers to beef up your numbers so we wondered whether people could be ‘audited’ as a way to expose ‘Twitter fraud,'” Caplain said. Most people assume that followers are earned, but that doesn’t always turn out to be the case.
Caplan cites Klout and SocialBro as other programs that offer similar metrics. The differences are that Klout bases its “scores” on the activity of the account being examined, rather than the activity of its followers, and SocialBro gives great insight into the quality of followers.
With TwitterAudit, Caplan and Gross wanted to create a “simple, free tool to quickly get an ‘audit score’ for any Twitter user and find out how many real, active followers they really have.”
Caplan confirms that the results are pretty valid from a statistical point of view. TwitterAudit takes a random sample of 5,000 followers (if the person has less than 5000, it looks at all of them) and examines their activity.
“This sample size might look small,” Caplan said. “For example if the person has millions of followers, but keep in mind that highly regarded polls taken in the USA only use a sample size of 1000 to gauge the entire country’s opinion!”
Still, he said that there can be a margin of error of 5-10%, and he’s working on a way to estimate that.
Caplan also noted that the audit scores for users with followers in the millions can reflect the fact that about 30-50% of Twitter users are completely inactive. That’ll put a damper on things.
“I would say that any score above 40% should be considered good. Anything less and I would begin to worry.”
Looks like our top political leaders are okay for right now, though Obama might want to keep an eye on that 46%.