I’ve written a lot over the past few years about the growing popularity of astroturfing, which is the purchasing of various fake things to boost your social media metrics. Whether that’s buying more followers on Twitter, obtaining extra likes on Facebook or acquiring more views of your YouTube video, there are any number of ways you can fake your online stats.
That such artificial inflation is still so popular underlines the misguided nature of most of our means of measuring success online. After all, the only people in the business of counting followers tend to be the people selling them. For everyone else, there is a real business, with customers and sales, that are what they’re trading to try and improve. Those are the things that should be measured, rather than the vanity metrics so many persist in using.
A recent study, lead by Emiliano De Cristofaro at University College London, set out to explore just what impact this industry has on the social landscape.
The researchers created 13 new Facebook pages around the topic of ‘virtual electricity’. None of the pages had any content, and all were given a description highlighting that the pages were not real and instructing people not to like it.
Do Facebook ads attract bots?
They then proceeded to create adverts on Facebook to generate traffic to five of the pages, spending $90 over the course of 15 days to do so.
Simultaneously they purchased a number of likes from four popular ‘like farms’ to buy likes for the remaining eight pages. These services generated 1000 likes for the Pages over the same 15 day period, for fees ranging from $70 to $190.
The researchers then measured the level of activity each page generated over the next three weeks. They did this both by using the analytics provided to each page owner, but also by trawling through the profiles of those people that had liked.
The results highlight the difference between those supporters generated authentically (via the ad campaign), and those from more nefarious means. The ad campaign attracted 32 supporters, with a nice split between men and women, and the majority being based in the US. Likewise, when the ad campaign was run in India and Egypt, nearly all of the people attracted were from those nations. When the campaign was global however, practically all came from India.
When the profiles were analyzed in more depth, these followers had an average of around 300 friends each, which is broadly in line with the average around the world. Where things deviated from the norm however was in their linking behaviour. Each was found to have liked somewhere between 600 and 1000 pages. This compares to a norm of around 40.
Inside the like farms
Strange though that ‘demographic’ was however, the people bought from the like farms were even more peculiar. The first abnormality was in the speed of liking. With an ad campaign, the popularity of a page would go up gradually, but when buying likes from a farm, it tends to go up in huge numbers, very quickly.
Not only is such behaviour not particularly helpful to you as a page owner, it’s also a very easily detectable signal for Facebook, although they have strangely declined to act on it thus far.
The profiles of the fake accounts used by the like farms are also unusual. They tend to like somewhere between 1200 and 1800 pages each, a huge amount more than is normal.
Whilst the ‘profiles’ recruited via the like farms were peculiar, what was equally worrying is the abnormal profiles attracted via the ad campaign. The researchers didn’t explore the validity of the ad generated profiles, but it would have to raise some serious flags you would think, and cause one to question the value of Facebook advertising if that kind of outcome becomes commonplace.
Suffice to say, with the page the researchers created doing all it could to deter ‘real’ users from liking it, it should perhaps not be too surprising that the number of apparent fakes signing up was so high, but nevertheless, it does highlight that there does appear to be a significant level of fake activity going on on Facebook, whether you specifically pay for it or not.
Buyer beware.Original post