Facebook referrals are crucial for traffic to hyperpartisan and fake news sites
Facebook is an important source of traffic for virtually all news outlets in the United States. New data, however, indicates that the social network is a far more important channel for some of the largest hyperpartisan and fake news sites in the country.
Analytics company Jumpshot tracked Facebook referrals to 20-plus websites between September 11th and November 15th. It then calculated the share of unique visitors that figure represented. The data, seen by Poynter, refer to PC traffic only.
Almost 80 percent of unique visitors to hyperpartisan news pages Occupy Democrats and American News came from Facebook, according to Jumpshot. Both of these websites have Facebook followings in the seven figures, and both have had several run-ins with the fact-checkers. American News, for one, peddled a Megyn Kelly hoax that fooled Facebook's Trending section — even several weeks after it was exposed as a fake.
For comparison, Jumpshot says that in the same period, Facebook accounted for about 20 percent of traffic to The New York Times and 11 percent of visits to CNN. Our own analytics indicate that Facebook accounted for 24 percent of total pageviews on Poynter.org in the first four weeks of November.
The completely bogus news site abcnews.com.co – which can credit 60 percent of total visitors to Facebook – also relied disproportionately on the social network. Among other hits, the site carried a fake story about paid protesters at Trump rallies that fooled his campaign manager and the President-elect's own son. It also claimed, in a debunked story that had 2.2 million engagements on Facebook, that President Obama banned the Pledge of Allegiance.
70 News, which fooled Google News with its bogus story about the popular vote, obtained a lower but still respectable 44 percent of its traffic from Facebook. (The figure could be depressed by the strength of its Google referrals on that one story.)
Jumpshot tracks 100 million devices at a global level. For this exercise, the sample monitored per site ranged from 10 thousand to 800 thousand devices. Data scientists at the company told Poynter they removed sites for which they didn't have enough data and that they believe the samples are large enough to be representative.
The data serves as a reminder of how important the social network is as a distribution channel for fake news and hyperpartisan sites.