Facebook’s making it harder for fly-by-night publishers to monetize low-value websites.
On Wednesday, Facebook said it’s starting to use artificial intelligence to lay down the law on links that shunt users to websites with too many ads, as well as ads of a "disruptive, malicious and shocking" nature. Frowned-upon ad types include excessively disruptive units like pop-ups and interstitials, sexually suggestive ads and plain old deceptive ads.
The move is a strike against viral publishers that use Facebook's ad tools to do content arbitrage, buying eyeballs at low prices and then turning a profit on the click using voluminous ad loads.
"It’s not about large or small, short or long,” said Jason Kint, CEO of publisher trade org Digital Content Next. “If they’re cleaning up the crap, I applaud it.”
Facebook analyzed hundreds of thousands of web pages linked from the news feed in order to train its AI model to pinpoint sites with little or no substance and vexatious or injurious ad experiences.
If the model determines that a post likely links to a trashy site, that post will be deprioritized by the news feed algorithm and won’t be eligible to be promoted as an ad.
For now, the update will ban the owners of crummy links from buying paid media to promote those links on Facebook, Instagram and Audience Network, in addition to suppressing organic posts in the Facebook news feed.
Based on what Facebook learns, it may extend the policy to organic posts on Instagram.
Facebook warned low-quality publishers to expect a decline in traffic as the algorithm update rolls out over the next few months. On the other end of the spectrum, Facebook said quality publishers might experience a concurrent uptick.
As to how this will all play out, it's still a bit early on that front, said Purch CTO John Potter.
"It's not clear to me how they are doing the ranking, how much they will be penalizing the links or if they can even keep up with the all the different domains," Potter said.
But this isn’t the first time Facebook has attempted to clean up the news feed. In 2014, for example, Facebook started rooting out clickbait by analyzing user behavior, like time spent reading an article and the number of clicks vs. subsequent shares, to determine if content is valuable.
To some extent, the move is enforcing rules that were already on the books. Facebook created a policy last year to prevent misleading ad positioning, like overly sensationalized headlines and links that send people off to unrelated or low-quality content sites.
The policy calls out most of the tools in the social arbitrage publisher’s toolkit, including deceptive ad copy and images that are excessively cropped and require users to click to see the whole thing.
Shady publishers have long exploited Facebook and other social platforms to buy cheap clicks and drive traffic to ad-heavy sites where they can monetize the ever-loving snot out of visitors.
But Facebook’s move to tighten the screws jibes with the increasingly loud battle cry in the industry for a cleaner and more transparent ad ecosystem, and ties into Facebook’s other efforts to curb the spread of fake news on its platform.