These are not good headlines they are more like those which have phrases like “what happens next is hard to believe”, “I saw this and I was SHOCKED” “This dog bit the postman and his result was priceless.”
Facebook has manually classified tens of thousands of headlines with a clickbaitiness score to train the new algorithm. It scores each one and the higher the scoore the more the algorithm punishes the entire Page that shared them or site they link to by making all their posts or referral links less visible.
Facebook’s VP of Product Management on News Feed, Adam Mosseri was quoted as saying that if you post 50 times a day and post one piece a clickbait, this shouldn’t affect you. If you’re a spammer and post clickbait all day, this is going to hurt.
If a publisher reforms its ways and “they stop posting clickbait, their referral traffic will bounce back.” The algorithm identifies offenders on both the web domain level and the Facebook Page level, spammers can’t build a Page and then just launch new websites with different URLs to elude punishment.
Facebook already made an anti-clickbait algorithm change in 2014, focusing on user behavior. It looked for links that people Liked then clicked, but then immediately bounced back to Facebook and Unliked because they hated what they saw.
In February Facebook started looking more closely at how little time people lingered on clickbait sites, and identified links with lots of clicks but very few subsequent Likes. Today’s change focuses on the source of the problem rather than tracking the symptoms.