This week is marked by the controversy Facebook, which tests a newsfeed containing no publication page (see here and here ). But for now, brands are still present in News Feed and seek to improve their visibility. To help them, Facebook publishes an explanatory video on its algorithm and a page that references the guidelines to respect to maximize its chances of appearing in a good position in the user’s news feed.
How does the Facebook algorithm work?
If you are a social media professional, the operating principles of the newsfeed ranking algorithm, formerly titled edgerank, have no secrets for you. This video produced by Facebook, however, has the merit of simply explaining how this algorithm works and what is its interest for users.
The metaphor of restaurant choice
The social network compares its algorithm to mechanisms that allow us to choose a menu in the restaurant. We first look at the available dishes (inventory). Then we examine the content of these dishes (meat, fish, vegetables …), we compare them to what we like and we take into account other elements such as the time of day (breakfast, lunch) , evening …) or the meals consumed the day before. Finally, we make a prediction: which meal will make me happiest? The Facebook algorithm works exactly on this principle.
The inventory of content available on Facebook
Facebook begins by referencing the content available to each user (Facebook page posts and friends, depending on the confidentiality assigned).
The importance of social cues
Then, Facebook analyzes signals for each available content: when was posted the content, who published it … The social network is particularly attentive to user feedback (reporting, hidden content, clickbait …) .Facebook also looks at other elements like the quality of your connection or the model of your phone, to provide adapted content in all circumstances.
Predictions and score of each publication
Depending on the content available and the associated signals, Facebook makes predictions for each user. What is the probability that a user likes, shares or comments on such content? He assigns a score to each post and then ranks them in descending order of the marks assigned in the newsfeed.