How does facebook rank news feed?
Last updated
Last updated
Stopped since 2011, replaced by Machine Learning with 100K factors.
https://mashable.com/2013/05/07/facebook-edgerank-infographic/
The new Facebook algorithm works by ranking all available posts that can display on a user’s News Feed based on how likely that user will have a positive reaction.
Facebook’s algorithm for ranking and displaying content on your News Feed is based on four factors:
The Inventory of all posts available to display.
Signals that tell Facebook what each post is.
Predictions on how you will react to each post.
A Final Score assigned to the content based on all factors considered.
The process is based on the Vickrey-Clarke-Groves algorithm, which “operates as a closed auction,” in which advertisers’ bids are kept hidden from one another, prompting them to bid their real value.
The new Facebook algorithm works by prioritizing content posted from friends over publishers, with a focus on “meaningful interactions.”
Since the data controversy erupted around the social network in late 2017, Facebook has worked to improve transparency around how it ranks content on the News Feed.
Between Facebook’s F8 conferences, News Feed webinars, and algorithm presentations — we can now say that Facebook’s new algorithm is no longer a complete black box.
Facebook went public with changes to the algorithm in their post “Meaningful Interactions” update back in January 2018.
The goal of Facebook’s algorithm is to “show stories that matter to users,” according to Adam Mosseri, VP of Facebook’s News Feed Management. With that in mind, you should know how Facebook’s different algorithm factors work together to determine which stories “matter” to a user.
Inventory represents the stock of all content that can display to a user on Facebook’s News Feed.
This includes everything posted from friends and publishers.
Signals represent the information that Facebook can gather about a piece of content.
Signals are the single factor that advertisers have control over. These are your inputs that Facebook interprets; type of content, the publisher, its age, purpose, and more.
You want your content to signal to Facebook that it’s meaningful and relevant to your target audience.
Predictions represent the behavior of a user and how likely they are predicted to have a positive interaction with a content piece.
Score is the final number assigned to a piece of content based on the likelihood the user will respond positively to it.
As advertisers, the only part of the process that we have control over are the signals of our content.
These signals can be divided into two categories: passive and active.
Passive signals include view time, story type, time posted, and other metrics non-active metrics.
Active signals include likes, shares, comments, and other active events that prompt engagement.
You should tailor your content to promote positive engagement, or what Facebook has defined as “meaningful interactions.”
4.75 billion pieces of content shares – daily
300+ million photo uploads – daily
510,000 comments – every 60 seconds
293,000 statuses – every 60 seconds
