How Facebook Algorithm Works ?

We all are more or less active users of Facebook . Have you ever wondered how does the news feed show you different types of content and posts from various other users including those in your network ?

Let us quickly have a glance on the working of the FB algorithm

The crux of the solution is a machine learning ranking system which is a fusion of different algorithms that coherently work together in phases . Some of them focus on selecting posts to show in a news feed, removing posts with misinformation while others deal with creating lists of friends that a person interacts with, topics of likes/ dislikes etc all these factors put together determine the overall ranking of elements in the feed .

The objective is to predict the relevance for a member and serve him/her the most appropriate content based on their preferences and past actions. Ranking happens based on the interests and higher probability of interaction from user's standpoint .

Thousands of signals are at play , I have outlined few of the important ones

Ranking Signals

• Characteristics of post : Quality/feature of a post and determining whether this is the kind of thing that a user tends to interact with more

For example, if a post is accompanied with an image and a member has a history of liking such posts then they rank higher . If John always likes and comments on Dick's updates /posts then Dick's posts would get prominence and more visibility on John's news feed

• Time as a factor : Recent posts get higher weightage and based on popularity it can get a ranking boost
• Engagement and Interest : Predicting whether a user will be likely to be interested in or engage with a post

Facebook uses machine learning models to forecast different aspects. There’s a model that foretells what content a user will like, a separate model predicts which post the user will comment and so on .

Each of these forms of engagement receives a ranking score and are subsequently ranked.

Personalization Of Ranking Signals

An interesting insight into ranking factors is that they are weighted differently from one user to the next.

Weights may vary for each user , some may have comments as the primary factor while others may have likes or shares .

Diversity and Context : Diversity and context plays a key role in showcasing content in a user's feed. Objective is to ensure that user’s feed doesn’t become repetitive.

Key Takeaway

Machine Learning has powered FB algorithm which is ever evolving to incorporate different variables and behavioral aspects which makes the ranking signals dynamic and can vary for every user .

About The Author :

Aditya Krishnan is a Digital Strategist & Consultant with 10+ years of experience across different sectors and practices . He has led and managed business transformation engagements ranging from setting up digital platforms , designing marketing strategies to leading conversion optimization, analytics projects . An active contributor and participant in the Webmasters , SEO , digital marketing, analytics professional groups/forums including trends on cyber security . He has authored several publications on various topics across portals and shares his views regularly on latest trends in the digital / marketing space . In addition to it he is also a reputed author , poet and an avid researcher .

Follow me on Twitter : https://twitter.com/adityaskrishnan