Ranking Newsfeeds: How Does Facebook Do It?
Posted on February 16, 2014
Filed Under Content Marketing Strategy, Facebook, General, Internet Marketing, Marketing, Social Media, Social Media Marketing, Social Networking, Social Networks, Traffic Building
Facebook’s EdgeRank algorithm is dead. In fact, it’s been dead for the last two and a half years. Instead, the social networking service is using a more complex algorithm based on machine learning. The current News Feed algorithm doesn’t have a catchy name but it is estimated that as many as 100,000 individual factors are involved in organizing and producing a Facebook user’s News Feed.
Facebook’s original News Feed ranking system worked around 3 variables:
3. Time decay
Affinity is a one-to-one function between a Facebook user and an edge. In Facebook’s EdgeRank, an edge is anything that happens on Facebook. It can be a status update, a like, a comment or a share. For businesses, affinity quantifies and qualifies the connection between a brand and a fan. Affinity is built when a fan repeatedly interacts with a brand’s edges.
On Facebook, a comment is deemed more valuable than a like because it involves more engagement. In the “weight” system used by Facebook’s EdgeRank, certain actions have more value than others. All edges are assigned a value.
Time decay is a measure of how long an edge remains alive. Mathematically, time decay equals 1/(time since action) and the older an edge, the less valuable it is. This serves to keep Facebook’s news feed fresh.
So with EdgeRank, the way people used Facebook determined what stories showed up in their News Feed. This is still the case but now, there are some other factors that also come into play.
Each month, more than a billion people use Facebook. Users are accessing millions of Pages (over 18 million Pages to be precise), many of which actively look for ways to show up in people’s News Feeds and they are using a diverse range of mobile devices to do so with different technical capabilities for processing and displaying content. With all that going on, Facebook faces the challenge of controlling what stories users see each time they log in on Facebook and what content from Facebook Pages make it into people’s News Feed. Affinity, weight and time decay are still important but now, there are also categories and sub-categories of affinity.
Facebook is now measuring how close a user is to specific friends and Pages based on not only personal interactions but also global interactions. For example, if Facebook shows an update to 100 users and only a few interact with it, it may not show in your News Feed but if many people actively interact with it, chances are, the story will show in your News Feed.
1. Relationship Settings
Facebook now allows users to choose different relationship settings. For each Facebook friend that you have, you can label the person a “close friend” or an “acquaintance.” With liked Pages, users can choose to “Get notifications” or “Receive updates”. These factors are taken into consideration to determine what shows up in people’s News Feeds.
2. Post Types
The current News Feed algorithm also takes into account the type of post that each user tends to like. If you engage with a certain type of post often, chances are, you’ll see more posts of that type than any other.
3. Hide Post, Spam
Content that shows up in News Feed is also impacted by a user’s ability to hide posts or mark them as spam. As Facebook users report a post, chances that you will see that post is less.
Basically, for each News Feed story, there is some computation that happens. For example, given a story and a user’s Facebook usage history, the probability that the user will like the story is calculated or the probability that the user will hide it. This in turn influences what shows up in people’s News Feed.
4. Facebook Ads
Although the News Feed algorithm is separate from the algorithm that organizes Facebook ads, clicking on ads or looking at other timelines may influence what stories show up in a user’s News Feed.
5. Technical Considerations
Last but not least, unlike its predecessor, the current News Feed algorithm also takes into consideration things like the device being used to access Facebook and the speed of the user’s Internet connection to filter the stories that show up in the user’s News Feed.
All in all, it’s fair to say that Facebook is showing its users content they will find interesting and will want to interact with.
Facebook’s News Feed algorithm has developed into something significantly more complicated than EdgeRank. The challenge it seems is for marketers to constantly create and publish content that will attract likes, comments and shares.