EdgeRank: The truth about Facebook’s secret sauce

As most avid Facebook users know, there are two ways you can consume your feed: “Top News” and “Most Recent”. The default view is “Top News” and uses the secret algorithm called EdgeRank to filter through the multitude of posts and only show us what is supposedly more relevant to us. Although the “Most Recent” view should list all the updates from your friends and pages views, it probably also uses a filtering algorithm of some sort (similar to EdgeRank) when you have more than 250 friends (and pages).

So, what is EdgeRank anyway? Exactly a year ago, at the  f8 conference, Ruchi Sanghvi and Ari Steinberg revealed a formula that was picked up and explained in excruciating details by the technical press.

Unfortunately, judging by the number of “Social Media gurus” who have attempted to crack the code in the past year, there is still a lot of confusion as to what EdgeRank really is.

The first confusion is around its name. EdgeRank is not ranking the edge!

Which leads us to the $64,000 question: what is an edge? In a social network, an edge is a relationship or, as Facebook really defines it, it is the result of an interaction with an object. When a user likes or writes a comment on a post, for example, he creates an edge. Note that the act of creating an object (e.g. status update) is also an edge.

So, if EdgeRank is not ranking an edge, what is it ranking? It is really ranking an Object. Objects are at the core of Facebook’s Graph API. They represent entities such as a User, Page, Status message or Photo for example.

I am certain that Facebook had fun sharing this cryptic formula to their developers’ audience last year:

Although it looks like it, you don’t need to be a rocket scientist to understand the formula. It is nothing more that the sum of all the “edge ranks” (yes, here we are really talking about edge ranks). Facebook calls this the EdgeRank (I call it object rank, but more on that later)

The confusion that was cultivated by a number of people in the past year comes from the fact that an EdgeRank is user-specific by definition, and not a score that can be attributed to a Page or user as former SEO experts and EdgeRankChecker would like us to believe. News Feed Optimization (NFO) is very different from SEO (Search Engine Optimization). The EdgeRank of an Object will not only be different for every single user, but it is also a rather ephemeral score that will be changing over time. This is radically different from Google’s PageRank, which is somewhat static over (short periods of) time and is the same for everyone.

As a Facebook Page owner, here is what you need to know to improve the factors influencing the calculation of your posts’ EdgeRanks. Let’s assume you are posting some content on your Page wall. What would determine whether this content is published to your fans’ wall?

Affinity: Friends or fans with whom you regularly interact receive a higher affinity score
This affinity is calculated for every “edge”, which means it is calculated for your post, but also for all the “edges” (or interactions) users had with the post.
So here, we are talking about affinity between:
- the viewer and the Facebook Page
- the viewer and the edge creator (for every edge)
Recommendation: This factor is probably the most difficult to improve as it is based on affinity that is out of our control. However, it would make sense to target friends of fans to increase overall affinity score for existing Fans in the long run. As an example, I am a Fan of “The Office” show. However, none of my friends are. I see posts from “The Office” page occasionally, but only after the post reached a certain threshold of “edges” (typically 30 minutes after post was published). If I had friends who were also fans and viewed, clicked, liked or commented on “The Office” Fan Page posts, I would probably see more posts, and also probably see them before they reach 500+ edges.

Level of Interaction
Although there was no details provided by Facebook regarding the weights of each edge type, empirical studies that were run over the past year confirmed the obvious: a view or click weighs less than a like, which weighs less than a comment. Also, the type of posts seems to influence the level of interaction. A video or photo will typically have generate more edges than a text post.
Recommendation: Focus on posts or applications that engage the users. Open-ended questions, surveys, trivias or engagement applications have demonstrated increased interaction.

Timeliness
This factor is the obvious one. People want to see timely and relevant posts. Newer posts will eventually replace older posts.
Recommendation: Keep content fresh. If content does not work, move on. Don’t overpost either as it would probably eventually decrease and dilute fan participation.

EdgeRank is definitely important but your Facebook strategy should focus on engaging your fans. This is definitely not a one-time event but a process. Webtrends Analytics 10 provides a number of metrics via different data collection mechanisms, including the Facebook Insights API that can help you improve and optimize your interactions with your fans. One of those metrics is the number of impressions. Following this measure can help you identify what type of content needs to be posted (and when) in order to be viewed by the maximum number of people.

Unlike Google’s PageRank which influences SERP (Search Engine Results Page) position, EdgeRank is a rather ephemeral and dynamic score that is used to filter posts that show up on users’ walls.

Google’s PageRank was named after Larry Page and fortuitously, it refers to the ranking of a web page ( it is also a cooler name than BrinRank!) Maybe Facebook should rename its EdgeRank algorithm to something that makes sense like ObjectRank, PostRank (although already taken by one of our partners)… or ZuckerbergRank?