10

Understanding Facebook’s EdgeRank Algorithm

by Nicholas Scalice on December 30, 2011

Understanding Facebook's EdgeRank Algorithm

As you can probably imagine, the default setting on your Facebook news feed only displays what Facebook thinks you’ll find most relevant. How do they do this? Through a rather complex algorithm which they call EdgeRank.

The name sounds familiar to Google’s PageRank algorithm, which Google uses to deliver the most relevant search results based on your query. While the idea of filtering content in order to promote relevancy is similar between EdgeRank and PageRank, that is where the similarities end.

You see, while Google deals primarily with filtering website content, Facebook is challenged with the task of filtering dynamic social content in real-time, based on a number of very unique and individual factors. Getting into the actual algorithm, Facebook looks at three major variables, which are affinity, weight and time decay. The EdgeRank of any given object can be thought of as the affinity score times weight score times time decay score.

Keep in mind that when we refer to an “object” we are talking about anything that is posted by users on Facebook. Examples of just a few types of objects on Facebook include status updates, videos, images and links.

Getting back to the variables, think of affinity as the relationship between the viewing user and the content creator. For example, if I frequently engage with a certain friend on Facebook, either by commenting on their updates, liking their content, visiting their profile or in a multitude of other ways, I am much more likely to see content from that particular user in my news feed. Facebook basically acknowledges that we interact with certain friends and brands more than others. Simply put, they want us to see more content from those in the first group and less content from those in the second group.

Moreover, just as any two Facebook users have an affinity score between them, a similar score also exists between any user and any Facebook page. So while I might have “liked” both the Nike page and the Adidas page, if I routinely interact with Nike more than Adidas, Facebook interprets this as meaning Nike’s content will have a higher affinity score for me and I would therefore like to see their content in my news feed.

As for the second variable, the weight score can be further broken down into three major subsections. I call them the type weight, interaction weight and preference weight. Type weight simply means that Facebook places more value on dynamic content types such as videos and images and less weight on links and status updates.

As for interaction weight, Facebook also places more value on certain types of interactions with the object. If an object is shared or if someone adds a comment, this has more weight than if someone were to simply “like” it or merely click on it. Of course, an object with some type of interaction is better than no interaction at all, so even an object with just a few clicks will have a higher EdgeRank than something with no interaction whatsoever.

Lastly, the preference weight basically means that each user has preferences on what type of content they like to interact with. Some people click on every video they see in their news feed, while others never play videos at all. Some people love browsing photos and others just like clicking links to blogs. All of these behaviors are summarized through the EdgeRank algorithm, with the intentions of providing each user with more of what (Facebook thinks) they want to see.

The last component of EdgeRank is time decay and this is probably the simplest variable of all. Since relevance in a digital world is so closely related to timeliness, this variable basically says that as an object gets older, the score gets lower. Facebook wants us to see not only the most relevant content, but also the most recent. A link that was posted three days ago is not as valuable as a link posted three minutes ago.

Even if some of this stuff seems overly complicated, the main idea to take away here is that we’re swiftly moving into a social environment that is deluged with more social content than anyone can even begin to comprehend. Every 60 seconds, over 600,000 status updates occur on Facebook. Imagine if there was no way to filter any of this. It wouldn’t be a very pleasant user experience.

That is why Facebook saw the need for EdgeRank. Now, also keep in mind that this is Facebook’s first major attempt at filtering our content. Just as Google’s PageRank progressively improved over the years, I’m sure the same thing will happen with EdgeRank.

That means that by tomorrow, the core principles of EdgeRank can potentially be completely different from what was described above. Nevertheless, one thing that won’t change is the need for some type of ranking and filtering algorithm. Currently, Facebook seems pretty quiet about EdgeRank. There are no official Facebook pages discussing it and a Google search doesn’t turn up much.

My prediction however, is that it’ll be an even bigger part of Facebook’s success in the years to come as more content is added through new methods, such as “frictionless sharing.” In a future post, I will discuss some of the methods you can use to leverage the power of EdgeRank for your content. Stay tuned for much more on this topic.

The following two tabs change content below.

Nicholas Scalice

Founder at FastBlink
A native of Boca Raton, Florida, Nicholas founded FastBlink in 2009. He has a diverse background in direct sales, affiliate marketing, domain name investing and content marketing.

Latest posts by Nicholas Scalice (see all)

{ 4 comments… read them below or add one }

Peter Womersley January 11, 2012 at 6:37 am

A very interesting article which certainly provides a lot of insight into FB’s EdgeRank which I certainly never appreciated. Thanks

Reply

Nicholas Scalice January 12, 2012 at 3:55 pm

Thanks Peter! I have more on EdgeRank in the works so stay tuned!

Reply

Digby January 11, 2012 at 8:28 am

Great article covering the factors involved in Edge Rank. I think it’s worth highlighting from a marketing perspective that all of the above can be considered when posting except ‘preference weight’, which is something that, as you say, entirely depends on the user. I imagine that this factor is calculated and applied after the edgerank of posts.

It would be really interesting if Facebook as well as showing which content received the most interactions, provided you with a ‘fan content preference’ demographic in its analytics to indicate the likelihood of different types of posts appearing in news feeds.

Reply

Nicholas Scalice January 12, 2012 at 3:58 pm

Yes, preference weight is something that can be entirely different between users. Have a statistic to track it would be extremely helpful. I hope to see something like that from Facebook.

Reply

Leave a Comment

{ 6 trackbacks }

Previous post:

Next post: