Today I had the pleasure of moderating a panel at #PubCon on Edgerank. Edgerank is the algorithm that Facebook uses to determine what you see in your Facebook newsfeed.
First up was Jonathan Goodman, CEO of Halyard Consulting.
Here is an analysis of how EdgeRank is working in Facebook Today.
1 out of every 20 stories in the newsfeed is an add (and this is increasing)
Before the edgerank althorithm had 3 main factors (affinity, weight and decay); now it has over 100,000 – it is based on MACHINE LEARNING and is (allegedly) always getting smarter
Affinity still matters, but it is now broken down into categories and subcategories
Generally, the more interactions, the more the post will be displayed in newsfeeds
Post types matter based on the individual – if a person prefers photos, they are more likely to see photos, if they generally like links, they are more likely to see articles with links
Interactions with ads can influence the newsfeed
Device also matters – some content does better on desktop vs. mobile
Story bumping means that posts may show up more often and decay is less relevant. Older unseen or uninteracted with posts may be shown again
If you interact with things you will likely see them again
Last Actor is the last 50 rolling interactions which assist sorting in the newsfeed
Complaints – if people hide or report as spam your content will be less seen
Relationships matter
In notifications you can set which notifications you want to get from pages – you can encourage your fans to change their settings to get all of the notifications from your page
COMING SOON TO FACEBOOK EDGERANK
Chronological ordering for certain types of posts, like live events or sports. Rather than showing these posts based on the traditional algorithm, Facebook will try to create a storyline for messages about events so that you see them in the order in which they are posted.
Today I had the pleasure of moderating a panel at #PubCon on Edgerank. Edgerank is the algorithm that Facebook uses to determine what you see in your Facebook newsfeed.
