Social edgerank

Published on August 28th, 2012 | by Paul Morris

0

Facebook Advertising – EdgeRank and the Future

By Paul Morris
I’m interested in Facebook Advertising hence essentially went on a mission to find out about the Facebook algorithm that governs its advertising platform. Figure that out and you stand a better chance of success to counter…

All the damn negativity. From Facebook themselves admitting that 5-6% of users are fake, to the BBC report suggesting the problem is worse to companies such as Limited Run saying the bot click problem is endemic and makes Facebook advertising a joke there are many people suggesting that Facebook ads are a waste of time.

I myself have written about there being many challenges to the Facebooks advertising model however rather than just writing off this advertising platform let’s delve in to what makes it tick.

And what makes Facebook’s advertising platform tick is EdgeRank (view the below video for an overview).


 

EdgeRank? Yep their algorithm is called EdgeRank as every bit of content or interaction in the News Feed is called an ‘edge’. As a result the newsfeed more resembles a chart of important edges than a list of recent news.

An example of its impact is that circa 95% of users use the top news feed and as this is driven by EdgeRank you need to understand it to crack Facebook advertising.

To those more akin to Search Engine Optimisation view EdgeRank as Facebook’s version of the PPC Quality Score or SEO Page Rank

Edge Rank is all about Affinity, Edge Weight and Time Decay (or Recency as conventional marketers might say).

Affinity – A score based on how friendly/ close you are with someone. Some people are not too happy at this part of EdgeRank as it creates Filter Bubbles thus a self fulfilling prophecy however I spent an entire post waffling on that one hence I will now shut up.

Edge Weight – Certain pieces of content are more likely to appear in news feeds than others e.g. photos, videos and links.

Time Decay – Due to the reliance on the temporary nature of content/ news the newer something is the more likely it is to appear.

How It Works – Theory & Application

Edge Rank example 1 (simple):

The fan in the past has clicked on your messages and engaged with your page several times.

There is a good degree of affinity with them hence the affinity score might be 5.

There might be a link in the post that might give an edge weight score of 3.

The post is a day old hence the time decay score for that might be 2.

Thus total edge score (there is only one edge in this example) is 5 X 3 X 2 = 30.

If your score is higher than everyone else’s score then you appear at the top of the News Feed (and then a sliding scale if you are 2nd, 3rd, etc)

Edge Rank Example 2 (more advanced):

The fan from the last example made a comment on the post you made.

The fan’s friend logs in. Will the friend see the post?

In the last example there was no additional edge/ engagement to include. However by commenting on your post the fan added an edge.

As a result Facebook needs to rank two edges.

First edge:

Affinity between friend and your brand = 1 (poor)

Weight of your post = 3 (same as in the last example)

Time decay = 2

Edge score of first edge = 1 X 3 X 2 = 6

Second edge (fans comment):

Affinity between friend and fan = 7 (good)

Weight of a comment = 5

Time Decay = 4 (due to relatively quick log in after fan comment)

Edge score of second edge = 7 X 5 X 4 = 140

Total object score = 6 + 140 = 146

Facebook is likely to show this post in the friends top news feed however it all depends on the other objects it is competing against.

Note to the above figures: The above workings are based on detailed analysis from Kluriganalytics.com and my own interpretation e.g. Applying methodology and detailed understanding of SEO PageRank and PPC Quality Score. Facebook do not want to let you in to the inner workings of EdgeRank hence they have only given a high level view of its algorithym. As a result whilst the above should be pretty accurate I’m not putting my je risque gros, c’est moi qui prends tous les risqué

‘How do I optimise for EdgeRank’ I hear you exclaim?

Truly understand the above as a starter for ten.

Ascertain what your campaign goals are, decide on KPI’s, make sure you have strong and engaging goal messages and then optimise for high CTR/ engagement to get new people to your page to interact.

Test, measure and learn with multiple ad images, call to action messages, etc to maximise EdgeRank.

You essentially need people commenting, interacting with and sharing post. ‘Simple’ things like engagement via answering questions and/or commenting on it produces a lot of edges. Other key advice includes maximising CTR to aid placement by constantly optimising your headlines (do not think in a PPC styley; instead think question headlines) and creative (that thumbnail photo really counts) and think short punchy ad copy rather than War and Peace marketing blurb.

If you are serious about FaceBook advertising its also worth using the Edge Rank Checker (although treat it with a pinch of salt as it cannot measure the Affinity part of Facebook’s algorithm)

What next for Facebook Advertising?

Much of the same with its reliance on EdgeRank yet with ongoing tweaks.

DSP’s/ RTB’s will be introduced soon (See Forbes article).

On the topic of DSP’s they still need to develop hence see my DSP article on their future/ the future of Display on FaceBook.

Facebook will ‘borrow’ ideas from Bing and create their own Social SERPS thus creating its own advanced Social PPC platform

On the topic of Social PPC they have just started to toy with this area via their recent development of search bar ads

And finally to patch up their Achilles heel, mobile (see a nice eye tracking study here that highlights part of the problem), they have to tackle Mobile advertising in a bespoke way. Again see my article on Facebook Native advertising development


About the Author

Global Digital Director. Interests include: my family/ friends, new technology, Martial Arts, cycling, sport in general, God & loving life.



Back to Top ↑