Thursday, December 8, 2011

Solving the mystery of Edgeranking on Facebook

When you're managing your facebook page, every time you create an update, you have an edgerank. That object will have a single edgerank. The new news feed on Facebook impacts how that all rolls out. EdgeRank is calculated using affinity x weight x time decay. Here are some interesting thoughts from EdgerankChecker co-founder Chad Wittman.

Author Archives: Chad Wittman
Did Facebook Really Fix The 3rd Party API Penalty? Posted on December 1, 2011

In early September we decided to tackle the rumor that 3rd Party APIs (using a platform to publish other than had some sort of penalty affecting EdgeRank. Our article caused quite a stir reaching many of the major tech news outlets. We heard many social marketers concur that they have been seeing similar affects on their data. A few weeks after our report, Facebook overhauled the news feed (correlated with the recent f8 conference). 3rd Party APIs developers submitted a bug to Facebook detailing an issue that seemed to severely affect Impressions and Engagement. Facebook acknowledged the issue and vowed to fix it. This seemed to somewhat validate that there was some sort of effect impacting 3rd Party APIs. Approximately 1.5 months passed and Facebook updated the bug and proclaimed that it had been fixed.

How was the data analyzed?
Facebook announced their bug fix on November 6th. We grabbed data from the two weeks before November 6th vs the two weeks after November 6th. The sample size included over 4,000 Pages that posted during both of those time frames. We took every Page that posted over this time period and averaged their Post Impressions (per Post per fan at time of Post). We then noted how the Posts were published (directly from Facebook or 3rd Party API). In order to fairly assess how the bug fix affected each Page, we needed to only include Pages that posted directly on Facebook and used a 3rd Party API during our time frame. With this in mind, our data was most accurately analyzed by only looking at the most widely used 3rd Party API (in our data set). After imposing several rules to accurately compare the data, our segmented sample size was 391. To clarify, we required a Page to have posted with a 3rd Party API and directly on Facebook once each, during each time frame compared.

No comments:

Post a Comment