Does Facebook Punish Third-Party Clients?
"Using a 3rd-party API to update your Facebook Page decreases your engagement per fan (on average) by about 70%," a blog post at EdgeRank Checker warns. The data behind that claim: EdgeRank Checker's own analysis of more than a million updates on 50,000+ public pages.
Does that mean any given post would do 70 percent worse via a third-party client than if made directly on Facebook.com?
And EdgeRank Checker is refreshingly clear about this in the infographic they posted.
"Our study illustrates that we've found a correlation; however, not necessarily a causation," the infographic cautions. "Engagement Ratio can vary for a variety of reasons ... random variations in Pages sampled, quality of Facebook campaign management and disparity in sizes of the Pages sampled."
Why then does the blog post declare that using a third-party tool will decrease engagement, even while admitting there could be multiple reasons for third-party clients showing up so poorly?
The blog post gives a clue: EdgeRank Checker, a service that aims to measure Facebook content's "EdgeRank" (importance), believes that Facebook "punishes" 3rd-party APIs like HootSuite and Tweetdeck by not ranking posts from them as highly as those made on Facebook.com. This would make such posts less visible in some users' feeds, giving that content less of a chance to be liked or commented upon. "We ran a study across our entire data set to provide evidence that our hypothesis is true," the post says.
It's hard to admit your numbers, while compelling, can't be deemed conclusive if the whole reason for the study is to, well, be conclusive.
My preferred conclusion? "Facebook posts from third-party tools were found to have a substantially lower engagement rate than those made on Facebook.com, although a number of factors could be responsible for the difference." Not quite as attention-grabbing, though.
What factors? The blog post itself starts us off:
We have a few theories explaining the results:
Facebook Penalizes 3rd Party API's EdgeRank
Facebook Collapses 3rd Party API Updates
High Chance of Being Scheduled or Automated
Content is not Optimized for Facebook
A lively discussion thread on the blog offers some more:
- Posts using third-party tools may be "more likely to be absentees on Facebook and less engaging."
- Posts via tools may be less likely to include embedded photos or videos, which could drive more engagement. (EdgeRank Checker's Chad Wittman responded by parsing the data for text-only posts and said there was a 73% difference between Facebook.com and HootSuite text-only statuses; links had a 46% difference.)
Let me add a few:
- Length of posts may differ between the two groups
- Time of day may be different
- Certain characteristics of pages using Facebook.com vs third-party clients may be different: size of site, staffing, commitment to Facebook engagement, etc.
Remember, a study could find that coffee drinkers have higher cancer rates -- without realizing that a higher percentage of coffee drinkers also smoke. If you don't screen out other contributing factors, you're at risk of jumping to inaccurate conclusions.
The encouraging news here: Anyone who reads the full post and comments on EdgeRank Checker's site will get a complete picture of the data's limitations, despite the initial somewhat overhyped "decreases your engagement 70%" claim.
That's not always the case as the story gets reported around the Web, though.
For example, Inside Facebook says the 70 percent number "is in part due to Facebook's direct punishment of the EdgeRank of posts by third-party APIs" as well as the problem that updates from a tool are more likely to be collapsed. Inside Facebook's report has no mention of EdgeRank Checker's other theories, though -- greater chance of automation and non-Facebook-optimized content -- or the correlation-isn't-causation warning.
Bottom line? I happen to agree that you're better off posting on Facebook's Web site than via third-party tools when possible. Doing so tends to encourage more best practices, whether or not third-party APIs are being "punished." But there's still a difference between evidence that appears to support a theory and data-backed proof.
Sharon Machlis is online managing editor at Computerworld. Her e-mail address is firstname.lastname@example.org. You can follow her on Twitter @sharon000, on Facebook, on Google+ or by subscribing to her RSS feeds:
articles | blogs .