Facebook Pages that automatically publish content to the news feed through third-party apps such as HootSuite, TweetDeck, and Networked Blogs receive an average of 70% fewer Likes and comments on their posts per fan, according to a new study by Applum, developer of Page tool EdgeRank Checker. The study says the difference is likely due to Facebook reducing the prominence of posts published by third-party APIs, and Facebook collapsing updates from the same API from across a user’s friends and Liked Pages.

As Likes and comments increase a post’s prominence in the news feed — thereby driving more impressions and clicks — all Pages using auto-posting apps should look to switch to manual posting if possible.

Many companies, public figures, organizations, and news outlets (including our own) use auto-posting apps to create Facebook Page updates by syndicating their Twitter posts or converting their blog post headlines. This increases efficiency by relieving the admins of these accounts from having to copy and paste headlines and links from one platform to another.

The practice is subjectively considered  sub-optimal, though, as different platforms have different publishing capabilities and norms. Facebook for instance allows for rich media posts, so authors can select a thumbnail image and caption along with posting a link and headline. It’s typical for Twitter accounts to post up to a dozen times a day, but that volume could be viewed as spam on Facebook. Therefore, auto-posts can appear robotic and less compelling.

EdgeRank Checker’s study is the latest of several reports we’ve covered that reveal obstacles to engagement on Facebook. A recent PageLever study showed that the average Page gets only 7.49 daily unique news feed impressions on its posts and only 3.19 daily unique Page views per 100 fans.

EdgeRank Checker’s Data

Now, EdgeRank Checker has revealed empirical data that automatically published posts perform worse than manually published ones. EdgeRank Checker analyzed over 1,000,000 Facebook updates by more than 50,000 Pages with a combined reach of over 1,000,000,000 fans including duplicates. It then calculated the engagement ratio of the total Likes and comments on a Page’s post divided by the total fans of the Page at the time of the post for the ten most popular third-party publishing APIs.

The study determined that compared to the engagement of posts published manually to Facebook’s web or mobile interfaces, the reduction in engagement ratios of the top third-party publishing APIs are:

• HootSuite – 69% reduction
• TweetDeck – 73% reduction
• Sendible – 75% reduction
• Networked Blogs – 76% reduction
• RSS Graffiti – 81% reduction
• Twitter – 83% reduction
• Publisher – 86% reduction
• twitterfeed – 90% reduction
• dlvr.it – 91% reduction
• Social RSS – 94% reductions

These averaged out such that posts published through a third-party auto-posting app saw roughly 70% fewer Likes and comments than those published through Facebook’s first-party interfaces. This is in part due to Facebook’s direct punishment of the EdgeRank of posts by third-party APIs. Also, if a user’s news feed contain multiple posts from a single API, whether from a single author or several different Pages and friends, the posts are collapsed and must be unfolded to be seen.

The takeaway point of the study is that it is well worth it to take the extra minute to manually craft a Facebook post rather than auto-post. This might require changes to workflows or allocation of additional human resources. Still, Page owners could be sacrificing a lot of their social media performance and return on investment to save a small amount of time and effort.

Update 9/9/2011: We’ve now learned that Facebook maintains a secret whitelist of companies that are exempt from having content posted through their publishers consolidated across different Pages and clients. This protects them from a reduction in news feed impressions. The whitelist includes some top enterprise Page management tools from the Preferred Developer Consultant program including Buddy Media, Vitrue, Involver, Context Optional and Syncapse. Read more.