Facebook adjusted their News Feed algorithm this week, and apparently it’s a significant enough change to warrant a news update.
Every time Facebook tweaks the News-Feed-algorithm-formerly-known-as-EdgeRank, people get concerned that their content is going to vanish off the face of Facebook.
It seems that for most brands, that won’t be the case. So let’s break down why.
What does it mean?
Facebook is trying to reward the content which users want to see on top of their news feed, and give less presence to content where it detects brands are artificially spiking engagement or response.
The key quote from the Facebook news release is: “Pages might see some declines in referral traffic if the rate at which their stories are clicked on does not match how much people report wanting to see those stories near the top of their News Feed.”
That means there are posts that are going to generate good “engagement” — meaning clicks, likes, comments and shares — while still being penalized with fewer impressions.
How Facebook is determining which posts to penalize isn’t entirely clear. They mention a Feed Quality Panel of over a thousand users. A survey of tens of thousands of people each day. And a 5 star rating system. Thus they’re taking a sample of Facebook posts from which they’re learning things about content trends that would indicate something ingenuine about the post’s initial appeal.
The question is, What’s the post characteristic Facebook can detect in a survey and map to all similar posts to reduce the impressions those posts receive in the News Feed?
There are numerous post characteristics on Facebook. Some that Facebook already classifies posts by, including media type, time of day, and day of week. Of course, those would be far too blunt a measure by which to reduce posts globally.
My guess is that Facebook is using the survey data to identify common text and character strings employed by marketers and publishers to artificially spike engagements. Facebook would then look for posts that contain similar elements, and downgrade them accordingly.
What types of text or word strings that could be? Think about a fair percentage of the headlines you see online. Over-hype, over-promise, and a general sense of deception. Maybe the word “free” in all-caps with a few exclamation points might be a good example. Maybe anything sounding like “And you won’t believe number 7”.
Their system could also employ image detection, among other possibilities.
What should you do?
Facebook seems to be wanting to keep pages from being overly promotional in their content, driving users off Facebook and onto their website with sensationalist headlines.
Thus, for posts with a lot of clicks, likes, shares and comments, Facebook is using user ratings of whether they would like a post to appear at the top of their feed as a way of qualifying that those links are of genuine interest, and not just a gimmick or overtly promotional.
So if you’re not resorting to gimmicks, Facebook says you shouldn’t have anything to worry about. And if you are using gimmicks to goose your content, you should probably start figuring out more genuinely appealing ways to increase engagement.
How can you analyze your content for this?
The low ratings that lead to a drop in impressions aren’t available to page admins as opposed to the engagement numbers which are available from Facebook Insights.
However, it is possible to identify posts that might have been given reduced impressions due to low ratings or another negative factor like too spammy or look like ads. You need a listing of your link posts with sortable columns for Engaged Users and Reach metrics.
Zuum’s Post Insights report lets you set this up quickly.
Per the chart image below, you’ll want to sort descending by Reach, and then compare the Engaged Users metric. As Reach decreases, Engaged Users should decrease as well. Look for posts where Engaged Users increases while Reach decreases. Those are the posts that may have been penalized in the News Feed and are getting less reach relative to how much engagement they are getting. (Keep in mind this would only apply to posts receiving no paid impressions.)
The reasoning is, under normal circumstances, the better engagement a post gets, the more impressions Facebook will allocate to it, and the higher the reach will be. However, if a post generates high engagement levels artificially and Facebook can detect it and in turn penalize that post, then Facebook would then be reducing the impressions for a post getting good engagement levels, and that post would then have lower reach.
Any thoughts on how you think Facebook is identifying posts they’ll penalize, and what metrics might reveal that impact?
Post Insights Chart
If you’d like to see how your Facebook page content compares using these metrics, contact us at Zuum.