Facebook algorithm: How the shift in engagement can favour newsrooms
Facebook’s algorithm change to favour content from ‘Friends and Family’ has reduced the impact of news publishers posting on the platform. Articles brought to Facebook by individuals now account for the majority of all news engagement. How is this a good change for publishers?
In this guest post, we intend to present information regarding Facebook engagement trends over the past year along with some potential takeaways for news publishers.
Before we start this summary of our analysis we need to describe the data we are looking at, most importantly describing the difference between the two key Facebook metrics we have studied.
Post Engagement versus Web Shares
You’re most likely already familiar with Post Engagement, the most widely understood type of engagement on Facebook.
When a publisher makes a post on one of their Facebook pages, individuals may see this post in their Newsfeeds and engage with it there. Post Engagement covers reactions, likes, comments and shares of that Facebook post.
If Post Engagement is a record of engagement prompted by the publisher, then Web Shares can be thought of as a record of engagement prompted by individuals.
When an individual visits an article on a publisher’s website, they may decide to share this article to their own Facebook feed. This is done via the social share buttons on the publisher’s website, or copying the url and pasting it into a new post they are creating on Facebook. Web Share engagement includes counting this initial share, along with the engagement (likes, comments and shares) on the post made by the individual.
The key difference to the user is that the post they interact with is delivered by a friend, not from a news publisher. This type of sharing was prioritised in Facebook’s major newsfeed algorithm change in early 2018.
Over the past few years, we have observed a steady increase in Web Shares as a proportion of total engagement. Taking a previous sample of France as an example we can see that in 2016, almost 80% of all engagement from the top 3 publishers came from Post Engagement. Mid way through 2018, the ratio was 60:40 Post Engagement vs Web Shares.
To take a deeper look with a greater sample size, we gathered data from the past year across 5 different countries, focusing on the top 12–15 news publishers in each territory. The following charts show how their engagement split between Post Engagement and Web Shares looked like over the period.
The immediate takeaway from this is that Web Shares now account for well over half of all Facebook engagement in all countries studied, aside from France where it currently sits at a shade under 50%. Across the other four countries, Web Shares make up between 62–70% of total Facebook engagement.
We did not expect Web Shares to make up such a large proportion of total engagement so soon.
Is this better for news publishers?
In short, yes. Over the last few years, the top performing publishers in terms of Post Engagement have been viral or tabloid publishers like Lad Bible or the Daily Mail Online. These types of publishers have been willing and able to optimise for Post Engagement to a greater degree than more traditional news publishers. This put traditional news publishers at an apparent disadvantage on the platform.
Facebook’s much publicised algorithm change to favour content from friends and family and to incorporate ratings from Facebook’s unofficially titled ‘trust index’ are likely the key drivers behind the acceleration of this ratio change.
For news publishers, it means that website performance has greater significance than optimising for engagement on their Facebook page. While both are still important, success in terms of generating traffic is now less about what the publishers post themselves on Facebook, and more about the behaviour of individuals away from those pages.
Let’s take one example from the UK: BBC News (a traditional news publisher with a strong website) and the Daily Mail Online (a tabloid who have been extremely successful with Post Engagement). We can see in the graphs below that in terms of Post Engagement (top), total numbers are very similar.
In contrast, when looking at Web Shares (bottom) the picture looks completely different.
Around the end of May 2018 you can see a significant divergence in Web Share trends. Over the last year, BBC News generated about 3.5 times as many Web Shares as the Daily Mail Online.
How significant is this change across the Facebook news ecosystem?
Although it was not immediately apparent when the algorithm change was first brought in, we believe that this is the most significant change since we began tracking data on Facebook in 2013/4. By lowering the effectiveness of on Post Engagement, (which will undoubtedly have had a negative effect on publisher reach), and simultaneously increasing the effectiveness of Web Shares, Facebook has introduced a fundamental shift in the way news articles are distributed across its platform. The obvious beneficiaries are traditional news publishers with a strong web presence.
How the Facebook Algorithm Works in 2018 (And How to Work With it)
Since Mark Zuckerberg's 2018 announcement that the Facebook algorithm will prioritize "meaningful interactions" from…
Does different engagement mean different content?
In our experience, articles with many Web Shares tend to be more focused on serious news, whereas Post Engagement favours lighter-hearted and more viral type stories.
Although a larger analytical study would be able to draw out some concrete trends, we analysed the top ~25 posts by engagement over the past year from both the US and UK. Stories were grouped into 4 general categories:
- Serious News
- Light/Viral stories
- Celebrity Deaths
- Other (e.g. Op-Eds, special features unrelated to current affairs)
The immediate thing that stuck out about this data was the popularity of articles regarding a celebrity or person of importance dying. These were so popular in terms of both types of engagement that they warranted a separate category — a quarter of the top stories were about these deaths.
From the US, the data seems to bear out that articles that gather Web Shares tend to be about “Serious News” more often than posts that generate Post Engagement, 50% of stories vs 29.2%.
Viral items and lighter news articles (e.g. Billboard tells liberals “Keep on driving until they leave Texas”) accounted for 20% of top stories by Web Shares, versus 37.5% of stories by Post Engagement.
In the UK, celebrity deaths are also represented in both categories. However, while they accounted for only 9% of top Post Engagement stories, a whopping 32% of top Web Share articles concerned this subject.
There was a much bigger difference in this dataset regarding the lighter and viral stories vs serious news. When ordered by Post Engagement, only 22.7% of the top stories were about “Serious News”, in contrast to 54.5% of top stories by Web Shares.
This suggests a fairly large difference between content delivered directly from publishers vs what individuals choose to bring to the platform themselves. It is worth noting that the dominance of viral type stories in the UK is influenced by the size of the Daily Mail Online. As Post Engagement is sidelined in favour of Web Shares by the Facebook algorithm, it is likely to affect this publisher more than others.
Summary: What does this mean for news publishers?
For more traditional news publishers it appears to be a positive trend. It means that optimising for Post Engagement becomes less important, as it no longer dominates total Facebook engagement.
Despite a turgid time PR wise for Facebook as a company, the health of mainstream news appears to be improving on the platform. Putting more responsibility on users to determine the content that appears on Facebook remains a risk, yet in the data analysed so far, we see that users favour sharing serious news articles over viral content with their friends. If the current trend continues we expect to see “Serious News” elevated further at the expense of lighter and more viral type content.
It is worth pointing out that even with a full year’s worth of data, we have analysed a small sample of publishers in each territory. A larger study of individual countries would be needed to see if the positive takeaways in this article are true across all news publishers (and in fact, all types of publisher) in a given territory.
Based in Helsinki, EzyInsights tracks news and engagement data in real-time for publishers, journalists and news agencies.
All the data in this article was taken from EzyInsights Intelligence, which gathers historical data on over 100,000 Facebook pages worldwide.
For a full list of the data shared in this article, including article headlines used in this analysis, follow this link.
Author of this article is Steve El-Sharawy, Head of Insights at EzyInsights.