What publishers post on Facebook
I took 32 publishers on Facebook representing some of the most-followed publisher pages across a variety of industries and analyzed all their posts since the start of 2015. That’s 742,490 posts. Here are my findings.
On average, publishers are posting twice as many posts per month in 2017 than they were at the start of 2015.
Use of in-feed video has grown four fold. In-feed video as a percentage of total posts has increased from 4% in January 2015 to 21% in March of 2017:
From the graph above, you can see that barely any posts are statuses (plain text). The number of photos posted per month has decreased 55% since the start of 2015.
Despite the growth of video, links still represent the majority of posts, with 75% of publishers’ posts on Facebook linking to a third-party page (e.g. the BBC sharing a link to bbc.com via Facebook):
Sidenote: ‘Video’ only includes in-feed video. Publishers routinely share video on their own web pages, which would be a post categorized as ‘Link’.
Despite the desire of publishers to share links to their own pages, links perform poorly when it comes to in-feed engagement. On average, videos get 16x more shares than links and images get 6.7x more shares than links:
But that’s not to say links don’t perform better in non-Facebook engagement categories like page views or ad revenue generated.
When we remove outliers from this data (i.e. a handful of posts that went viral with very large share counts), the number of shares and comments of in-feed videos vs. other post types decreases to the same level as images. This suggest that videos are more likely to go *very* viral than images or links.
Post message lengths have been increasing a little since the start of 2016, shown by the average number of characters per post:
The average video or photo is accompanied by about 105 characters of text whilst links are accompanied by just 85 characters.
Average video length, however, increased a lot during 2016:
Today, the average in-feed video shared by publishers is 179 seconds, up from 79 seconds a year ago, in April 2016.
Standard deviation for video times is very high, however, at 340 seconds. 75% of videos shared are less than 89 seconds.
So video is getting longer, but is that a good thing? As video length has increased, the number of shares, reactions, and comments for in-feed videos have all dropped:
In the graph above, video length is drawn faintly in gray. Whilst there may be many factors causing this decrease in Facebook engagement metrics for video (video saturating feeds, higher use of phones on cellular, etc), there is a negative correlation between video length and in-feed engagement when videos are longer than about 140 seconds.
It seems like the optimal video length for publishers to get maximum engagement is about 140 seconds.
It’s no secret that Mark Zuckerberg believes Facebook will be mostly video in five years. However, this data suggests that publishers are still resistant to the idea of in-feed video. Publishers are still focussed on driving users to owned webpages. In-feed video may be a great way to raise awareness of the publisher’s page and attract more followers.
I’ve got more data to come in this space, including a brands-only study, so stay tuned!
The raw data
I collected this data using Facebook’s Open Graph API. You can collect this data yourself and analyze it using the code written for this study in this Github repo.
Cautions about the data
- I do not have any data on how the size of each publisher’s audience has changed through time.
- I do not have data on the content on third-party pages shared via links. Some of these pages may include video, which would be counted as a link, not a video, in this study.
- I also do not have data on live videos, which are excluded from this study.
- Other types of Facebook posts, event and note, were excluded from this study because they were used so little — usually not at all — per time period studied.
- There are many other variables, like device type (e.g. mobile) and current events (e.g. the US election), which may have caused some of the underlying behaviours of consumers on Facebook to change temporarily of permanently.
- The publishers are all Western, suggesting that the data here better describes US and UK consumer habits than people in other parts of the world.
