Following on the results of a study carried out in July 2019 by Nonli’s Data Science team on Facebook engagement excluding clicks (reactions, comments and shares) generated by all types of posts and based on the data analysis of posts published between 2016 and the 1st quarter of 2019 by 100 pages of the largest E-commerce companies doing business in France, we have decided to condense its results.
The fact that this document specifically focuses on Facebook posts is justified by 2 main elements:
- the #1 social network in France is Facebook with 22 million daily active users;
- Facebook is the common social network to all the e-shops in scope and the one on which the activity is high enough to develop a statistically significant analysis.
On average, the volumes of posts published were relatively similar between 2016 (284) and the first quarter of 2019 (64, i.e. 256 for the whole year by extrapolation).
The 3 most published types of posts were photos, videos and links, with a majority of post type photo that represented between 61% and 70% of publications each year over the period in scope.
When looking into the engagement data of the period in scope, we notice that the median engagement per post decreased from 248 in 2016 to 131 for the 1st quarter of 2019 - accounting for less than 1 daily publication on average.
The decrease in engagement year over year can be explained in part by the decrease of the reach on Facebook.
The publication types generating the most engagement are:
- Video: 359 in 2016 to 310 in Q1 2019 - for less than 0.1 post/day per page on average;
- Photo: 272 in 2016 to 126 in Q1 2019 - for less than 0.5 post/day per page on average;
- Link: 119 in 2016 to 20 in Q1 2019 - for less than 0.1 post/day per page on average.
The daily engagement per piece of content increases with the number of fans of the page that published it: the more popular a Facebook page, the more engagement its published content will generate.
The question of volumes also arises for the number of publications here.
We notice a decrease in the median normalized engagement per post when we look at all the pages aggregated,the engagement tends to decrease by half between the 1st and the 6th post (very few of the e-commerce companies in scope publishing beyond that, the numbers corresponding to higher volumes are not statistically significant).
There are 2 things worth noting:
- this evolution is not universal, certain pages experience a decrease in engagement after the 1st post, others an increase or steady engagement, if not variations;
- on the same sample of pages, the total daily engagement increases (the more we publish, the more the daily sum of engagements increases).
When analyzing all consecutive publications with gaps of less than a minute or over 30 minutes between 2016 and Q1 2019, with equivalent numbers of fans, we could notice that the gap between two publications had an impact on their engagement - two publications published with an interval of at least half an hour experience a higher engagement than those published with a very short interval.
Regarding the time and day of publishing, a granular analysis of different e-shops allowed us to demonstrate that there is no universal truth in terms of best practices. We could observe that the trends from that standpoint depended on the pages analyzed.
We have also observed that not only the times and days of publishing varied from one page to another, but also that the same page experienced variations from one year to the other. Thereby, where the engagement is higher on Sundays for a given year, it will be Thursdays and Sundays for the following ones. Same goes for times, a page that engages more between 3pm and midnight for a given year, will experience a higher engagement between midnight and 10am the following year.
Engagement is therefore highly dependent on the audience.
Certain results are also victims of self-fulfilling prophecies: certain pages never or very rarely publish during certain periods, and therefore their engagement is lower, when not nil, during these time frames.
Over the 4 sub-periods considered (2016, 2017, 2018 et Q1 2019), we have aggregated the results and created 2 rankings.
The first ranking is a top 10 by total engagement of all the Facebook pages belonging to the e-commerce companies in scope in our study.
The results are justified in part by the volumes of publications of the pages that appear in this top 10.
The second ranking is a top 10 by total normalized engagement, and therefore compares the Facebook pages in scope considering them as if they all had the same number of fans, providing a different perspective, where their engagement is relative to their number of fans.
On Facebook, the impact of the decrease of the reach on the engagement has in general not been compensated by an increase in the volume of publications by e-commerce companies.
Besides, despite the fact that the type video publications have been those that generated the most engagement, they represent a minority of publications. Although they have increased from a 1:10 to a 1:5 ratio compared with type photos between 2016 and Q1 2019, their experienced an increase in engagement on the same period, from 32% to 146% higher engagement than photos.
The total engagement grows with the volume of posts, nevertheless, unlike our study on the same subject for the media, the absence of data passed 6 posts does not allow us to conclude beyond that.
Here are our recommendations:
- test the times/days of publishing - posting in off-peak hours and days, those can end up being unexpected pockets of audience;
- the reach keeps decreasing, therefore it is necessary to publish more at the risk of experiencing dropping engagement;
- multiply the presence and intensify the activity on different social networks, in particular those where your target audience is or when the subjects apply.
*the full study is available here