SOCIAL DASHBOARDS ANALYSIS
Did TV and Movie Studios get Hammered by Facebook?
Prepared for: Jason Keath, SocialFresh.com firstname.lastname@example.org Prepared by: Dennis Yu BlitzMetrics email@example.com
December 14, 2012
Movie studios and TV shows have been complaining of suddenly losing traffic at the end of September. So we decided to investigate by pulling insights on 137 of these pages, covering the 30 days before 9/20 (the date that many reference) and 30 days afterwards.
Various social analytics players have drawn dramatically different stories on seemingly the same data — some cry foul, while some are saying nothing has changed. Facebook has been claiming that nothing significantly has changed and has rebuffed naysayers by claiming their data is incomplete. So what’s the real deal here?
Surprisingly, we found that small to medium-sized pages actually got a significant INCREASE in newsfeed exposure and engagement, while the large pages (over 500k fans) were significantly penalized. Many reasons for this, which we’ll cover shortly. And to no surprise, those pages that ran ads fared better than those who didn’t run ads. They had more organic impressions, too. We know that running ads gives an organic algorithmic boost, in addition to the fact that a larger base, all else being equal, will get more impressions.
FIGURE 1 The small and medium-sized pages experienced significant increases in impressions, while clearly, the largest of pages had a decline. Before you draw a snap conclusion here, note that the largest of fan pages have a very different audience base than smaller pages that have their core, devoted fan bases. The large pages might have done heavy advertising or contests a long time ago, leading to a current fan base that is largely inactive. Some of the medium-sized pages are new TV shows at the start of the season, getting high engagement on a new fan base. On Facebook, we are competing in the news feed against the friends of our fans — a high bar for being more relevant and engaging. It’s a lot easier to do this on a small base than trying to be specifically relevant to all fans of a South Park, for instance.
FIGURE 2 The pages that grow the fastest typically tend to be the ones that run ads. And TV/movie studios rarely run Facebook ads. When they do, it’s during campaign periods, as opposed to the lightweight, “always on” layer that Facebook recommends. Low engagement leads to low fan growth here, too.
FIGURE 3 Impressions per fan mimics the change in total impressions, since fan bases overall didn’t change much. However, for pages that may have doubled in fan base size, but kept the same number of impressions, we’d consider this a loss, as impressions per fan would have fallen.
FIGURE 4 This is interesting. The number of stories generated per page dropped only single digits, compared to a more significant drop in impressions. This means that Facebook is doing a better job in choosing what to show in the news feed. In fact, most of the impressions they removed didn’t have a significant impact to engagement. As there is more content being generated by more users and more pages, we’d expect that each page would get a smaller share of a user’s attention. So the drop in page stories doesn’t seem unreasonable.
Fan base -
Start Date End Date
9/21 10/21 page_fans
8/20 9/20 page_fans
50,000 34 25,349 24,043 We took a random selection of 31 TV studio pages from networks that we have admin 188,504 access 50,001 500,000 50 176,877 to, comprising a total of 243 million99,999,999 fans. This covers 481 million impressions, 157 Facebook 500,001 63 7,642,223 7,772,088 million consumptions and 34.5 million page stories. Impression per fan
-5.2% 6.6% 1.7%
Pre Post (+ vs. -) Impressions End of the full / Date The 31 TV shows are a subset avg. Fans 137 10/21 Start which includes movies, news, and other pages, Date 8/20 9/21
entertainment pages. 50,000 Fan base34 50,001 500,000 50 500,001 99,999,999 63 50,001 Impressions Total 500,001 147 Per Fan
50,000 500,000 99,999,999
34 10 50 24 63 41
0 0 39 0 106
% 10/21 27 79.4% page_fans 40 80.0% 61.9% 72.1%
1 0 0
60% -30% -38%
Post impressions Date Start Impressions
8/20 9/21 Increased decreased End Date 9/20 10/21 Having 3 3 Fan base paid impressions Sample page_posts_impressions page_posts_impressions Having No paid impressions 3 11,502 22 50,000 34 17,498
% 50% 88% 52%
50,001 500,000 50 40,933 30,457 -26% 500,001 99,999,999 63 2,804,699 1,777,365 -37% The biggest hit came in September when Facebook made a change to its algorithm “EdgeRank” which
Start Date End Date
Pre decides which content users most want to see in their Newsfeed. The falloff forPost Hollywood studios has been Impressions
8/20 9/21 Sample 9/20 page_posts_impressions_organic 10/21 page_posts_impressions_organic
7,226 61% 50,000 For example, 72% of movies and34 network TV shows experienced a drop in11,661 the number of people 50,001 their Facebook posts, 50 500,000 38,918 27,820generated during -29% who saw based on the 9 billion impressions these posts 500,001 99,999,999 63 2,747,212 1,689,276 -39% the 60 day period.
Pre Post Impressions -Paid End Date 9/20 10/21 pages saw a reduction in the number of “engaged” users — people who click on a post, share it with
That decline took a toll on another Date watch closely: engagement. 23% of the biggest studio Start factor marketers 8/20 9/21 friends, or write a comment — because of decreased exposure in the Newsfeed, BlitzMetrics found. 34 3,860.1 5,081.4 32% 50,000 50,001 500,000 50 478.9 903.5 89% 500,001 99,999,999 TV posts 63 17,077.6 online friends fell 39%. 77% of the 73,311.6 329% The number of movie and Facebook users shared with their large pages saw a drop in reach (unique impressions) of 45%, but the decline in engaged users
Fan base Sample page_posts_impressions_paid page_posts_impressions_paid
Pre Impressions was only 23%. At the time, the number of paid exposure to new studio postsPost than tripled – suggesting more
-Viral Hollywood has begun to pay to getDate End noticed on the platform 9/20
Fan base 50,001 500,001 50,000 500,000 99,999,999 Sample
9/21 10/21 page_posts_impressions_viral
34 50 63
416 1,536 40,409
756 1,733 14,777
82% 13% -63%
Fan base -
Start Date End Date Sample
8/20 9/21 9/20 10/21 page_posts_impressions_unique page_posts_impressions_unique
50,000 500,000 99,999,999
2,000 7,671 746,254 firstname.lastname@example.org
5,038 152% 5,180 -32% 414,057 -45% Facebook.com/blitzmetrics
What does this means?
No surprise: Just like we saw with the TV shows, the biggest studio pages got hit the hardest. They lost impressions and interactions. Interesting: Medium-sized pages actually grew their impressions. This is less about medium pages being “better”, but more about the pages that grew to be large tend to have a more diffuse audience, which is harder to target, please, and activate (especially with competition everincreasing to show up in the newsfeed). The medium-sized pages are ones where people became fans despite whatever the studios did — usually meaning no ads. • The newsfeed is more competitive because there is more content vying to your audience's attention, both organically and paid. Nobody can tell to what degree this is a content increase vs ad inventory crowding out the feed, since only Facebook has access to total newsfeed impressions. We can't prove that Facebook is monkeying with the newsfeed.
• We do know that 6 months ago, ads in the newsfeed were limited to one per day per user, and prior to that, didn't exist in the newsfeed at all. Today, the newsfeed ads are at a frequency as high as 1 in 15 stories, though it varies by user and type of device. There are an increasing number of ads on mobile, but nowhere near as much as most advertisers would like. • The trade-off of giving users what they want vs satisfying advertiser demands is a tricky one, endemic to all ad-supported models. Hulu started out great because people could watch full TV episodes with only a few short ads. Now users are complaining, because the share of ads to play time now approximates that of set-top boxes. Facebook has said that negative feedback decreases page reach, so that if people are marking your posts as spam or hiding them, then you'll get penalized. While this may be true and increasingly a factor, this does not explain an across-the-board decline across thousands of pages that we monitor.
Further: Movies tend to be on or off. They are heavily promoted prior to release and then abandoned. TV shows will have lasting or even building engagement over a season. The good news is that Facebook is doing a better job choosing what content users are most likely to interact with.