In Social Video We Trust (or Not)
Facebook has hit the headlines again for the wrong reasons.
This time it’s because of the accuracy of its own reporting of views of video on the platform.
It turns out that over the last two years, it has over reported video viewing times by between 60 and 80%. That’s a huge discrepancy.
Apparently it failed to factor in any video views below three seconds. Anyone who has used the site on a mobile device and the vast majority of Facebook users do, will understand why this makes a big impact: autoplay and fast scrolling on a small screen can occur frequently. Failing to report these as views, over inflates viewing times.
Given the fact that GWI reported that nearly two thirds of Facebook users have not posted any update in the last three months, over 1 billion users (yes, 1 billion) do a lot of scrolling.
So it amounts to a whole lot of over inflated views.
Facebook’s answer to address this? Change the name of the video view metric from ‘average duration of video viewed’ to ‘average watched time’.
Bingo! Everything is ok then.
This highlights the problem that Sir Martin Sorrell has highlighted in measurement: You can’t have the player be the referee as well.
Third party independent measurement can only answer this and not one of Facebook’s Instant Measurement Titles (my own emphasis).
Facebook has been changing newsfeed algorithm resulting in the reduction of prominence of external publishers’ posts. You then have to pay you know who to regain that prominence in the feed. Result? — Instant Revenue.
In fairness to Facebook, it maybe a social network, but it is a commercial social network.
Facebook’s most recent algorithm changes affect the prominence of news and media sites prominence once more in that it wll highlight friends and family above all else.
In fairness to Facebook, it is a commercial entity, but it is a social network also.
Facebook earned $17.08 billion in ad revenue in 2015, up 49% year over year, as advertisers sought the attention of its 1.7 billion active users.
Turns out only a third of them are really active (i.e. posting) and even then based on this revelation, it is hard to know what the average duration of video viewed/average watched time of the video ads they saw was anyway.
