More than 70k videos appeared in YouTube’s ‘trending’ section last year. That’s the tab that spotlights videos that YouTube’s algorithm thinks are currently taking off, and it can drive significant traffic to them.
So how does it work? Data scientist Ammar Alyousfi has been doing the hard work, analysing the data from those 70k videos to identify interesting patterns.
Among the findings: some videos appeared on the trending list when they only had around 50,000 views; most trending videos have titles between 36 and 64 characters long and descriptions between 500 and 1,500 characters long; on average videos take around one and a half days to appear on the trending tab for the first time; the average number of tags used is 21; and the most common objects in thumbnails for trending videos is a person.
Some of this may sound obvious, and some of it may simply reflect general trends on YouTube – a comparison of how trending videos differed from non-trending ones would be required to understand that.
Still, in an analysis of this depth, there’s plenty to consider, so Alyousfi’s work may be handy for music marketers looking to give their videos the best chance of trending.
Sometimes I wonder… how relevant certain data are. For instance the length of a title (36 and 64 characters long). Could it be that NOT successful videos have also a title between 36 and 64 characters long? My point is that nearly nobody analyses non-trending videos and compare the data.
For instance, wouldn’t it be more relevant to analyze other intrinsic aspects of the video itself, such as: is it black and white? How many camera positions? How many seconds are between camera changes in the perspective? Is music all the time? Is any dialog? Is there any “message” at the end?