I read a great arinitticle yesterday written by Alexis Madrigal in The Atlantic which attempted to decipher how YouTube recommendation works.
The article describes a study of 175,000 YouTube sessions which involved picking a random video, then a random video based on the top five YouTube recommendations suggested after that video, and so on for a few more times. The idea was to see the types of videos being recommended.
The top fifty videos recommended had been viewed hundreds of millions of times each with only 5% of the recommendations going to videos with fewer than 50,000 views. Put simply, popular videos will get more recommendations (social proofing in action!) but if there is an early spike in views, more traffic can be directed to a video. One other finding in the study is that the more videos viewed in a session, YouTube is likely to recommend longer videos.
This study has some limitations as it doesn’t account for a user being ‘logged in’ and learning a user’s viewing preferences over time but there are perhaps a few takeaways here for DIY musicians.
1) Getting traffic early to your YouTube videos once live will help it being recommended so a large YT subscriber base will help. Once live on YouTube send an email to your newsletters, tweet a 30 seconds clip with a link to YouTube to get an intitial momentum. You might even think about using paid traffic on YouTube itself.
2) If longer videos are being recommended, do you have any longer form content on YouTube that can be served/viewed?
Read the article in full here.
Got any questions about this post or how you can better market yourself?