Today is Saturday, it is sunny and warm outside. Perfect day to go shopping ! You are obviously going to spend this afternoon with your favorite side-kick. Your mission is to find the perfect outfit that will fit your shape, your personality, and evidently match your preferences.
You finally find these really cool jeans, but you are still hesitant about the top : the blue and black, or the white and gold one ? The retailer comes to you and says that the jeans and the blue and black top would be the perfect match. On the other hand, your friend who has similar tastes as you – so similar you have a couple of clothes in common – votes for the second option.
What would you do ? Who would you trust ? How to make sure which one is the right choice ?
As a matter of fact, when you are browsing the internet, you might encounter the same problem-resolving process.
When a Youtube user surfs on Youtube, they first type the title of the video. On the right side of the video, a section displays other videos. In this section, two types of recommendations are presented :
- Featured / Related videos :
They are based on the Youtube’s product network ; that is to say that they are based on the site recommendation algorithm only.
- “Recommended for you” :
These are based on the Youtube’s social network. In fact, another user marked the video you just watched as favorite, as well as another one. Therefore, the second video will be “recommended for you”.
Thus, which of these videos matches users’ preferences the best ?
In a study, participants were asked to watch videos on Youtube, and rate each of them from one star – Poor – to 5 stars – Awesome. One group had access to the product network only – Related-Featured videos – the second group to both product network and the social network – Dual Network – and the third group to user-generated links only – Recommended for you.
What emerges from the study is illustrated in the graph below :
The fact that the second group curve is the lowest shows that finding a liked video takes less time when using the dual network than any of the other networks. Thus, rather than proposing either one or the other, offering both at the same time gives the user more possibilities, more choice, and therefore, more opportunities to reach the right video.
Therefore, the most efficient way for Youtube to satisfy its users, is to offer them as many choices as possible using different methods : the product network as well as the social network.
So, next time you will go shopping and do not know what choice to make, try the clothes on in the fitting rooms ! – or buy both.
Jacob Goldenberg, Gal Oestreicher-Singer, Shachar Reichman, The Quest for Content : How User-Generated Links Can Facilitate Online Exploration, Journal of Marketing Research, August 2012, 462-468, 17p.
Yang Sok Kim, Ashesh Mahidadia, Paul Compton, Alfred Krzywicki, Wayne Wobcke, Xiongcai Cai, Michael Bain,People-to-People Recommendation Using Multiple Compatible Subgroups, AI 2012 : Advances in Artificial Intelligence, 2012.