Search
  • Rob Haisfield

What social media news feeds like Facebook and LinkedIn could learn from Spotify


Photo by George Pagan III on Unsplash

Music discovery services like Spotify have long used AI to determine what you like to listen to and to give you more. How does Spotify get better and better at giving you what you want to hear, and how is that different from social media news feeds?


To start, it would be helpful to mark the distinction between revealed preferences and declared preferences.


If you're looking at someone's revealed preferences, you're inferring what they like from the choices they make. The problem with that is people would often say that they would prefer to prefer doing something else, but because of habits and failures of self-control, they keep making the choice that they aren't happy with. For example, I'd prefer to prefer reading a book over watching Netflix but I watch Netflix anyway.


On the other hand, if you're looking at declared preferences, you're trusting what people say they like and don't like. The problem with that is that this data isn't always the most useful because it often isn't in alignment with what people actually do.


The heart tells Spotify you want more, and the minus says you want less

Social media news feeds mostly rely on revealed preferences. They keep track of what you like, comment on, and generally pay attention to, and they give you more of that to keep you on for longer. Spotify explicitly asks the users through their design to tell the algorithm what they want to see more or less of with their heart and minus symbols. I'll call this UX technique the declared preferences question. This means that they can still use revealed preference metrics, but they give the users an obvious way to declare their preferences. As a result, the user has more of a sense of ownership over what music is given to them and Spotify is better able to provide their users with music they'll be happy with.


You can still declare your preferences on social media if you put the pieces together yourself and you're savvy with how the algorithms work. You can treat your timeline as holy and like what you want to see and hide or unfollow everything you don't want to see.


I do this on LinkedIn and basically only see posts from people I trust with valuable resources now. I don't mind spending a lot of time on social media if everything I gain from it is useful.



No matter how hard I try, I can't escape Gary Vaynerchuck

The problem is that the declared preferences question isn't explicitly given to the users- it's implied and underemphasized. To hide a post, you usually need to do some variation of clicking three dots in the top right corner and then navigate the menu to hide or unfollow. Additionally, liking is perceived to be more about you giving recognition to the content creator than you communicating to the algorithm to what you want to see.


Facebook, LinkedIn, and Twitter could give their users a declared preferences question by putting a check mark and an x in the top right corner to distinguish it from socially interactive features like liking or commenting. This could give social media users a way to anonymously endorse or reject a post and posts like it for their feed.


By posing a declared preferences question to users, products end up with better trained algorithms and happier customers by opening up a direct line of communication between the two.


Are there any other products you've seen that pose a declared preferences question? Are their other ways that products can ask the user what they want?

83 views

©2019 by Influence Insights.