Overview
In late 2008, Netflix DVD by mail subscription service was on the rise, its core focus was trying to get people to constantly find new and interesting movies and shows to watch. While streaming had already begun, the streaming catalog was still far from being as compelling out complete as the DVD one. As such Netflix had to find a way to make sure people could find things to watch and there make their subscription worth its price.
Goals & Challenges
While each user has distinct tastes, there is also a set of top level content that normally has such broad appeal that its easy to recommend. However, the more nuanced titles that require that the service know more about its user, was something Netflix needed to solve for. How to get the best initial signal possible, so that the service can on day one begin to provide value while still being accurate on its recommendations.
Building a Taste Preference Profile
The taste preferences that a user has are not by any means simply a like or a dislike. While we often tend to state that we are Horror or Sci-Fi fans, there is nuance to those statements, in that in some cases we may enjoy Sci-Fi but really really enjoy Sci-Fi and know this is can totally change the way to experience a service thats trying to provide you with the best content possible.
signing up for the service
As part of the sign up flow, we considered adding a taste preference flow that would aid in the creation of a profile. This flow is something that the user would have to grow through before beginning their experience in the service.
The users ability to understand the different types of genres that the service has is key to establishing the profile.
The user then has to make a selection for each genre of how often they see content related to that specific genre.
The end results should be a service that has targeted its recommendations to your preferences and this in the end should drive further engagement. The recommendations will also serve as a mechanism to get a stronger signal base on the selections that had been made. Ideally with this kind of signal, the service will only get better and better. The service should also utilize the choices that have weak signals as a means to serve you content that shares some aspects of genres you like, but probably land in genres you may not have signaled as top priority preferences.
Measuring initial Taste Preferences.
In the early days of the DVD by mail business Netflix had established a pattern by which they could collect information on a users movie preferences. One of the early learnings is that proving too much options about how you felt towards a genre of movies, would lead users to not complete the sign-up process.
Instead they narrowed it down to basically a clear sense of signal for liking or disliking a genre. As well as still letting users be neutral. This simple three prong approach was the basis for crafting how Netflix would collect info to then cater the service to those specific choices.
This simplified approach would go on to guarantee that not only would most people sit through the sign-up flow, but they would in fact fill in the taste preferences honestly. The redesign of the overall experience was all about streamlining how this new 3 choice option could be best presented to users.
Qualitative Testing of Different flows.
Content Led - Stacked
In this iteration, of the taste preference flow, we planned to show actual content at the top of the view, this would hopefully drive people to engage with the flow when they see titles that they might watch. The choices are stacked which means some may only be seen above the fold and the rest require scrolling down the page.
Concept Art - Stacked
In this iteration, of the taste preference flow, we planned to a more conceptual visual that aids in the understanding that this is all about your taste preferences. The choices are stacked which means some may only be seen above the fold and the rest require scrolling down the page.
Concept led - side by side
Like in the first version, we lead with content as the main visual. However this time around you are seeing your taste profile choices all above the fold for the most part, hoping that this will guarantee that the user will engage with both modules and have a more complete profile.