Interaction Design / Information Architecture / Research


Identify the current information architecture of Netflix and compare it to similar streaming video sites. Our team proposed a redesign in which users are better able to find the content they desire through suggestions.


Modify the sorting algorithm of Netflix to be more user-centered by enabling users to create collections of content. Based on these collections, Netflix and other users can then suggest movies and TV shows.

Netflix is a popular online video streaming service available through subscription on nearly any mobile or desktop device, including but not limited to Smart TVs, game consoles, cell phones, and tablets.

Despite the ubiquitous presence of Netflix, the current navigation structure is often viewed as cumbersome to users, who complain about the organization of its categories, the accuracy of the “suggested for you” content, and the difficulty of finding the film or television episode title that they desire.

1 In order to better understand how real people use and navigate Netflix, we performed exploratory interviews Netflix users. We found that the users we spoke with fell into the following categories:

  • Passive viewer — This user prefers browsing Netflix's recommendations rather than search for a movie or TV show.
  • Casual viewer — This user enjoys browsing Netflix's recommendations and searches for a particular movie or TV show occasional. The casual viewer is a mix of the intentional and passive viewer.
  • Intentional viewer — This user knows exactly what he is looking for on Netflix. They use the search bar all the time to search for a particular movie or TV show.
  • We asked questions like how users approach key tasks like finding an interesting title to watch as well as behavioral questions like how often and how long they use Netflix. These questions aim to reveal insights on our user's goals, needs, and use patterns and hence inform our design decisions.

    2 We took a content inventory in order to illuminate what elements currently exist on the site. This became our basis for our selected methodology and further analysis of the site.

  • Who's watching
  • Browse
  • Kids
  • DVD
  • Search
  • Notifications
  • My List
  • User Profile
  • Furthermore, we discerned the categories Netflix uses to suggest movies and TV shows to users.

  • Recently Added
  • Trending Now
  • Continue Watching
  • Top Picks (for user)
  • Because You Watched...
  • Watch Again
  • Popular
  • In addition to the standard categories, other categories are presented dynamically and vary from user to user in response to personal rathings or what the user has watched before.

    We performed a LATCH analysis to understand the different methods in which Netflix displays its content. This activity explored various possibilities for how movies and TV episodes are categorized.

  • Genre (categorical)
  • Language (categorical)
  • Year released (continuum)
  • Recently watched (time)
  • Ratings (hierarchical/continuum)
  • Alphabetical (alphabetical)
  • Geography (location)
  • 3 Our approach was to see how users would categorize titles in an open card sort. We were interested in observing the similarities and differences between the results of our open card and how Netflix currently groups and organizes television and movie titles.

    We selected 30 titles (both movies and TV shows) and created an open card sort using the program Optimal Sort. Interestingly, after a very limited number of participants were tested, the results started to show the common categories based on genre of comedy, drama, and crime. The initial card sort had one major flaw — not everyone recognized the titles that were picked. We thought it was important to consider if this would significantly skew the results.

    We created a second card sort with 44 titles in which we selected titles that we anticipated would be more widely recognized by the participants. We also changed the settings of the card sort program so that if a title was not recognized, the participant didn't have to categorize it to complete the card sort.

    4 We identified the core tasks because they are essential to understanding what users are trying to achieve and help construct models for how they navigate and interact with the site.

  • Watching a title
  • Resume watching
  • Browsing titles
  • Searching for a specific title
  • Adding titles to lists (proposed)
  • View other users' lists (proposed)
  • Create a new list (proposed)
  • Select a movie
  • Rate titles
  • Add a tag (proposed)
  • Write a review (proposed)
  • View recommendations
  • 5 We conducted a competitive review of other prominent websites that provide online video streaming services to identify industry standards and understand information architecture that users may already be familiar with and potential best practices. We analyzed Amazon Video, Vimeo, and YouTube.

    This enabled our team to recognize the format in which users were already familiar with navigating other streaming video sites, which would allow the users to easily learn a new format of Netflix.

    6 We then turned to the existing Netflix information architecture. From the current Netflix format, we charted a site map, then proposed a new site map which considered the needs and wants of users that were communicated during our interview process.

    7 Comparing the existing site map to our proposed site map allowed us to design wireframes in a way that enhanced new features while maximizing learnability. Many of the existing features of Netflix remained, with new features in close proximity to make them discoverable while maintaining the original look and feel of Netflix. Therefore this new system would be easy to learn for a wide range of users depending on their preferences.

    View full prototype