Role
UX Designer
Tools
Figma, FigJam, Zoom, Maze
Timeline
January 2024 - March 2024
Problem
Why Users Still Struggle to Find Something to Watch on Netflix
Netflix users often struggle to find something new to watch and don’t trust the platform’s recommendation engine to match their current mood. Instead, they prefer to rely on personal backlogs of recommendations from friends, family, and social media. If they can’t find appealing content, they may turn to competitors, risking long-term subscriber loss for Netflix.
Solution
Making Content Discovery Social to Reduce Decision Fatigue
The new social feature adds a friend activity feed to the Netflix desktop app, allowing users to see what their friends are watching in real time. Users can send and receive show or movie recommendations directly within the platform, making it easier to discover content from trusted sources. This creates a more personalized, socially driven viewing experience that builds on users' existing habits of relying on peer suggestions.
Competitive Analysis
Uncovering the Opportunity to Leverage Social Connection
Research was first performed on Netflix’s deprecated feature, "Play Me Something", to understand why this feature, which was implemented to randomly play content for a user based on the Netflix algorithm, did not work for the product. It was relevant to understand why the feature failed because its purpose was helping a user choose content, something I hypothesized users struggled with when watching in groups. A competitive analysis of platforms Hulu, Max, Prime Video, and Spotify revealed only Spotify to offer a native social feature—highlighting a clear opportunity across the streaming landscape to improve content discovery through trusted, social connections.
User Interviews
Users Struggled to Decide on Content Individually Rather Than in Group-Viewing
5 participants who use or have used Netflix and other streaming services were interviewed over Zoom. Their interviews were then synthesized into 3 key insights.
Group Viewing Choices
In group settings, users often choose familiar, crowd-pleasing shows like "The Office" or something pre-agreed upon, as the focus is on socializing rather than the content itself.
Mood-Based Decisions
When users don't want to spend energy deciding, they turn to quick, familiar shows for background noise, especially when nothing on Netflix fits their current mood.
Trust in Word of Mouth
Recommendations from friends are preferred over Netflix’s suggestions, as users trust their friends' taste and enjoy discussing mutually watched shows.
User Persona + Customer Journey Map
Meet Rachel: The Overwhelmed Relaxation Seeker
The key insights from research were then used to develop a user persona and a customer journey map, which were used to illustrate a user's thoughts and emotions, highlight pain points, and opportunities for improvement along the Netflix customer journey.
How Might We
Two HMW Statements to Tackle Two Separate Pain Points
User research revealed two key insights: first, users trust word-of-mouth recommendations more than Netflix’s algorithm; second, they often struggle when suggested content doesn’t align with their current mood. To address both challenges, two separate “How Might We” statements were developed—allowing room for ideation and flexibility to prioritize or explore solutions for each as time allowed.
Ideation
Choosing my HMW After Brainstorming Ideas
With my ‘How Might We?’ statements top of mind, I held a brainstorming session using FigJam. After reviewing the ideas generated, I decided to build on the feature of recommending content with friends within the Netflix app to solve for the first HMW statement due to complexity of the survey idea and project time constraints.
Feature Prioritization
Prioritizing Features for MVP Using the MOSCOW Method
Next, I prioritized feature development based on user insights, ultimately focusing on enhancing content recommendation and social interaction within the app. Additionally, I drafted a user flow for recommending content to a friend to help map out necessary key screens.
Must Have
Search for friends
Add, remove, or block friends
Friend activity feed
Recommend content to friends
Privacy setting
Share recommendations with non-Netflix users
Should Have
Collaborate on watchlists
Could Have
Add profile information
See friend's series or movie progress
Friend content ratings
Will Not Have
Match My Mood
User Flow: Recommend Content to a Friend
Sketches
Putting Pencil to Paper for Must Have Features
Once I decided on the features I was going to focus on, I started sketching to visualize the screens I would need to create. I solicited feedback through group critiques, thinking through interaction flows and interface designs in this phase before spending time digitizing these screens.
Activity Feed/Add Friends - Version 1
Activity Feed/Add Friends - Version 2
Recommend Content to Friends - Version 1
Recommend Content to Friends - Version 2
Testing and Improvements
3 Main Improvements From Usability Testing
Users were asked to complete the following tasks during a moderated usability testing session over Zoom.
Add a friend
Recommend a show to a friend
Accept recommendation
Streamline the navigation bar and establish an intuitive task flow for accepting a recommendation
3 users expected recommendations and friend requests to appear in Netflix’s existing notification bell dropdown.
Before
After
1
Removed inbox icon and nested all notifications under the existing bell icon
2
Added and reorganized the notification dropdown to include "Friend Requests & Recommendations"
Separate Section For Friend Recommended Shows
Users expressed that once the recommendation request was accepted, they would want a separate section where they could specifically see the shows that a friend recommended in addition to having it added to the "My List" section.
Before
After
1
Added a separate watchlist specifically for content recommended by friends
Updated the Friend Activity feed with Timestamp Information
Adding timestamp information to the friend cards should enhance the social feature and make the activity feed more relevant.
Before
After
1
Added last activity timestamp to friend cards and ordered them by the most current person online
Final Screens
Minor Changes Made After Testing the High-Fidelity Prototype
5 participants (4 moderated, 1 unmoderated) were asked to complete a second round of usability testing on a high-fidelity prototype via Maze. From the insights gathered, most changes were not made since the majority of users took the expected path to complete tasks. However, the final iteration was modified to include a visual indicator to the Friend Activity feed to make it apparent friends are online as users mentioned this would be interesting to know.
Adding a Friend From the Search Bar
1 user attempted to use the Netflix search bar to add a friend, while all others clicked on the friends icon in the navigation bar.
Sharing a Show Using the Friend Activity Panel
2 of 5 users initially chose the Friend Activity panel to locate a friend’s page, whereas the other 3 went the expected path by first accessing the show.
Accepting a Recommendation is Intuitive
All users followed the expected path with no variation.
Friend Activity Feed + Recommend Content to Friends
Add Friends + View Friend Requests and Recommendations
Learnings
Next Steps
Add a way for users to select a friend from the Friend Activity panel and search for a show from there to recommend to the friend. A minority of testers (2 of 5) intuitively tried to go through this user flow.
Key Insights
You can’t always solve for every user problem in one development sprint. As much as I would have loved to solved both ‘How Might We'?’ problems I stated, I had to evaluate my constraints and prioritize what features were most useful and realistic for MVP.
Spending the time in lower fidelity iterations pays off. I didn’t have to worry about the images for my mid-fidelity screens and was able to focus on fine tuning the interactions of the feature. It laid the groundwork for me needing to spend less time perfecting my high-fidelity screens and final prototype later on.