Tailoring YouTube Recommendations
UX Case Study
Overview
This case study explores the potential of leveraging YouTube's existing recommendation algorithms to create a more engaging user experience by suggesting videos that align with viewers' diverse interests, as reflected in their curated playlists. The goal is to increase watch time and foster a more engaging experience, benefiting both YouTube and its viewers.
TIMELINE
Nov - Dec 2023
DESIGN ROLE
UX Research
UI Design
Prototyping
Tools
Figma
Miro
How Might We...
… empower users with greater control over their YouTube content recommendations to create a more personalized and engaging experience.
Despite YouTube's vast features and content, users encounter challenges in discovering videos tailored to their evolving interests and preferences, leading to suboptimal user experiences and decreased engagement.
Problem
Research
Understanding user's relation with YouTube Recommendation
Methodology - Interviews + Affinity Map
I tested the sub-hypothesis by preparing a script of open-ended and unbiased questions and conducted 3 interviews.
I grouped user interview responses and feedback into categories to identify recurring pain points and preferences related to recommendations.
I often discover new channels and content through YouTube's recommendation algorithm.
The 'Up Next' feature on YouTube introduces me to some amazing videos that I wouldn't have found otherwise.
I appreciate how YouTube suggests videos based on my viewing history, making it easier for me to explore new topics and interests.
I'm frustrated with YouTube recommending clickbaity content instead of informative or entertaining videos.
I've noticed that YouTube tends to prioritize popular or trending videos, neglecting niche or lesser-known content that I enjoy.
Sometimes YouTube's recommendations feel disconnected from my viewing habits, suggesting uninteresting videos after a single watch.
I wish for more options to customize my YouTube recommendations, like selecting specific topics or genres.
While I enjoy personalized recommendations, I wish there was an option to provide feedback for further refining my preferences.
Find new content
Refine recommendations
Prioritize popular or trending videos
"The recommendations can be hit or miss. Sometimes they're relevant, sometimes not so much."
Key Research Insights
Users express a desire to find new content. Playlists offer a way to curate and organize content based on specific interests.
Users feel frustrated with recommendations that do not align with their interests. Playlists can provide a clearer signal to the algorithm about user preferences.
Users wish for an option to refine their preferences and provide feedback. Playlists can act as a feedback mechanism, indicating the type of content users want to see more of.
Playlists demonstrate user interest in specific content areas, which aligns with the idea that playlist-based recommendations can potentially lead to more personalized content and ultimately a more engaging viewing experience.
Assumption Mapping
I was mindful of the assumptions I was making and wanted to select the most critical one to be tested. For this, I mapped all assumptions based on risk and knowability.
Users prioritize watching content aligned with their interests (reflected in their curated playlists)
High-Risk
High Knowability
Low-Risk
Low Knowability
A significant portion of users create playlists for specific topics or interests.
Users expect more playlist recommendations to be relevant to the playlist's theme.
By providing more relevant recommendations, users will watch more videos and spend more time on YouTube.
Users are comfortable using basic recommendation features like "liked videos" and "watch history" to personalize their experience.
The visual design of the recommendation interface can influence user perception of its effectiveness.
Hypothesis Statement
Using the most critical assumption, Icreated a hypothesis statement that I could test through the prototype.
I believe that
Users who create playlists on YouTube prioritize watching more videos aligned with the themes and topics of those playlists.
Assumption
(A statement I believe to be true)
I will know this is true when I see
10-15% increase in average watch time per user session
User Insight or Market Feedback that will validate the belief
Sub-hypothesis
I felt that the above hypothesis could be broken down to be more specific for testing.
I believe that
recommending new videos specifically tailored to the themes and topics of user-created playlists
Building X
(proposed solution)
I will know this is true when I see
10-15% increase in average watch time per user session
User Insight or Market Feedback that will validate the belief
for
users who actively create playlists for their diverse interests
User
(who will benefit from the proposed solution)
will achieve
more engaging personalized experience
Outcome
(benefit to the user)
Persona
Alex Parker
Demographic & Psychographic Traits
Age
38
Gender
Male
Occupation
Software Engineer
Tech Savviness
High: Proficient in various software tools and programming languages.
Personality
Analytical
Driven
Detail-oriented
Alex has a passion for lifelong learning. Even after a full day's work, he dedicates time in the evenings to expand his knowledge base. He's an avid consumer of educational content, from tech tutorials to TED Talks, always seeking ways to grow personally and professionally.
Quotes
"Researching online feels like a rabbit chase! Wish things were more organized."
"Give me some control over what I am watching! Surprise is great, but relevance is even better."
Goals
Alex's primary goal is to continuously learn and stay updated in his field.
He values platforms that offer high-quality educational content in an easily digestible format.
Frustrations
Scattered Recommendations: Alex craves a structured learning path, but YouTube's current recommendations jump between unrelated topics, disrupting his focus.
Inefficient Learning Flow: As an analytical learner, Alex values efficiency. YouTube's current system might require him to jump between searches and playlists, hindering his ability to efficiently build his knowledge base.