top of page
Frame 1_edited.jpg
iHR_secondary_White.png

AI POWERED PERSONALIZATION FEATURES

Designing the first AI Powered features for iheartradio.

iHeartRadio is a free, ad-supported streaming service that offers radio, podcasts, and music. It's owned by iHeartMedia, the largest radio broadcaster in the United States, I consulted for iHeartRadio in the capacity of a Product designer
to design their first AI powered expereinces.

  • Designed iHeartRadio’s first AI powered personalization features impacting 4 million paid subscribers.​​

​

  • Developed the UX strategy for audio recommender systems in collaboration with the machine learning team.

​

Project Duration - Jan - June 2024 (6 months)

​

Group 1073716982.png
Group 1073716984.png

THE TEAM

Collaborated with several teams including Machine learning and analytics team

I spearheaded the collaboration across several teams ranging from machine learning to UX research to define the business problem and user problem and deliver user centered design solutions.

My Role : Lead product Designer


 

Machine learning and analytics

User Research &
Insights 

Product
Management

Comms and Makerting

Technical Operations

iHeartRadio operates in a crowded and rapidly evolving audio streaming market, competing against platforms like Spotify, Apple Music, and Pandora.
 
Retaining users is critical to sustaining revenue growth, particularly in subscription services and ad-supported models.
 
The Opportunity with AI :

AI offers a strategic pathway to address these challenges by enhancing user engagement, streamlining content discovery, and delivering personalized experiences.

BUSINESS PROBLEM.

How to Leverage AI to Enhance User Retention

CUSTOMER PROBLEM

Audio Recommendations don't meet user expectations

Many iHeartRadio users are dissatisfied with personalized audio recommendations that often repeat content or miss the mark, leaving their preferences unmet.

Additionally, the overwhelming variety of available audio content makes it challenging to decide what to listen to, with users unwilling to spend excessive time searching for something enjoyable

THE SOLUTION

AI based audio content recommendation features and a robust recommender system

I led the research and design of three AI-powered content recommendation features and developed the overarching UX strategy for the audio content recommendation system.

Fresh Finds :

​

Transforming Homepage Discovery with AI-Powered Personalization

​

​Fresh Finds delivers a curated list of radio, podcasts, and music, using AI to personalize recommendations based on user behavior and preferences. To discover more content, users can refresh Fresh Finds

​

Personalized Recommendations:

 

AI curates content based on user preferences, location, time, and day, continuously learning from interactions to improve future suggestions.

​

​​

​

Harmonize :

 

Create AI powered playlists with friends and family​

​

​Harmonize allows users and their friends to co-create AI-powered playlists, perfectly tuned to everyone’s mood. Users can enjoy music discovery through social networks and let AI craft shared listening experiences.

​

Dynamic Collaborative experience:


The playlist updates regularly, ensuring it reflects the users' evolving preferences, making it a fun way to discover new music while strengthening social connections.

iHeartAI 

 

Conversational AI to help users find audio content​

 

iHeart AI is a conversational AI assistant that helps users quickly discover audio content based on context or mood using natural language processing, sentiment analysis, and machine learning for personalized recommendations.

​

Personalized Search Results:


iHeart AI delivers personalized search results by analyzing user behavior, preferences, and popular audio trends.

​

​

Personalization Strategy :

 

Foundational UX Strategy for recommender systems powering personalization features such  as fresh finds

​

Implicit and Explicit Signals on users preferences based on GDPR guidelines :

 

Curated and designed a comprehensive suite of features to gather user preferences, including thumbs up/down interactions, onboarding quizzes, playback behaviors (skips and plays), and engagement metrics like followed artists, in accordance with GDPR guidelines driving enhanced personalization

​

AI models learn over time :

​

Developed a UX strategy that enables the recommender model to learn and adapt over time, using diverse techniques to gather insights and deliver increasingly personalized audio recommendation

Final UX Strategy Poster.png

USER RESEARCH

Collaborated with the Research team to uncover user insights
 

I collaborated with the user research and insights team to gather more data on user's audio listening habits and needs

Surveys :

Conducted a survey with over 200 participants to explore how users interact with audio platforms and identify the factors driving their choices.

Interviews :

Facilitated in-depth interviews with 14 participants to uncover user preferences and pain points in discovering audio content they enjoy.

Competitor analysis :

Analyzed the features of key competitors, including Sirius XM, Pandora, Amazon Music, Apple Music, TuneIn, and Audacy, to understand how they leverage AI to enhance user retention.

 

Key  Findings

Finding 1: Personalized audio recommendations don't meet user needs​

​​

​Many people are dissatisfied with personalized audio recommendations due to repeated content and inaccurate suggestions.​​

​

​

Finding 2: People often rely on their immediate social circle for audio recommendations

​​​

People rely on recommendations from close friends and family on audio content they like.

​

​

Finding 3: Moods and activities influence user's listening preferences

​​​​

Users like to specific audio content based on the time of the day, mood and activity and often 

​

​

​​

BRAINSTORMING AND IDEATIONS

Organized cross functional design workshops to ideate solutions for each finding

Global buy in was critical for this project. In order to achieve an agreed upon path I included members of the research team, engineering, product and design in large scale collaborative workshops focused on reviewing our personalization strategy and asking the question;


Addressing Finding one: Personalized audio recommendations don't meet needs
​

Collaborated with multiple departments to build the UX strategy for a recommendation system that provides new and relevant audio suggestions.

Our UX strategy focuses on delivering a personalized and engaging content experience by leveraging both explicit and implicit user feedback.

 

The approach evolves through four key phases to cater to users at different stages of their journey:

​

  • Cold Start Experience: For new users, we gather explicit feedback through genre preferences during onboarding and implicit feedback from initial interactions.                        

  • Initial Experience: For users with a few weeks of activity, we use enriched data from likes, search keywords, and playlists to offer tailored recommendations.                                 

  • Intermediary Phase: For active users over several months, we analyze interaction patterns, like the time of day they listen to specific content to provide Contextual suggestions,                                                             

  • Final Phase: For power users, we refine recommendations with more intrusive feedback methods, including surveys and emails. Cross-domain recommendations, such as introducing podcasts to music listeners, broaden content discovery and deepen user engagement.

Planning board

Screenshot 2025-01-20 212025.png

Read about the full strategy here

Finding two: People often rely on their immediate social circle for recommendations​

Designing social features for Iheartradio app

We brainstormed several ideas and concepts for social features for the iheart radio app to facilitate audio content discovery through social network

Idea board

Frame 10.png

Finding three: Mood and activities influence user's listening preferences

Enable users to input mood- or activity-based prompts to receive AI-powered personalized recommendations

We ideated on diverse approaches, including digital assistants, chatbots, and conversational AI, to empower users to express their moods and receive tailored recommendations

Idea board

Frame 9.png

Conducted concept testing by gathering feedback directly from users and engaging with cross-functional teams to identify and address technical and business constraints, ensuring the development of well-informed and viable final design solutions.

CONCEPT TESTING

Collaborative Concept Testing and Refinement to finalize solutions

AI POWERED SOLUTIONS

Group 488.png

“Why is this Recommended?”
Users are provided with clear and concise explanations of why particular songs, podcasts, or radio stations have been suggested to them.

iHeart AI
Users can use iHeart AI to find specific content using prompt-based searching.

Content Recommendations
Users can select a curated list of radio, podcast, and music content based on user habits, preferences, behaviors, location, time, and day.

Refresh Recommendations
Users can refresh content recommendations to get their desired recommended content.

Fresh Finds: Transforming Homepage Discovery with AI-Powered Personalization​​​​

​​Fresh Finds delivers a curated list of radio, podcasts, and music, using AI to personalize recommendations based on user behavior and preferences. To discover more content, users can refresh Fresh Finds​

iHeartAI: Conversational AI to help users find audio content​

iHeart AI is a conversational AI assistant that helps users quickly discover audio content based on context or mood using natural language processing, sentiment analysis, and machine learning for personalized recommendations.

Frame 1073718084.png

Content Type Specifications
As an option, users can select what type of content they are looking for specifically in order for iHeart AI to provide tailored, accurate responses.

Trending Suggestions
Users can view trending suggestions to get prompt ideas to search for on iHeart AI.

Conversational Search
Users can interact with iHeart AI by entering prompts or conversational text to discover their desired audio content.

Tailored Search Results
iHeart AI provides specific, tailored results based on users’ behaviors, preferences, and trending audio content.

Surprise Me
If users don’t know what to search for, surprise me offers them random, novel, or unexpected content.

User Feedback & Refresh
Users can provide feedback on iHeart AI’s responses and refresh content if they aren’t satisfied with existing responses.

Harmonize: Create AI powered playlists with friends and family​

Harmonize allows users and their friends to co-create AI-powered playlists, perfectly tuned to everyone’s mood. Users can enjoy music discovery through social networks and let AI craft shared listening experiences.

lala.png

Intro screen for Harmonize

Add Friends

Add content type

Enter mood based prompts

Combined playlist of the user and friend

FUTURE WORK

Integration with Live Events

iHeartRadio is known for its various live events, making it a natural progression to integrate these events with the platform’s music discovery capabilities. This integration can create a more holistic and engaging experience for users by connecting their music preferences with real-world music events

bottom of page