Method: Survey, Interview, Usability Testing

Team: Justin Lee, Jay Han

Timeline: Q1 - Q3 2023

Role: UX Researcher & Team Lead

CONTEXT

The increasing popularity of AI has been taking over - even within music.

The music industry has embraced AI to deliver personalized experiences. Spotify, a leader in music streaming, introduced the Spotify DJ AI feature to provide listeners with tailor-made music for specific moments. The challenge is to ensure the AI selects the right song for the right occasion

In the age of digital abundance, choosing the perfect song from a vast library can be overwhelming. The Spotify DJ AI aims to solve this by curating music for users' moods and situations. However, user adoption and satisfaction with this feature remain uncertain.

Reasons for why I embarked on this project

As an experienced and dedicated Spotify user with a passion for music, I started on a research journey to gain valuable insights into user's initial perceptions of this innovative approach to music comsumption and see if there is any room for improvements.

RESEARCH GOALS

To figure out if Spotify DJ meets the listener's needs, I first outlined my main goals 📌

Understand how users perceive the personalization capabilities
Identitfy any usability issues and pain points
Explore the impact of the feature on users' music listening habits
Collect user feedback to inform iterative improvements

APPROACH

UX Research Process

I. EMPATHIZE

Heuristic Evaluation 📃

After I outlined my goals, it was time for me to start researching. I wanted to first try the feature myself before doing my research, so I opened up Spotify and began searching for any potential issues. Using the Jakob Nielsen's 10 usability heuristics, I evaluated the overall usability of the feature implementation.

It took me 20 minutes to figure out where this feature was

I discovered that Spotify DJ breaks these heuristics:

Heuristic #1: Visibility of system status
Heuristic #7: Flexibility and efficiency of use

🚨 THE REAL PROBLEM 🚨

Once I finally figured out where it was, I thought the challenge was over. Yet, my enthusiasm was soon met with disappointment as I began using the feature. I found it quite frustrating to listen to a song that I wasn't familiar of and limited control over what type of music i wanted

Desk Research 💻

Feeling underwhelmed, I went on the internet to read more about what people had to say about it. To have a look on what people have said, you can open this link

Big common issues I figured out through the analysis

DJ is hard to find
Music recommendation are not really catered towards the listener
Find the voice either annoying or does not do well narrating
Seem to repeat the same playlists or stations

SURVEYS 📝

I gather over 30 survey responses on what users thought and felt about Spotify DJ. I aimed to make sure it as not only finding if the DJ is easy to use but also something users needed. Spotify already has created a survey, so I took my advantage of my resources and tweaked some of the questions to align with my research objective.

My 2 Survey Criteria

Active Spotify users
Users who have recently used the DJ feature

Findings 🔎

Key Insights from the Survey

71% of listeners find repetitive music problematic
60% find the DJ button confusing
70% desire of diverse music that aligns with their taste
100% want to know what their friends are listening to
85% seek insights into the history of songs and artists

What this really means

DJ tends to only play familiar songs or playlists
Clarify the DJ button's functionality for users
Want "new" songs but somewhat familiar and similar to their own taste
Enable sharing and discovery of friends' music choices
Users value additional information about the songs and artists in DJ's selection

Because survey only tells the "what", I had to conduct in-depth survey to dig deeper and understand the "why "

PARTICIPANTS 👥

I created research participant requirements. Specifically, I wanted to focus on Spotify users that have recently used the DJ feature for few weeks and see their overall impression.

Disclaimer: The participants I chose are mostly people that I know of which made it easier for me to recruit and gave me an opportunity to ask each user with deep engaging questions.

INTERVIEW ✨

Given that the majority of participants didn't find it DJ very useful, I sought to delve deeper into the "why" by formulating questions to explore the future outlook of music listening, and whether Spotify DJ is even needed for users.

I have selected 5 users from my survey samples and conducted an in-depth interview

Here is some of the participant's quotes 🔎

Goods

"Especially in the car with friends, I like using the DJ because it helps to decide a song for us"

"I like how it is a very novel feature that is exclusive to Spotify"

"Sometimes Dj will play a songs that I haven't heard in a while; especially the throwbacks

Bads

"It doesn't have the granular controls needed to tune it to exactly what I need"

"When playing something from throwback, I wish I can control the year"

"Can't control station, will skip over stations if I skip too many songs. Will automatically change stations after a while even though you are still enjoying it"

"Eventually, songs will become too repetitive and familiar.."

"I don't like that DJ's voice, it would sometimes pronounce the artist/song name wrong"

II. DEFINE

AFFINITY MAPPING 🌟

After I finished writing notes and documenting data from my 5 interviews, I decided to create affinity maps on Miro.

Seeing my interviewees' frustrations drawn out infront truly motivated for this project to skyrocket. Their reaction to my idea was very positive; many STRUGGLED to find the right music. I found a few different patterns: those who prioritized personalized music over song their choice tend care more for the shuffle feature.

I found that people:

😊 People Like

Ideal when users are unsure about what to listen to
Enhances the music experiences when with friends
Appreciation for the novelty of the DJ AI
DJ's selection of nostalgic and familiar songs

😣 People Dislike

DJ's automaic station change after 5 songs
Difficulty in locating the DJ feature
Concerns about the DJ's voice and narration content
Desire for more personalized music
Uncertanity about specific feature functionalities

Follow Up Interview 📞

I conducted a follow up In-depth interview with the same 5 participants to see if their overall preceptions have changed over time.

To my surprise, 4 out of the 5 users exhibited a positive attitude toward Spotify DJ

The novelty of the DJ feature seemed to have played a significant role in the change of perception. Users adapted to the feature over time.

Users found DJ particularly useful when they were unsure about the music they wanted to listen to, especially during driving or when seeking a quick and hassle-free listening session.

III. IDEATE

How Might We? 🤔

Given the changes of user's overall satisfaction of DJ or not, I had to finalize what specific features to settle on. Thus, utilizing HMW was essential for me to focus on the imperative changes that were needed.

HMW improve the DJ feature to better cater to users who want to discover new genres of music based on their mood?
HMW address the issue of repetitve music playlists on Spotify DJ?
HWM design a new Spotify DJ feature that allows users to interact and collaborate with friends?
HWM enhance the DJ's user interface to make it more intuitive and user friendly, and provide better control to users over their music listening experience?

Okay so what now? 💡

Based on the research findings, I have found 2 main categories that were needed to improve the Spotify DJ

1. User Interface 📱

Giving like or dislike option

"When a song doesn't match the listener's vibe, they should have the power to say so. Adding a thumbs-up/thunbs- down option to help DJ fine tune its picks based on the listener's mood"

Quick Mood Adjustments

"Develop a streamlined interface for genre switching, offering users a convenient way to transition between music styles - such as Discovery, Friend's favorites, Today's Hits, Nostalgia, and Recently Played.

Easily discoverable within the app

"No more hide-and-seek with DJ. Placing the DJ on the homescreen, so users can easily spot and quickly access it.

Users can control the DJ's narration

"Sometimes DJ talks a bit too much. There are times where listeners does not want to hear DJ's chit-chat. Have the option for listeners to control when DJ speaks or stays quiet"

Tool Tip for Clarity

"Incorporate informative tooltips to improve user understanding of interface elements. Especially helpful for first time users"

2. Song Recommendation 🎧

No More Suprise Genre Changes

"No more genre whiplash. DJ won't abruptly change the station after 5 songs, so listeners can enjoy a consistent music experience"

Address repetitive playlists

"Create an algorithm that reduces song reptition by taking the songs played recently and favoring a more diverse selection"

IV. PROTOTYPE & V. TESTING

One Big Challenge I faced.. 😓

The primary limitations of our study was that our prototype, created on Figma, couldn't fully replicate the authentic Spotify DJ experience. It lacked the capability to play actual songs, register user interactions such as 'like' or 'dislike', and consequently recommend different songs

Solution

I tackled the challenge of assessing DJ feature usability by making the testing as close to real life as possible. I achieved this through 2 key methods:

Contextual Inquiry: This method allowed me to observe users in their eveyday environments, providing insights into how they naturally interacted with DJ
Scenario-Based Testing: I presented users with scenarios that mimicked real-life situations, like road trips or workouts, making the testing feel like their daily experiences.

Overall Finding from the usability testing

Through this research journey, several critical insights emerged that hold the potential to shape the future of the Spotify DJ feature:

I found out that:

People find the thumbs up/thumbs down feature extremely useful
Making sure that the users can have better control over the DJ's narration
Songs should not change after 5 songs, instead allow the users to decide that for themselves.
Make the DJ more easily accessible and user-friendly for first time users by including tool tips

View Prototype

Click here to learn more about the usability testing

REFLECTION

Valuable Lessons I have learned

Effective Communication of Findings: Power of presenting findings with engaging insights to make data accessible and promote team discussion and action
Value of a Research Guide: Having a well-defined research guide maintains focus and aligns research with goals
Action Over Speculation: Prioritizing empirical evidence over assumptions for informed decisions
User-Centered Focus: Valuing participant feedback and behavior over personal bias

Page updated

Google Sites

Report abuse