Spotify Casestudy
Project Type: Independent Project, Desktop UI/UX
Brief: Spotify is one of my most used applications that is a part of my daily life cycle. I have been heavily influenced by the journeys of a few Indian indie artists (namely, Nucleya), this has intrigued me to study the functioning of Spotify and understand how users interact with the application.
The project is about solving user problems with appropriate UX principles and an intuitive interface.
Process
I started out by studying Spotify’s revenue model, technology models. Spotify was late to the Indian market, so it was important to study its competitors and their marketing plans to strive ahead.
Alongside, I conducted 2 surveys, an Instagram poll & post-survey questionnaires. To understand the user’s frustrations, desires, and goals, I carried out the card sorting method and used affinity maps.
Survey Insights
20+ participants
#1
The reason why users switch to other streaming services is due to the frequency of ads. (Despite India having the lowest cost per 1000 impressions.)
#2
Features important to the users are curated playlists, friend’s activities, and suggestions.
#3
Users of Spotify invest a lot of time and effort to curate their own playlists. Unlike other services where users feel less interaction.
Affinity Mapping
I started to map the whole data collected across the surveys in Figma and grouped certain behaviors, attributes together which helped me to analyze efficiently.
Problems
Users use Spotify in different ways but a few tasks stay constant like their continuous search for better playlists, checking recommendations, and seeing what others are listening to.
#1
How can I make users understand recommendations better?
#2
How to solve the abundance of playlists in the user’s library?
#3
An efficient way of sharing song recommendations?
Card Sorting
I found a few inconsistencies in the contents within the menu of Spotify across all its applications. It was important to understand where users expect certain features to be.
Number of participants: 8
Optimizing Radio
Many respondents were either not happy with the song recommendations on Spotify’s radio or feel other applications understand their taste better.
Machine learning will continuously try to improve which means it might not be perfect at the start which may negatively impact user satisfaction.
How do we solve it?
-
It is important for users to know why a certain suggestion has popped up in the playlist. Categorizing a playlist helps users to model what kind of songs might be present in it.
-
There’s no option for the user to give feedback while listening to the radio. Adding the ‘like & dislike’ metric could improve the overall user experience and train AI better. In the current app, the suggestions act as a black box with no option to interact.
-
When a user provides his feedback about the recommendations he does expect the result to be reflected in the playlist. Adding a refresh button or auto-update features could motivate the user to interact more.
Sorting Playlists
From the survey, on avg users saved over 40+ playlists (with a max 78) to their library but only listen to 6-7 playlists. I have over 37 playlists in my library and use a maximum of 3 playlists.
Spotify just lists them vertically in a menu, imagine how a person having over 70+ playlists would look like. Scrolling through these excess choices would make the user feel fatigued making the decision moreover user might avoid making the choice.
How do we solve it?
-
Breaking information into smaller chunks helps users to grasp it better. Similarly, grouping playlists into realizable categories will reduce the cognitive load on users. Also, in accordance with Fitt’s law, rearranging vertical columns to horizontal groups will optimize discoverability.
-
Adding the recently played playlists to the feed would make the users prioritize their choices and help to improve the discoverability as well.
Sharing Recommendati-ons
From the survey, users find it troublesome to share recommendations with their friends. Though a link can be shared across various messaging services there’s no repository where they can collect and revisit it.
How do we solve it?
The solution would be, to add a feature where users can suggest songs to their friends and the suggested songs would add to a mix.
The only problem would be how can we restrict who shares the songs with the user.
Logic Map: To explain this we’ll have a ‘test user’ who wants a song suggestion from his friend Monarch. The user has to follow Monarch’s profile which enables Monarch to recommend a song, that song would go into Monarch’s mix. Additionally, the user can manually add a song from Monarch listening activity to the mix. Following a profile would be mean willingness to accept recommendations.
Implementation: The UI of the interface includes a section for suggested mixes by the user’s friends. These mixes include all the songs that the user’s friends suggested & songs added by the user from friend’s activity.
