Opportunity
Music streaming services like Spotify have provided efficient and convenient access to music listening anywhere you are. Users can create unique and shared listening experiences that add to their lifestyles. With the excursive music recommendations from Spotifys algorithm, and the increase of users introducing new content; discoverability of unique music has become an exhaustive and tedious process. How can Spotify provide improved personalization in music recommendations to retain its users?
For this speculative project my objective is to provide a simple and easily discoverable feature in Spotify that offers users more control over their music listening preferences.
For this speculative project my objective is to provide a simple and easily discoverable feature in Spotify that offers users more control over their music listening preferences.
Research
Let's take a look into Spotify users’ music listening habits.
First, we'll study the usability of the navigation scheme so we understand
what a more personalized music listening experience can look like.
What does music have to do with mood?
It has been well studied and documented throughout history how music has numerous psychological benefits.
The initial research begins with understanding how user’s music listening habits can be used to help track and manage their moods over time. With these findings, Spotify can give more accurate personalized music recommendations.
Where does Spotify fit into the intersection of music and wellness?
Since Spotify users have different music listening goals and motivations, I will focus on understanding the correlation between music taste preference, activity, and mood change patterns over several weeks.
Target audience research
Understanding the user's needs, motivations and goals.
Business environment research
Current feature offerings in other platforms in Spotify, SWOT analysis.
Contextual interviews with Spotify users
Pre-selected based on screener survey criteria to further expand insights about the users and their behaviors.
Target audience research
Before meeting with our users, it’s relevant to know the background, what our competitors offer for personalizing music recommendations, along with navigation, UI, & integrations.
Business environment research
Understanding the strengths and weaknesses of Spotify’s
top competitors gives us clear direction for the business, user,
and technical needs of our users to have a continuous positive experience.
top competitors gives us clear direction for the business, user,
and technical needs of our users to have a continuous positive experience.
Contextual interviews with Spotify users
To understand who is having these pain points, we first need to understand who is using Spotify and how, when they are using Spotify. Some background statistics give us insight into demographics, usage, and feature preferences:
Gathering Spotify user feedback and visualizing survey data
Using this dataset, we determined a user base of 30 participants on select criteria for an initial survey that can be used to determine the right candidates for future interviews. Users who answered questions with keywords including mood, activity, and type of influence will be invited for a contextual interview.
What do Spotify users want us to hear?
Empathizing with the users through understanding their reasoning, motivations, and mental models will help us to
- 1) measure the severity of our problem
- 2) define if the feature we’re developing is useful, functional and readily adoptable into Spotify‘s ecosystem.
Strategy
With this knowledge, we can understand how to create a feature that offers users control over music search to generate recommendations better aligned with their mood and music taste.
Based on the user mindsets outline, I defined segmentations based on how: self-sufficient, or reliant on others; and how painful or reactionary users are as related to their music-listening behavior.
How might we address the limitations of Spotify’s music recommendations?
To ensure I was surveying as many possible opportunity areas, I reframed the core needs from user interviews into How Might We statements.
Insight
We know: Users face indecision by abundance of choice. Spotify premium users need to have a more robust search filter because they are overwhelmed with the discoverability features and are unable to narrow their search.
We know: Users generally use the search bar to filter search results for specific music that appeals to their current mood.
How Might We...?
How might we improve music recommendations with consideration to mood and listening context so that users are choosing to listen longer and more consistently on Spotify?
How might we help users track and view their history of mood-based music listening so that they can better understand their moods and how to improve their moods?
Setting the Sensori feature into the existing Spotify architecture
Before I began generating potential solutions to the How Might We statements, I first wanted to assess whether the Sensori feature can be integrated into the existing sitemap without completely restructuring the existing contents. Next, I created 3 user flows to understand whether Sensori feature can interact with existing screens. While the feature does integrate into the system architecture, it seems that several new screens are required.
Ideation
Ideating early concepts for the Sensori feature
Users want to feel their music and have an enjoyable music listening experience. With this objective in mind, a creative constraints brainstorm produced 2 viable ideas:
Concept 1
A dashboard integration that provides a hands-on listening experience where users can choose music based on mood, activity, and track their music selections over time.
Concept 2
In the users' "Liked Songs" library, integrate an advanced search filter to sort through and re-discover their favorite songs and artists.
From two concepts emerges one idea:
Introducing
Sensori
the hands-on listening approach.
Find what you want by mood and activity,
and for those really involved, we have advanced filters
to give you an enjoyable sensory experience.
Find what you want by mood and activity,
and for those really involved, we have advanced filters
to give you an enjoyable sensory experience.
Because, let’s admit it, an AI won’t know you like you know you.
Sketching low-fidelity screens and digital wireframes to test early
Since our Spotify listeners use the mobile app primarily on Android devices, we’ve visualized key mobile screens based on 3 flow scenarios. Transforming the sketches into mid-fidelity wireframes will be effective in early stage usability testing to scope out any issues in the flows. To evaluate feature usability, I will set measurable parameters for task flows, then iterate on the design before developing the final prototype.
Analyzing the cognitive walkthrough feedback will shape the direction of the high-fidelity designs
To ensure that user’s needs and stakeholder expectations are still met, and to ensure the three core flows were effective in addressing the problems I had defined, I conducted a brief cognitive walkthrough with 3 participants. Since this precedes the feature usability test, fewer participants were recruited for the purpose of quickly identifying flow errors.
Goal 1
Ease of navigation to the feature
Goal 2
Gather feedback on desirability of feature design (interface)
Goal 3
Identify and address any parts of the flows that are not intuitive for users
Post-Interviews: At A Glance Summary
- Conducted mixed-method interviews including a feature inspection and question-asking protocol.
- 3 participants1 who use Spotify premium for music listening, daily.
1participants interviewed are heavy mobile users, thus, the desktop wireframes were not tested. - 3 Task scenario walkthroughs: login, discover & use Sensori History and generate a personalized mood-based playlist using Sensori feature.
Task 1:
Existing user log in
Existing user log in
Task 2:
Discover and test the Sensori History feature
Discover and test the Sensori History feature
Task 3:
Generate mood-based playlist using Sensori
Generate mood-based playlist using Sensori
Feedback was organized into a severity-frequency matrix and revealed key insights:
- High priority & most severe: users can't easily find Sensori and don't understand its function.CTA's are unclear, and language is unclear where user input is required.
- Mid priority & most frequent: presentation of the dashboard design is too clunky and text-heavy.Simplification of complex terminology needed on advanced filter options.
- Low priority: minor changes to components and icons that won't disrupt the main flows.
Design
Making design revisions based on priority ranking and most severe issues
The “liked songs” list view is long and cumbersome to comb through history of favorites without using filter or sort. How can this be simplified?
Problem 1: Mis-interpreted function of Sensori History for a genre.
Revision 1: Separate feature display.
Problem 1: Mis-interpreted function of Sensori History for a genre.
Revision 1: Separate feature display.
Sometimes users don’t know what to search for when they don’t remember the specific artist, album, or song.
Problem 2: Users would like to access Sensori separately from Sensori History.
Revision 2: While Sensori History is an extension of Sensori, this feature will have its own card for quick access.
Problem 2: Users would like to access Sensori separately from Sensori History.
Revision 2: While Sensori History is an extension of Sensori, this feature will have its own card for quick access.
Users don’t understand how the Sensori History dashboard functions– CTA’s are unclear, graphics and the language is too complex.
Problem 3: Breadcrumb navigation is not intuitive to guide users. There is an overwhelm of information in the dashboard and CTA’s are too long and unclear.
Revision 3: Updated breadcrumbs to contrast active vs nonactive states. Clearer visualization of favorite music history organized by date, top 3 moods, and activities consistent with Spotify UI patterns.
Problem 3: Breadcrumb navigation is not intuitive to guide users. There is an overwhelm of information in the dashboard and CTA’s are too long and unclear.
Revision 3: Updated breadcrumbs to contrast active vs nonactive states. Clearer visualization of favorite music history organized by date, top 3 moods, and activities consistent with Spotify UI patterns.
Similar to Sensori History feedback, users were unclear about what Sensori feature does and how to use it.
Problem 4: Breadcrumb navigation was not intuitive, and there is an overload of micro-interactions within the Sensori feature card.
Revision 4: Updated breadcrumb navigation. Sensori feature card includes a brief overview and one clear CTA. Bottom content from screen removed.
Problem 4: Breadcrumb navigation was not intuitive, and there is an overload of micro-interactions within the Sensori feature card.
Revision 4: Updated breadcrumb navigation. Sensori feature card includes a brief overview and one clear CTA. Bottom content from screen removed.
Emotions can be difficult to name, and we want our users to find the right music to fit their needs without thinking too much.
Problem 5: Multi-select options for feelings. Longer time spent deciding on options.
Revision 5: Inspired by Plutchik’s Wheel of Emotions, the goal is to name a feeling related to color and enhance emotional literacy.
Problem 5: Multi-select options for feelings. Longer time spent deciding on options.
Revision 5: Inspired by Plutchik’s Wheel of Emotions, the goal is to name a feeling related to color and enhance emotional literacy.
For audiophiles and explorers alike that want more control over music discovery, Sensori includes an advanced filter option.
Problem 6: All users were automatically directed, unknowingly, to advanced filters.
Revision 6: Language was updated to reflect the type of filters users suggested based on usage, and graphics were updated based on competitor evaluations.
Problem 6: All users were automatically directed, unknowingly, to advanced filters.
Revision 6: Language was updated to reflect the type of filters users suggested based on usage, and graphics were updated based on competitor evaluations.
Integrating the Sensori feature design into Spotify’s design system
From our user research, we know that Spotify users represent a universal range of listeners, therefore, the Sensori feature should seamlessly integrate into the existing brand and style. Additionally, the UI style should be consistent across all devices.
.png)
Testing & Iterations
Testing the Sensori feature on navigation scheme, visual design, and usability
To ensure that user’s needs and stakeholder expectations are still met, and to ensure the three core flows were effective in addressing the problems I had defined, I conducted a brief cognitive walkthrough with 3 participants. Since this precedes the feature usability test, fewer participants were recruited for the purpose of quickly identifying flow errors.
In order to see if my iterations to the mid-fidelity wireframes were effective in addressing the problems I defined, I applied tested my high-fidelity prototype with 5 participants. The unmoderated usability test was conducted through Maze with the same 3 tasks were assigned.
Post-usability test survey analysis reveals that the Sensori feature can enhance discoverability and further customization of music recommendations.
Overall, users found that the Sensori feature is easily navigable, steps are simplified where user input is required, and the information is displayed in a way that is easily digestible and informative.
Making priority revisions
Revision 1: Sensori History dashboard designUpdated to reflect the Spotify branding and iconography for a familiar and consistent feel.
Language has been simplified to reduce possible errors users’ may make.
Language has been simplified to reduce possible errors users’ may make.
Revision 2: Quick access to Sensori History Seamlessly access the Sensori History suggested tag located at the top of the Library page.
Alternatively, the Sensori History can be located in “Search” and the icon will also appear on the Homepage for returning users.
Alternatively, the Sensori History can be located in “Search” and the icon will also appear on the Homepage for returning users.
Revision 3: Simplified terminology for advanced search filtersUsers can either “generate playlist” or continue on to “advanced search filters” during their interaction with theSensori feature.
There are clear instructions indicating the advanced search filters section along with a progress bar.
There are clear instructions indicating the advanced search filters section along with a progress bar.
Revision 2: Easy access to SensoriSeamlessly access Sensori suggested tag located at the top of the Library page.
Alternatively, Sensori can be located in “Search” and the icon will also appear on the Homepage for returning users.
Alternatively, Sensori can be located in “Search” and the icon will also appear on the Homepage for returning users.
Solution
Future directions for Sensori feature
Optimization
Optimize Sensori & Sensori History feature for Spotify
desktop web-app users
desktop web-app users
Voice-User Interface
Adapt a voice-command integration when using Sensori feature (improve hands-of listening)
