How do users find new podcasts to listen to?

Academia shows that there are millions of users that listen to podcasts, and of these millions of users, the average listener listens to 7 podcasts per week. This brings up the question: How do users find new podcasts to listen to?

The objective of this study is to determine if more podcasts would be discovered if users had a centralized and personalized database to find new podcasts, therefore increasing revenue and reach for podcasters. My role in this study is a UX Researcher and Designer.

  • Problem

    There is no easy way to find podcast recommendations based on a user’s listening history.

  • Goal

    To create an easy-to-use and personalized database of podcast recommendations.

Discover


There is an opportunity gap for personalization.

I began my research by figuring out what the need for a podcast recommendation database is and, if there was a need, what was the target audience to test with. Spotify, which is the leading podcasting app, has over 28 million users that listen to podcasts, with Apple Podcasts trailing with about 28 million users. Next, I found that the primary user is a 12-34 year old, white male, living in the United States. Moreover, according to a 2021 Forbes survey (found here), 90% of users tend to listen in their homes and specifically to have “me time”, to relax, and to form deeper connections with tasks such as gardening, working out, or drawing.

I surveyed 20 candidates and picked 5 users that met the primary user criteria to continue user interviews with. I surveyed these users deeper about their listening habits. Almost all users noted they grab piecemeal advice on recommendations (i.e., from other podcast hosts or friends) and would benefit from a centralized database of some sort.

  • Job: Relationship Manager

    Favorite Genre: Comedy

    User A noted they liked to listen and watch podcasts. User A conveyed they are subscribed to shows on Patreon and Spotify, and when asked if they would continue to use Spotify if the features available on Patreon were available on Spotify, they agreed, mentioning it would be “one less step” to listening. User A finds all of their recommendations through podcast hosts.

    “I know what I want [on Spotify] and I click on it without thinking”

  • Job: Editor

    Favorite Genre: Comedy

    User B utilizes Spotify’s “offline listening” feature more often than not, noting this was important to them. User B also mentions their dislike for Spotify Podcast’s section, clarifying that the section seemed “clunky”. User B finds a majority of podcast recommendations through social media platforms, such as Twitter, Discord, and Instagram.

  • Job: PR Specialist

    Favorite Genre: True Crime

    User C notes that she enjoys all of her listening experience within one app. She enjoys being able to frictionlessly move from one form of media to another. User C notes she would like the ability to leave comments as well as star ratings, which is something Spotify does not have currently.

  • Job: UX/UI Student
    Favorite Genre: Comedy / Education

    User D uses search engines such as Google to find most of her recommendations. User D noted she likes being able to sort through her podcasts a certain way and has difficulty doing this when using Spotify, especially on the web-based program.

  • Job: Content Creator

    Favorite Genre: True Crime

    User E explained that she listens to podcasts one at-a-time, going through the entire list of available episodes prior to moving on to another podcast show. User E also conveyed frustration at Spotify altering their personalized playlists. User E finds new podcasts to listen to through podcast host recommendations and social media.

Design


MVP should include multiple genres to select from, a podcast show rating system, and an opportunity to watch video podcasts.

After knowing the pain points and likes of my 5 users, I was able to develop affinity maps, user flows, and a primary persona to help evolve the app. The biggest pain points noted were the inability to leave a comment or rating, the “clunky” user interface, and the inability to watch a podcast episode in the same app that the user is listening to the podcast in.

I used this data collection and developed John, the primary persona that I would work with to further develop wireframes and low-fidelity prototypes.

Figure 1: Interpretive coding.

Figure 2: The primary persona.

Ideate


Testing and iterating with new users to develop high-fidelity screens for Podium, the personalized podcast recommendation application.

I created wireframes and low-fidelity screens to perform usability tests on 5 new participants which resulted in the most critical updates of the project. These updates included fixes to simple usability issues, informational hierarchy updates, iconography changes, and new screens to incorporate features previously missed.

Figure 3: Red Route #1 and #2

Figure 4: MVP low fidelity mock-ups used in usability testing.

Figure 5: More wireframes.

Using a working wireframe/low-fidelity prototype, I performed a series of tests with 5 users. I asked these users to perform basic tasks, such as reviewing their recommendations, saving any 1 podcast and skipping any 1 podcast, and playing any episode from any of their Saved Podcasts. The information received from this usability study allowed me to perform my first bigger iteration which included changing the Home Screen, the bottom navigation panel, and some icons.

I also updated some other minor things, such as replacing the left and right facing arrows with a heart and a deny sign to symbolize “saving” (the heart) and “skipping” (the deny) a podcast.

Believing that my wireframe prototype was in a good place to move on, I began developing my style guide and used this guide to begin making high-fidelity screens.

I used colors found in galaxies, representing the vastness of the unknown as inspiration, and chose icons that were simplistic enough to get the point across but weren’t too detailed that they’d distract from the app itself. I also chose to use SF Pro font type, given this is a common typeface among apps.

Figure 6: Iconography, typography, and color scheme.

I chose to guerrilla test 5 new users with the new high fidelity prototype to get a sense of what can be updated to make it easier for the everyday user. My main findings were that users weren’t going in the requested direction when asked to review their recommendations; all users chose to click somewhere within the “Explore something new” area - or the search bar - on the Home Screen rather than the Recommendations button.

Understanding that I needed to make a change, I decided to use the prime real estate on the Home Screen for something else: a welcome message for the User. There was no better solution than giving a personalized message to the user in the area that used to be owned by “Explore something new”. Additionally, to add a little bit of unique-ness to the message I chose to add a microphone as an ode to the art of podcasting.

Furthermore, to give the Recommendations button a figurative neon sign that says “click me!”, I replaced the Explore button with the literal word “Recommendations”, and replaced the search button with a Home button. Now, the user has a slimmer pickin's when looking at the Home Screen. The user’s options now are: A, search for a specific podcast, B, listen to a podcast they were just previously listening to, C, go to their library, or D, go to their recommendations.

Figure 7: Example of high-fidelity registration/onboarding.

Figure 8: Final high-fidelity screens as a result of guerrilla testing.

Figure 9: More high-fidelity screens.

I continued tests and iterations until I honed in on what you now see as Podium, the tailored podcast recommendation app.

The biggest lesson learned from this project is to expect the unexpected. Going in, I had assumed that my participants would understand how to use the app as if they just similarly spent tens of hours staring at it, however I was very wrong. If the application is based on what I believe is common knowledge (such as using a left facing arrow for “no!” and a right facing arrow for “yes!”), but it really isn’t common knowledge, then I would be working on something that, in the long run, won’t make it very far. I found out pretty quickly that I needed to take a few steps back and look at the app with the freshest eyes that I could in order to see the issues right in front of me.

One thing I loved about this process was during my testing rounds, after the user was finished with their tasks, I asked if they could just poke around the app and let me know what they find. The user had no direction or destination, but instead had the freedom to use the app how it was intended to be used: educatedly blindly. I know that people download apps they have no idea how to use and need to take a few minutes to click on random things to get their bearings. I wanted to do this with my participants so they could get their bearings, and I could get important information. During these moments I found areas in my prototype that weren’t linked, some random gestures saying “oh, I thought XYZ would be here instead of there”, and of course some praise for making a unique experience.

I am very proud of Podium, and wish it were alive to be used by all. For now, please feel free to watch me poke around below, or choose to poke around yourself.