Spotify for Podcasters
iOS App
ROLE Research, interaction design, and Hi-Fi prototyping
CHALLENGE/PROBLEM Unlock the potential of podcasts as a more shareable media on mobile phones through the use of Spotify for Podcasters app’s tools.
SOLUTION Machine learning model identifies and extracts most compelling segments from podcaster’s content and creates snippets to be used as ads that would pull users into discovering their feed.
The context
Towards the end of every bootcamp, BrainStation runs an Industry Challenge/Hackathon. What this entails is BrainStation partnering up with an outside company and then breaking up the current cohort of designers, data scientists, and web developers into teams. These teams must then go on to design a digital solution for an ongoing issue that partner company is currently dealing with.
Luckily for me and my fellow classmates, Spotify was the company and the issue they were having trouble with concerned their newly released app Spotify for Podcasters.
The problem objective
Armed with the Spotify for Podcasters app, podcasters would be able to grow their audience and earning potential while giving fans the opportunity to enjoy and be inspired by their creations. The challenge podcasters - as well as my team and I - were facing was finding new ways to unlock the potential of podcasts as a more shareable media on mobile phones through the use of existing Podcaster tools.
In other words, our original “How might we…?” question was:
How might we utilize all of the resources that Spotify for Podcasters has to offer in order to make sharing podcasts as fun, easy, and rewarding as sharing music is today?
To find an answer to this question, the team needed to find out who it was we were exactly trying to help. With only 24 hours to create a design solution that met Spotify’s needs, we immediately went into the research stage of this challenge.
SECONDARY RESEARCH
The new generation
There was a reason why Spotify developed this new app and it was a simple one: the podcast market is exploding in growth as was shown through Spotify’s internal research.
There are nearly half a billion podcast listeners which is the same amount of listeners currently using Spotify.
Despite a 22% increase in smart speaker sales, listeners most commonly listen to podcasts on their mobile phones.
Spotify reaches 32 million podcast listeners per month.
The researchers over at Spotify were also kind enough to provide us with information regarding their target audience: Gen Z. Though often maligned and misunderstood, Gen Z are the demographic that’s helping push the growth of podcast usage.
Gen Z’s podcast discovery rate is growing at more than twice the rate of other age groups.
There was a 37% increase in average podcast listenership among this generation on Spotify between 2021 and 2022.
Digging a little further down, our own research confirmed Spotify’s documentation.
50% of Gen Z use Spotify for podcasts: the younger you are, the more likely you'll be listening on spotify.
In 2022, ages 13-17 especially saw significant listener growth showing increased engagement, with 42-49% growth.
Having determined the target audience, we then turned our attention to the podcasters.
SECONDARY RESEARCH
The (not so) old generation
So who are the people making podcasts? While millions of people create podcasts every year, there hasn’t been real data until now. After some digging on our own, some surprising facts came to light through our research.
69%
Podcast creators are primarily male.
of podcast creators identify as men.
are between the ages of 25-45.
Podcast creators are not young but they’re not old.
73%
have an annual income of $75,000.
55%
Podcast creators are educated and high-income earners.
They were also not just fans and supporters of other podcasts: 52% of creators were also excited about the use of podcast ads to increase their base and engagement. Luckily for them, the Spotify for Podcaster app was also equipped with other tools to help them do just that.
Creators would now be able to track how their show was growing over time.
Play count
They would know long how long people were tuning in and pinpoint the exact moments where they dropped off.
Audience retention
Podcasters could better understand their audience so that would be able to tailor their content and partner with advertisers.
Demographics
By understanding who it was they were speaking to creators would have a better chance at increasing their earning potential in the form of host-read and automated ads.
So the whole process seemed enough:
1. Understand and zero in on your target audience
2. Create content that would engage them
3. Listeners would subscribe to their podcast
4. Watch the money roll in
As you might have suspected, we would learn that this would be a lot more difficult than anticipated.
INTERVIEWS
Understand and empathize
We moved forward and conducted a total of 5 decontextualized interviews (phone and in-person) with podcasters and listeners whose age range was between 15-30 years old. From this we walked away with 2 key insights:
Finding podcasts was hard for listeners - 71% relied on friends for recommendations, while 40% preferred solo searching.
Despite the app's convenience in podcast creation, editing a promo ad could still take creators 2-3 hours.
The whole process of engagement was a two way street and frustrating for both users. For podcasters, however, it was even more so since the app was especially created to aid them in this endeavor. With everything that we learned, we updated the original problem space:
How might we make it easier for podcasters to generate compelling clips of their episodes in order to help listeners discover their feeds?
We realized that the answer to this question didn’t have to fall into the podcaster’s hands. There was actually “someone” else who would be able to provide us a solution who would be more effective than we could ever give.
THE PERSONA
Persona x 2
Noting earlier that engagement was a symbiotic relationship between the podcaster and audience, we decided to create personas for both users.
Persona: PODCASTER
Persona: LISTENER
We then created user journeys for them in order to find out where they may encounter their pain points.
User Journey: Podcaster
User Journey: Listener
For the podcaster editing the content was a large issue. Not only did it take a long time to edit the content, it took them just as long to find suitable content to edit as they scoured their past podcasts. Also they were unsure or just didn’t know how to leverage the app’s tools to better advertise themselves other than uploading an episode.
On the other side of this equation, listeners were forced to search among the hundreds of podcasts in order to find something new that engaged and bettered themselves (We learned through the research that personal learning and growth were important to this generation).
From these new insights we revised our personas user journeys.
Updated User Journey: Podcaster
Updated User Journey: Listener
So the crux of the problem was generating a snippet from a podcast that would in turn be used as advertisement for returning and potential listers. The time that it took for a podcaster to take a clip or snippet was an issue that ran parallel to the listener’s own pain point as they were forced to scroll through podcasts - favorites and new - to find something that spoke them.
For us, a machine learning model would prove to be the best solution.
THE SOLUTION
Machine learning model
IA machine learning model would identify the most engaging and relevant parts of a podcast from which it could then extract the most compelling segments of a podcaster's content. From there it would then be able to generate snippets from the content to be used as advertising for the podcaster.
Machine Learning Journey Map
A bonus to to this would be that the machine learning model would be able to identify listener preferences and then use this information to suggest other content that they would likely to enjoy.
The visual end result was simple and straight forward:
The machine learning model analyzes the podcaster’s content…
…and then produces a 20-30 second snippet that would then be advertised on the original Spotify app.
Utilizing Spotify’s basic cards, we placed the advertisement directly in view right at the top of the screen with the title of “Discover & Engage Feed”. Normally when a listener previously clicked on a podcaster card, it would send them straight to the actual show about that particular episode.
Previous…
This time around, however, the machine model would also provide a list of other podcasts that had been curated for that particular listener.
…New
KEY FINDINGS, NEXT STEPS
Spotify is impressed
Unknown to the team, the people at Spotify had been exploring similar solutions but were, however, unsure as to how to implementation it. Our winning solution provided them with another avenue to explore. The other teams had also gone down a similar road that we had taken but the Spotify executives were impressed as to how we were able to incorporate the needs of both the podcaster creator and listener as seamlessly as we did.
As far next learnings go, 3 major next steps stood out for us:
Our user sample size was too small. In order to really zero in on the solution we needed to dig deeper into both users needs and wants.
More data
We never had the chance to actually put our solution to the test with real users. A/B testing results would help a lot in future iterations.
Usability testing
This first incarnation of the Spotify for Podcasters app was simple, an MVP if there ever was one. Further iterations could stand to have a little bit more engagement in terms of UI.
UI improvements
Personal takeaways
During my time at BrainStation I had participated in a number of collaborative projects. This challenge, however, gave me the chance to work with not only other designers but also data scientists and web developers. I learned what it really took to complete a product from start to finish with people who had a professional mindset.
Though it wasn’t without a few bumps, in the end I learned that there were a lot of different ways to solve a design problem if stood back a little to see the whole picture.
I also learned another important lesson: I’m looking forward to doing it again.