#33 - Discovery, Controlling the Menu and Spotify's Podcast Trojan Horse
#33 - Discovery, Controlling the Menu and Spotify's Podcast Trojan Horse
How Product Design Can Help Drive Podcast Adoption
The real world is more uncertain than the Field of Dreams. If you build it, will they come?
Since early 2019, Spotify has spent hundreds of millions of dollars acquiring podcast content. As discussed in my Roganomics newsletter, shifting users from marginal costs music to fixed cost podcasts boosts Spotify’s long-term profitability. This post explores how product design can nudge users towards podcasts, driving business model leverage.
Discovery
Discovery is one of Spotify’s strengths. Helped by the acquisition of The Echo Nest, the company has built a sophisticated recommendation engine. Using collaborative filtering, content-based filtering, and audio features extracted through machine learning, the recommendation engine serves up personalized content. What’s going on under the hood looks like something this:
Spotify’s recommendation engine is the same general idea as Amazon recommending books based on your purchase and search behavior:
Here’s a more technical description and here’s a general overview.
With over 50 million songs and 1 million podcasts on Spotify, discovery is crucial. When done well, it drives better engagement and retention. According to Spotify’s CEO Daniel Ek:
We know that discovery is one of the core benefits that we provide to people. If we can get you to discover one new item of content that you absolutely love every single month, we’re pretty sure you’ll remain a user or customer with us for a very long time.
Controlling the Menu
If you control the menu, you control the choices. Tristan Harris, a former Design Ethicist at Google, notes that this is a very powerful position:
Millions of us fiercely defend our right to make “free” choices, while we ignore how those choices are manipulated upstream by menus we didn’t choose in the first place. This is exactly what magicians do. They give people the illusion of free choice while architecting the menu so that they win, no matter what you choose.
When considering a menu, Harris suggests people ask:
Do I know the menu providers goals?
Spotify’s primary goal is to keep users streaming so that they continue to pay for monthly subscriptions (premium service) and hear ads (free service). A lesser consideration is shifting some streaming from music to podcasts. In the future, Spotify’s CEO Daniel Ek expects about 20% of content to be non-music. An increasingly important question for the company’s long-term financial performance is their ability to orchestrate this shift while keeping users happy.
History provides reasons for optimism. Over time, users have leaned more heavily on Spotify to recommend content. According to the company’s F-1 filing from February 2018:
Many of our Users also rely on Spotify to help soundtrack their day...Our machine-generated playlists have been made possible by our investments in artificial intelligence and machine learning, which power our music discovery engine. We now program approximately 31% of all listening on Spotify across these and other playlists, compared to less than 20% two years ago.
While users can always go off the menu, about a third of listening comes from Spotify’s recommendations.
Menus matter, and Spotify controls the menu.
Product Design & The Podcast Trojan Horse
By controlling the menu, Spotify can put its thumb on the scale in favor of podcasts. There are a number of product design choices that can nudge users in this direction.
As the company’s F-1 notes, about 30% of streaming comes from Spotify generated playlists including Daily Mix, Discover Weekly, and Your Daily Podcast. Additionally, the F-1 states that:
Using our artificial intelligence and machine learning capabilities, we are able to find, promote and program songs whether they are established hits or “hidden gems.” This enables Users to find great content that is personalized for them, but that may not be currently in their personal libraries or at the top of the charts.
This is exactly what the company needs to do with podcasts. It has already proven its ability to do so with music. In Platforms, Promotion, and Product Discovery: Evidence from Spotify Playlists, economists found:
Clear evidence that Spotify has power to influence consumption decisions. We document large and statistically significant effects. The major global playlists raise streams for prominent songs substantially. Getting on Today’s Top Hits is worth almost 20 million additional streams...Playlists also affect the success of new songs and new artists. Getting on the top of the New Music Friday playlist in the U.S. is worth roughly 14 million streams.
Adding more podcasts to personalized playlists is a way for Spotify to get users to listen to more podcasts. This appears to be happening. The company has recently been peppering personalized playlists with podcasts. For example, 30% of the content in the playlist below is podcast and 70% is music:
Similarly, the personalized Daily Wellness playlist below is about 20% podcast and 80% music:
Caveat a very small sample size, but the podcasts were front-end loaded, meaning podcast mix will be higher than the numbers below if the user doesn’t listen to the full playlist:
Controlling valuable real estate is another way that Spotify can nudge users towards podcasts. Being featured in Apple’s App Store can drive a significant boost to downloads (see here and here). The analogous situation for Spotify is choosing what to feature on the homepage. Here’s an example of Spotify using the marquee spot on its desktop homepage to pump p p pump pump p p p pump Joe Budden’s podcast (sorry, I just had to):
Podcasts are also getting visibility in the Spotify app, where they’ve been added to the home screen:
And to the library tab:
Spotify has a strong culture of experimentation and has many product dials to turn to promote podcasts.
Defining Success
Small shifts in content mix from music to podcasts could improve Spotify’s income statement. To succeed, Spotify just needs to change some user behavior, some of the time.
External conditions here are favorable, with a growing audience of podcast listeners. According to Edison Research, the average podcast listener in the US listens to about 26 hours of podcasts per month:
The last time Spotify disclosed similar data was the fourth quarter of 2017. At the time, the average user streamed 25 hours of content per month. Given that the company’s push into podcasts began in February 2019, it’s fair to assume that the vast majority of this was music. Podcasts could be an avenue toward some combination of more users, more usage, and more favorable content economics.
Spotify is well positioned to crack podcasting. It has the tools (recommendation engine and ad tech), the users (286 million and growing), the data, the content (oh hey, Kim Kardashian), the focus, and the business model incentive to succeed. At the same time, podcasts are sideshows at most for tech giants like Apple and Google.
Nothing is certain in 2020, but I like Spotify’s chances in podcasting.
👉 If you enjoyed reading this post, feel free to share it with friends!
For more like this once every weekend, consider subscribing 👇
