Spotify Marketing Platform Analysis

Introduction

Leveraging large amounts of data and advanced machine learning, Spotify created an experience that became the new industry standard, indulging listeners with an endless stream of music. However, the platform faces backlash from it’s users for a growing lack of diversity in music recommendations, caused by the concentration of listeners towards major tracks rather than smaller artists. In this paper, we will explore Spotify’s role within the music industry, it’s strengths that have made it the industry leader, and how it can create a more equitable platform for smaller indie artists to compete amongst industry giants, resulting in more diverse and unique experiences for listeners. 

Background

With the decline of the iPod (2000-2010) and the rise of streaming platforms, Spotify emerged as the new leader in the music industry. Rather than owning the music as users did with iTunes, streaming platforms such as Spotify provided a service to users that gave them instant access to a vast library via the cloud. Spotify rose in popularity by giving users an endless amount of selection at their finger tips.

Platform businesses, like Spotify, operate by connecting a large number of suppliers (Artists) to a large number of consumers (Listeners), and acting as the intermediary between them. Depending on how the platform is architected it can often control the user experience and dynamics that happen at the edge of decision making, resulting in larger downstream impacts to how successful tracks are, often determined by it’s algorithms.

There are two sides of the market which are made up of different demographics. On the Supply side, there is a diversity in range, but a concentration in power amongst the biggest record labels. Known as the Big 3 record labels – Universal Music Group, Sony Music Entertainment, and Warner Music Group represent the lion share of the market, taking up 90% together. Artists backed by record labels boast higher streaming numbers from the time of release which creates a snowball effect in how the tracks are pushed and recommended throughout the platform. Due to their large marketing budgets outside of the platform, and possible deals behind closed doors with Spotify, these artists enjoy the benefits of reach that platforms like Spotify can provide. Their songs are pushed to the top of the charts and shared across the platform through algorithmic momentum.

However, other artists, such as smaller record labels and independents who release music from their bedrooms, struggle to gain popularity and break through the masses. Smaller artists tracks struggle to build momentum and ride the algorithm waves due to slower starts in listener adoption. Though Spotify doesn’t actively advertise that Artists that are backed by record labels, the unequal strength of artist’s backing can be felt throughout the platform. Smaller artist’s are slower to build momentum, and without critical early plays, they miss out on algorithmic gains that bigger artists benefit from. The result is an outsized impacts on stream performance for larger artists vs smaller artists. Major artists disproportionately benefit, while smaller artists feel the challenges of the platform and fail to compete effectively.

From a listener’s perspective, Spotify currently boasts 600 million users who listen to an average of 26,637 minutes per user (Reference 1) which shows the amazing scale that the platform is operating at. Users can’t possibly listen to all of the music on the platform individually, so they rely on Spotify’s recommendation algorithms, editorial playlists, and smart recommendations to find the music that best fits their tastes.

They fall into two categories of needs – finding music from artists they already know (the hits), and finding new music from new artists which they’ve never heard about (the niches). Being able to find a range of music adds to the novelty of their experience and caters to the unique tastes of each individual. Everyone wants to be in the know about artists that are hidden gems and future breakout stars, and everyone also wants to hear the hits which are making waves culturally. Spotify does a great job tracking user preferences to learn about the user through many features such as it’s Taste Profiles, and recommends tracks that it think the user might like that are also growing in popularity. (Reference 2)

As the platform acting as an intermediary between the artists and the listeners, Spotify has to be careful in how it manages the distribution of attention for larger artists vs smaller ones, and how it can best maximize the experience for listeners on the platform.  Given the amount of music on the platform vs. the amount of time that users listen to music, it’s impossible for all users to listen to everything, so they rely on Spotify’s platform to help them find the best music that fits their tastes and makes their time worthwhile.

Business Model, Strategy, & Competition

In terms of business model, Spotify has a variety of ways in which it conducts business & generates revenue. From the customer’s side, they have 2 tiers of subscription plans – ad-tier and premium subscriptions. From the business perspective, they generate money through royalty deals ranging in 30% cut of streaming profits. They also sell ads in the ad-tier to marketers who want to reach their listeners audience. All of these revenues combined create Spotify’s revenue strategy.

From an Artist’s monetization perspective, we see a common theme that the numbers really only make sense for big players as well. Artists receive a portion of revenue in a revenue share model with Spotify - resulting in $0.003-0.005/stream. To put this in perspective, where 1 million streams seems like an amazing feat, it only results in about $3000-$5000 dollars in revenue for the artists.

On the contrary, for bigger artists, it's easier to put together a few million streams, and the economics makes more sense past this point. If a song reaches 5 million streams, the result is about $15,000-25,000. Because the linear relationship of the payment model, the payout for smaller artists is relatively less given that the effort needed to generate those plays is going to be higher compared to bigger artists.

Other music streaming platforms such as Tidal, Apple Music, and YouTube music have been gaining in popularity in recent years. From a competitive standpoint, these streaming platforms are just as successful as Spotify in accessing the leading major record label releases (the hits), with where Artist’s only releasing on one platform being an anomaly.

Amongst it’s competition, Spotify’s platform stands out against it’s competitors for it’s strength in leveraging data and machine learning to create a seamless listening experience that is very powerful. In addition to having the must-have library, Spotify users crave novelty and discovery of new artists that they’ve never heard before.

Spotify boasts it’s strength in data and machine learning amongst a number of features such as – Song Radio where users can select one song and create an entire playlist based off of that one song, which is similar in genre, popularity and other things. It has a Smart-shuffle feature (Reference 3) which can enhance user’s playlists with additional songs based off of the one’s provided. Spotify also has a AI DJ feature (Reference 4) which can create a mix of music that it is confident you will enjoy in minutes.

Spotify Wrapped also shows the companies strength in data – showing a culmination of user’s listening habit over the year, which is also a great marketing tactic. As part of Spotify Wrapped this year, we got a glimpse of what is happening on the back-end of the data when Spotify revealed this screen to users (Exhibit A). On a similar note, Daylist (Exhibit B) is also uses machine learning to categorize your taste of music depending on the time of day and your preferences. Similar to how Netflix created 1,130 taste communities to categorize it’s users, Spotify’s Wrapped and Daylist features show that users are following a similar categorization and cohorting on their back-end. (Reference 5)

The Problem – Concentration of Big Player Dominance Creates Repetitive Experience for Listeners.

In an ideal world, artists without a record label can leverage platforms like Spotify to level the playing field where they can compete with the likes of the most popular artists. However, given the concentration problem that happens on platforms, this doesn’t necessarily become the case. From a data machine learning perspective, the most streamed artists become disproportionately successful against those smaller artists that don’t have a lot of data. The result is that the big players keep getting bigger, and the smaller players have trouble surviving. This problem is not only unique to Spotify, but affects other platforms such as social media, streaming, food/bev, etc.

As a result, Spotify is recently facing a lot of backlash. Users are increasingly complaining about the lack of diversity in recommended tracks, which often come from the same Artists or well-known artists that are similar styles. The machine-learning recommendation features, which were once effective, are as engaging as they were before. Spotify is failing to create an engaging experience that can help users discover new music that they actually enjoy (Exhibit C). Many users report difficulty in finding fresh, new tracks, leading to frustration, and decreased interest. This ultimately results in  lowering subscription rates, reducing engagement, and lessening time spent listening on the platform.

The ultimate goal for Spotify is to create the best listening experience for its users. This means recommending a blend of music from major and independent players alike, raising the benchmark for diversity. By introducing more spontaneity and novelty into it’s listening experiences Spotify can reignite engagement from its users. Solving this problem effectively, it can result in a substantial increase in listening time and longer subscription LTVs as users are satisfied with the platform and excited to hear new, engaging content.

Below are 3 recommendations that Spotify can implement to accomplish this task.

1)    Spotify creating tools for targeted advertising for Artists

To enable effective artist marketing, Spotify should develop tools that allow artists to target specific listeners. Similar to Keyword Auctions on Google Adwords, or Demographics targeting on Tiktok and Instagram, Spotify can help enable artists select the exact audiences they want to reach. Spotify can achieve this by leveraging its Machine Learning algorithms to create cohort-based advertising, as seen in the Spotify Wrapped feature. For example, an artist might want to target listeners of niche genres like lo-fi, female rap, or underground hip hop. These custom genres built through Spotify’s machine learning provide a strong foundation for creating targeted user groups that artists can advertise to. For instance, an acoustic artist like Mac Ayers might want to expand his audience by targeting listeners of Country or Jazz-hop. Spotify should enable artists to target these specific users directly.

Targeted advertising can also align with Spotify’s Ad-Supported Tier to create synergies. Artists could design 15–30 second audio clips showcasing a sample of their music with a call-to-action (CTA) to “tap to listen” or “save this song” as a form of direct marketing.  This feature allows listeners to discover new music via advertisements while enabling artists to precisely target who sees their ads for the best conversion rate. Additionally, this approach generates ad revenue for Spotify, and the same principle can be applied across other mediums, such as video ads, banner ads, or other in-app promotional tools.

2)    Better Training, Transparency, and Control of Algorithms

Spotify’s machine learning algorithm has been successful, and key differentiator amongst competitors, but there is still room to be improved. Currently it leans heavily on non-direct engagement metrics to learn about user’s preferences, such as ‘plays’, ‘add to playlists’, and ‘favorites’. While these metrics are directionally useful, they don’t fully capture the full range of user’s sentiment. For example, a user may love a song but fail to save it because they’re busy, or a data point may not be captured accurately.

Additionally, Spotify’s current system does not effectively account for disliking songs, which limits the ability for the algorithm to learn what users don’t want to hear. Introducing a like/dislike feature would enhance recommendation accuracy. However, Spotify should implement this feature carefully, as disliking tracks could create backlash from artists.

With the advent of AI, Spotify can take this further by enabling users to prompt the platform directly. Users should be able to interact with Spotify to express their likes, dislikes, and preferences. Users should be able to control their data by accessing their listening profile to ensure it reflects their preferences. They should also have the ability to edit or adjust recommendations based on what they want to hear or avoid.

Providing this level of transparency and control empowers users to take charge of their listening experience rather than depending on an algorithm that they don’t fully understand or control.

Releasing these features will help users escape the trap of repetitive recommendations that cause echo chambers in the way they consume content, opening the way for newer content to be shown to them.

3)    Establish Credibility through Social Sharing 

Spotify has already made significant progress in supporting smaller, independent artists through its Editorial playlists. These playlists, curated by professionals at Spotify, highlight tracks with potential and hand select them to be promoted to playlists with large followings. This process is very similar to the way that professional movie critics might endorse lesser-known films, such as those at the Sundance Festival, to bring them to mainstream audiences. However, this feature only puts power in the hands of critics instead of every day users, who’s voices and opinions can be just as powerful.

A feature Spotify could lean into is social re-sharing of tracks to leverage social endorsement and increase artist’s promotion. This feature would allow users to re-share what they’re listening to with their social circles, helping create virality for smaller artists, similar to ESPN’s Remarkable model. Re-sharing is especially valuable because it captures an important metric that goes beyond play counts. It enables users to express enthusiasm for tracks they love, even if they don’t listen to as much music as other power users.

The most passionate listeners can act as advocates for their favorite artists, boosting their credibility and visibility. Re-shares could appear on artist profiles and individual tracks, providing a new layer of credibility and encouraging other users to discover these artists.

This helps solve the biggest problem for newer artists on the platform – the cold start problem. For artists that are just starting out, they can leverage this word of mouth sharing to create momentum and establish credibility in their music. In the long run, solving the cold-start problem can help solve the machine learning algorithm’s recommendation problem, leading to more shares, greater virality, and engaged listeners throughout the platform.