AdobeStock_201393709.jpeg
VoxEU Column Competition Policy

Platforms, promotion, and product discovery: Evidence from Spotify playlists

Streaming platforms give a diverse range of music creators direct access to large audiences, but whether their songs reach these audiences can depend on the platforms’ decisions about what to promote. This column examines the impact of inclusion on Spotify playlists. The overall effect of appearing on the Spotify-curated Today's Top Hits playlist is about 19.4 million streams on average. Inclusion in top selling or new music playlists also has a significant impact on song success. The results suggest that growing concentration in the streaming market, as well as other markets dominated by one or a few players, may create a need for scrutiny of how platforms exercise their power.

Digitisation has vastly increased the amount of new music produced and, following the advent of streaming platforms, has raised the number of songs available directly to consumers. In 2017, Spotify added nearly a million songs to its catalogue, giving its users access to 35 million tracks.1  Streaming platforms have thus substantially levelled the playing field for music creators by allowing products from diverse sources – such as independent record labels and foreign producers – to find direct access to large audiences. Recent research shows that Spotify adoption indeed leads to an increase in the variety of music consumed (Datta et al. 2018).

Product discovery and playlists

Access to an increasingly large catalogue creates a daunting problem of product discovery for consumers, however. Against this backdrop, a major value-creating function of music streaming platforms is to help consumers find products that they like. Broadly speaking, they do this by suggesting music through playlist that are generally made available to their subscribers (like Spotify's Today's Top Hits) or personalised to each user (like Spotify's Discover Weekly). 

If these playlists affect individuals' consumption decisions, then platforms can play important roles in determining song and artist success, including the determination of which songs and artists are discovered in the first place. It follows that creators could be highly dependent on platform decisions about which songs and artist to promote. 

A handful of online platforms have become dominant – or nearly dominant – in their respective markets, such as Google in search advertising, Facebook in social networking and Amazon in online retailing. The music industry is no exception in that regard and is witnessing growing concentration in the market for interactive streaming. While the market has a number of major participants, including Apple Music, Google, Deezer, and Spotify, the latter is growing quickly and was reported to have 37% of the subscription streaming market in 2017.2

In a recent paper (Aguiar and Waldfogel 2018), we explore whether Spotify has the ability to influence users' listening decisions through its playlists. In particular, we ask whether playlist inclusion affects the number of streams that songs receive and whether it affects consumers' discovery of new songs and artists. We collected data on the daily top 200 songs streaming at Spotify for 26 countries as well as on the songs that appeared on Spotify's major playlists during 2016-2017. We then focus on distinct types of playlists to assess their impact on song performance.

Different types of playlists

The general playlists offered by Spotify – the ones that are not customised to each user – can vary along two dimensions: whether they are algorithmic or curated by humans, and whether they are global or country-specific.

Global playlists

Playlists like Today's Top Hits are curated by Spotify employees who choose which songs should be included on the list. These lists generally add songs that have been streamed on Spotify for some period of time and are therefore often used to promote already widely known artists. Because playlists curators usually choose to include songs that they expect will be popular, identifying the causal impact of playlist inclusion on song performance is empirically challenging. The fact that the songs appearing on these lists have already been streaming for some time nevertheless allows us to look at the change in streams right around the time that the song appears or is removed from the list. Because major playlists have a large base of followers, a song that is added to one of these lists will witness a sharp increase in the number of users exposed to it. Figure 1 shows the change in streams following the inclusion and removal from Spotify's most followed playlist Today' Top Hits. Streams rise (and fall) substantially at the time they are added to (dropped from) the playlist.

Figure 1 Today’s Top Hits events

We estimate that the average effect of appearing in Today's Top Hits is around 260,000 worldwide daily streams. Because songs tend to stay on this playlist for an average of 74.4 days, the overall effect of appearing on Today's Top Hits is about 19.4 million streams. And given Spotify's recent reported payments of roughly $4 per thousand streams, this translates to about $77,000 in payments from Spotify alone. 

Another global playlist provided by Spotify is the widely followed Global Top 50. Instead of being curated using human discretion, this list algorithmically includes the 50 globally most streamed songs of the previous day.  Since we know the inclusion criteria for this list, we can compare the streams of songs just making the Global Top 50 list to identify the effect of inclusion on streams. Figure 2 shows how the decline in streams of going from rank r to rank r-1 hovers around 2% for ranks 40 to 50 of the Global Top 50. But going from rank 50 to rank 51 leads to a decrease in streams of about 6%, indicating that being added to the list adds about 4% to streams. Given the average global streams for a song ranked 50, and given the fact that songs spend on average about 50 days on the list, we estimate the overall effect of appearing on the Global Top 50 to be around 3 million streams.

Figure 2 Global Top 50 and rank streaming gradient

Country-specific playlists

Every week, Spotify constructs a rank-ordered list of 50 new songs for each country in which it operates. These playlists – called New Music Friday – are updated every Friday and generally include songs that are new to Spotify, bringing new information to consumers and offering the possibility of promoting the discovery of new songs and artists.

Does appearing on the New Music Friday lists increase the probability of a song's ultimate success? Figure 3 shows the share of songs at each New Music Friday rank that ultimately appear in the corresponding country's top 200 and top 100 daily streaming charts. It shows that songs with better ranks are more likely to appear in the top of the daily charts. For instance, close to 85% of the songs ranked #1 in a country's New Music Friday list appear in the country's top 200 daily streaming charts. While this is suggestive that high recommendation ranks matter for performance, the relationship depicted in Figure 3 is also driven by the ability of curators to predict which songs are going to be successful.

Figure 3 New Music Friday and Spotify chart appearance

In order to identify the causal effect of higher-ranked recommendation on streaming performance, we exploit the cross-country variation in the New Music Friday rankings. Figure 4 illustrates the idea by comparing the US and Canadian New Music Friday lists released on 10 December 2017. The rankings are positively correlated, but they are substantially different. While The Wombats were ranked 20th in Canada on that day, they were ranked 37th in the US.

Figure 4 New Music Friday ranks in the US and Canada, 10 December 2017

Assuming that countries have similar tastes but are treated with different rankings, we can then measure the effects of the New Music Friday rankings by comparing the streaming performance of the same songs in different countries where they have received different rankings. As depicted in Figure 5, songs that obtain a number 1 ranking are about 50% more likely to appear in the streaming charts (relative to a song ranked 50th) and this effect falls sharply with the rank, to about 18 percentage points at rank 10 and to about 4 percentage points at rank 20. Focusing on new artists only provides similar results, indicating that the New Music Friday lists indeed help in the discovery of new artists.

Figure 5 New Music Friday rank effects: Overall

Songs remain on the New Music Friday lists for only seven days, so it is interesting to see whether these effects continue past the time on the list. Figure 6 shows that these effects are indeed persistent, indicating that the New Music Friday playlists bring new information to consumers and are note merely used as utilities for listening to the new music that they present. When looking at the overall effect of inclusion on total streams, our results show that the benefit of being ranked number 1 on the US New Music Friday playlist is worth about $55,000 in payments from Spotify alone.

Figure 6 Effect of appearing on New Music Friday on Top 200 charts (by rank, first 14 days)

The fact that playlists have substantial impacts on song success should be of interest to both music industry participants and observers of platforms more generally. Growing concentration in the streaming market, as well as other markets dominated by one or a few players, may create a need for scrutiny of how platforms exercise their power.

References 

Aguiar, L and J Waldfogel (2018), "Platforms, Promotion, and Product Discovery: Evidence from Spotify Playlists," NBER Working Paper No. 24713.

Datta, H, G Knox, and B J Bronnenberg (2018), "Changing their tune: How consumers’ adoption of online streaming affects music consumption and discovery," Marketing Science 37(1): 5-21.

Liebowitz, S J (2004), “The elusive symbiosis: The impact of radio on the record industry,” Review of Economic Research on Copyright Issues 1(1): 93-118.

Sorensen, A T (2007): “Bestseller lists and product variety," The Journal of Industrial Economics 55: 715-738.

Endnotes

[1] See https://www.sec.gov/Archives/edgar/data/1639920/000119312518063434/d494294df1.htm#rom494294_14  and http://everynoise.com/sorting_hat_closet/ for thetotal number of tracks available on Spotify and the lists of songs added weekly.

[2] See https://www.statista.com/statistics/653926/music-streaming-service-subscriber-share/.

[3] From that perspective, the study of playlist impacts on streamingresembles the question addressed in studies of the impact of airplay on recorded music sales (Liebowitz 2004).

[4] Measuring the impact of most-streamed lists on streams is related to existing work on the impact of best-seller lists on sales and product variety (Sorensen, 2007).

1,260 Reads