CSPC: Drake Popularity Analysis

Streaming Part 4

While the consistency doesn’t change much from previous albums, the novelty on Views is the terrific hits it contains. Of course, all his previous albums had hits but since they came out before the streaming era, they accumulated plays only as catalog songs, being streamed as a part of their albums quite often. Views got released last year with streaming being already massive when the likes Hotline Bling and Too Good were first issued. Both together are over 1,2 billion streams, a figure One Dance eclipses on its own. How incredible is that feat?

The near-total absence of One Dance from YouTube is one part of the explanation of its unreal Spotify results. Hotline Bling in the other side was huge on both platforms. Unsurprisingly, a streaming success as massive as this means massive equivalent album sales too, a mind-blowing 4,73 million in this case.

More Life has sold less than a fourth of Views units in pure album format. It isn’t a good reflection of their respective popularity. Indeed, streams are way more solid for More Life proportionally speaking. Its entire track list is well over 30 million streams at Spotify already. The choice of pushing the more valuable audio streams ahead of video streams is fairly obvious. Both Fake Love and Passionfruit average 400 million, much lower than One Dance but still huge. All in all, the playlist is worth more than 2,3 million equivalent album sales.

12 thoughts on “CSPC: Drake Popularity Analysis”

  1. Hi MJD!

    Nice analysis as always, but I have 1 main question concerning how you calculate streaming equivalent sales, particularly audio streams.

    Your method of calculating audio streaming equivalent sales of records is by multiplying the cumalative streams on Spotify with the ratio between the total audio market and Spotify in the current year, and finally dividing it by 1500.However, one problem I see is this: the growing share of Spotify among streaming platforms, hence a much smaller ratio for each year.

    For older acts (eg. Beatles), their streams increase at a relatively slow pace. If we were to multiply their cumalative streams with an updated ratio, their streaming equivalent sales will undoubtedly be lower every year as the ratio keeps getting smaller and smaller.

    However, this doesn’t concerns those acts as much as acts of today that achieve huge streaming (eg. Drake) While streaming only takes up a small percentage of the total CSPC total for the former, the latter is the total opposite, with some acts total CSPC sales having more than half coming from streaming. What’s more, with a dying market of album sales and downloads, these newer acts heavily rely on streaming to generate catalog “sales”. If the records of these acts don’t increase in streams at a certain pace on Spotify, then using a decreasing ratio every year will cause their totals to will remain more or less the same, perhaps even lower. Hence the concept of catalog sales is meaningless to them.

    What I suggest doing is to multiply the streams of a record it achieved on Spotify during a year with the ratio of that particular year, and using the updated ratio for another year on the Spotify streams it achieved for that year. For example, let’s say that an album achieved 500 million streams and 200 million streams for 2019 and 2020 respectively, while the share of Spotify among all audio streaming platforms is 40% and 50% respectively. What I suggest is to calculate like this: ((500m*100/40)+(200m*100/50))/1500= (1,25B+400m)/1500=1,1m audio streaming equivalents for 2019 and 2020

    Of course, this is only a suggestion. Tell me your thoughts on this. If you were to use this formula I proposed, the only real problem is to keep a large database that shows how many streams an album generated each year and the share of Spotify among all audio streaming platforms each year as well.

    1. Hi Raffi!

      The main error on the old formula for streams was the paid users vs total users confusion. The formula was using the percentage of paid users of streaming services that were using Spotify as there was no data on IFPI report for free users. The problem is that paid-only platforms appeared, creating a very different share of “paid users” and “free users” using Spotify. The formula used the share of paid users, while counts of streams were built by all users combined.

      Now that this is fixed and that each streaming platform took its place, I’m not expecting notable changes. For nearly 1 year now, Spotify goes up and down between 62% and 64% of the overall market, this is quite stable at the moment. The ratio is someway wrong for years 2012/2014, but the impact is minimal there. As noted on updates, the artists roughly doubled their streams during the last 10-11 months, e.g. last 2 years represent a good 80% of the total streams to date of an artist. If the formula deflates an artist results for pre-2015 years by 20% for example, that would be 20% out of 20%, an error around 4%, a percentage that gets smaller every week of new streams which lower even more the importance of ‘old’ streams.

      Obviously, if a new change of policy, for example Spotify going premium only, changes the share of all services, the formula will be adjusted. I do not exclude the possibility of ‘dating’ streams as you mention to apply the most relevant formula for each stream. I stay optimistic there though about shares staying flat during the next months/years!

  2. Hi, I just noticed for Drake all his features sales/streams are given to him 100%. Why is this?

    It doesn’t seem to make sense to give a feature 100% of a song’s sales/streams, even if they were on the song for 5 seconds. Janelle Monae shouldn’t receive 100% of We Are Young’s sales/streams.

    You also made the formula for video streams 11,754 = 1 album versus 1,500 = 1 album for audio streams, the reasoning being that video streams pay much less than audio.

    An artist gets paid much less of a fraction from features than as lead artist, so shouldn’t they receive a fraction of the sales/streams from features?

    1. Hi Anon!

      The case of features is definitely a tricky one. To be honest, I don’t like the 100% attribution that much, but I can’t see a better solution. I don’t like using a fraction, say 50%, for two reasons: 1/ it is artificial, 2/ not all features are as relevant. Many blame Rihanna due to her numerous featurings, but on most of them she has a role at least as big as the lead singer. In the other side, someone like T-Pain or Nate Dogg were never the #1 singer of a song, Quavo seems to be following the same road.

      For this reason, I can’t fix a frozen percentage. I also can’t dictate a distinct share for each feature as 1/ it wouldn’t be valid, 2/ that would imply to listen / gauge each and every song for each artist. That’s not possible.

      It must be said that I took that decision when starting the first CSPC article with Rihanna and having in mind that I would study mostly major artists that are also pretty relevant to the songs on which they contribute. If tomorrow I work on Lil Jon, I’ll most likely build an appendix to the concept to avoid getting a flawed total due to unweighted featurings!

      1. Thanks for the reply MJD!

        However, I think you should set a frozen percentage for these reasons:

        — Downloads have a frozen percentage. Even when some songs sell majority for 69 cents these days, instead of $1.29. Therefore despite half the price, they get the same representation in CSPC.

        –Same applies with physical singles, frozen 1:3 weighting. Many physical singles before the late 90s/early 00s sales crash were $3.99+; however when single sales crashed many artists sold singles for as low as $0.49. They still count similar despite a price difference of near 10x.

        –An album sale still counts as 1 sale, even if sold for $14.99 or 99 cents. Club sales are also counted which were commonly sold for below $1 through membership.

        –Streams have a frozen weight despite some streams coming from $10+ per month subscribers, and the others coming from completely free users who generate 10x+ less revenue.

        So seeing as all the other includings in CSPC have frozen percentages/ratios despite sometimes drastic changes in prices/revenue, why not create one for features? Sure there is features that are 15 seconds, and some that are 4 minutes in vocals but fully giving them 100% of sales/streams is just inflation for those that commonly do features. I think features should definitely be included when measuring careers, but not to the same extent as their own material in any case.

        1. Hi again Anon!

          I have to contradict the end of your message that states features are counted as their own material, they aren’t! In fact an artist featuring on an other song has 100% of its streams/downloads, but he gets 0% of the album sales generated from that! For example, Love The Way You Lie isn’t among Rihanna’s top 10 hits, because stand-alone streams/downloads aren’t enough to made up for the fact Rihanna was granted 0% of the album Recovery. That’s how this song is worth a total of 4,65m EAS for Eminem while only 1,8m EAS for Rihanna (both figures are with old ratios but it doesn’t change the idea).

          As you can see, the “inflation” is limited with a mere 1,8m for a strong featuring on the biggest global hit of 2010. In terms of function, it also makes sense:
          – if a consumer is interested in Eminem, he buys Recovery and Rihanna is awarded 0 sale. It’s natural as the person is going after Eminem only.
          – if a consumer is interested in the song only, not especially on Eminem, he buys the song and units go as much to Eminem than to Rihanna. It’s natural as among buyers there will be as many casual listeners of Rihanna than Eminem.

          True big hits like Yeah! by Usher generated a lot of album sales, so someone like Lil Jon gets only a tiny fraction of the song total units.

          Then there is the case of very recent releases. As the industry continues to evolve, album sales are disappearing while featurings are always more numerous, with now not only duets but real all-stars hits coming out like Feels or I’m The One. Those songs aren’t personalized though, they are party tunes, people do not really care about who’s the credited author of it. The urban community too doesn’t care about the leading act on a Drake / Rihanna / Nicki / Kanye song, they playlist all big R&B/Rap hits and that’s it. After all, if nobody buys albums of DJ Khaled and Calvin Harris and most of their streams are acheived from the pages of Justin Bieber and Katy Perry, why would they be granted a bigger rate? Plus, those streaming-only hits do not create that much value. It takes nearly 1 billion Spotify streams to account for 1 million album sales, something Adele achieves by clapping her hands only!

          One last point, by “frozen ratio” it wasn’t about the price/setting ratios, clearly I have no issues with that. The problem comes with categorizing features. It would require to create a rank of features, duets-like (Love The Way You Like, 4 Minutes etc), grouped tunes (Feel, I’m The One), average participation (say Jay-Z on Beyonce’s songs) and minor featurings (like Janelle on We Are Young). This implies to listen to every song something which is impossible to apply. The album factor does that perfectly, as the least shared a hit is, the most album sales it will generate, the more units will be assigned to the lead singer rather than the featured act.

          If all-star songs continue to come out at fast pace, I may think about introducing a ratio on songs involving 3 or more acts. Something like:
          – 1 featured act = 100% of albums generated to lead only, 100% of songs-related units to both the lead and the featured artist
          – 2 featured acts = 100% of albums generated to lead only (they would be most likely low), 70% of songs-related units to each artist
          – 3 featured acts = 100% of albums generated to lead only (they would be most likely low), 45% of songs-related units to each artist
          – 4 or more featured acts = 100% of albums generated to lead only (they would be most likely low), 25% of songs-related units to each artist

          1. Thanks for replying again.

            I think the proposed ratio is a great idea. Features are the new norm so I definitely do think you will have to use a ratio for simple fairness in the future.

    1. Hi Tom!

      Spotify counts only streams on which the artist is the main credit and also often excludes all remix versions, which explains the gap between their lists on those published on our site 🙂

    1. Hi Harry,

      This is a very wrong reasoning – his album sales are not spectacular in comparison to 90s artists because of streaming, they cannibalize each other, you can’t comment both sales avenues as if they were unrelated. If there was no streaming, he would sell an awful lot of CDs.

Leave a Reply

Your email address will not be published. Required fields are marked *