Browse the internet long enough and you’ll eventually run across Hunter S. Thompson’s meme about the music industry. The meme is actually a misquote, but it’s still a fair representation of what the music business is like. Or rather, what it used to be like 50 years ago. Today, things are different. ‘Thieves and pimps’ no longer run free. Instead, ‘corporate execs and lawyers’ run the show. The suits do just fine. But artists die like dogs.

It’s another victory for oligopoly.

In the 21st-century music industry, the vast majority of revenue is controlled by a few corporations. In 2018, the Big Three record companies (Sony, Universal, and Warner) raked in 68% of all music sales. The situation is nearly identical for music streaming. In 2019, the Big Three music streamers (Spotify, Apple, and Amazon) claimed 69% of all streaming revenue.

Because big firms can use their power to hike prices, we usually think of oligopolies as being bad for consumers. But oligopoly doesn’t always pan out this way. Sometimes oligopolists use their power in the reverse direction. Instead of hiking prices, they cut costs. How? By dictating prices to sellers.

Think of this cost-cutting strategy as the ‘Walmart model’. Yes, Walmart’s slogan is ‘always low prices’. But its actual business model is ‘always cut costs’. To achieve this cost cutting, Walmart uses its power to force sellers to lower their prices. That’s good for Walmart, but bad for sellers.²

The modern music industry runs on the Walmart model. Streaming services like Spotify charge low prices to listeners. But under the hood, the streaming business model is ‘always cut costs’. That means paying artists as little as possible — about 0.2 cents per stream. To put this rate in context, an artist with 1 million streams would earn $2000 — hardly enough to live on. For most musicians, then, Spotify is basically a marketing tool that pays a little on the side. Their real income (before COVID) comes from playing live.

This post, however, isn’t about the vast army of struggling musicians. (I was once one of them.³) It’s about artists at the top. Even at the pinnacle of the music business, competition is fierce. Hit songs have a brief window in which they are immensely popular. After that, they ‘die like dogs’.

To get a sense for this popularity window, I’m going to measure the ‘half life’ of Spotify hits. This is the time a song spends above 50% of its peak popularity. Like radioactive plutonium, hit songs decay quickly. Most artists have about 2 months to make dime on their song before it fades into obscurity.

Spotify hits of 2020

Let’s start the journey into Spotify hits by jogging our memory of 2020. Yes, it was a shit year … but the music was pretty good. On Spotify, 17 songs reached number one. Figure 1 shows the rise and fall of each of these songs, measured in terms of daily streams.

The chart above is admittedly messy, so let’s find a tidier way of looking at the rise and fall of Spotify hits. To better see the trends, it’s helpful to center each song around the date that its streams peaked. Like a rocket launch, we treat this peak as t = 0. Time before the streaming peak is negative (as in t =  − 5 days). Time after the peak is positive (as in t =  + 5 days.)

For comparison, it’s also helpful to plot daily streams not in raw numbers, but as a percentage of peak streams. When we plot the data this way, the location of the streaming peak (on the coordinate grid) is always the same.

Figure 2 shows what the Spotify hits of 2020 look like when plotted on this revised scale. The red dashed line is the date of peak daily streams, defined as t = 0. The blue curve shows daily streams, expressed as a portion of peak streams. If you let the animation play, you’ll see the various ‘shapes’ of each hit song. Some songs rise and fall quickly. Others stay popular for a long time. (The short-term jitter in daily streams is caused by weekly listening habits, which tend to peak on the weekend and fall during weekdays.)

If you want to hear these hit songs, watch the video below. (I couldn’t include Billie Eilish’s song my future because it was blocked by copyright … even though I excerpted a mere 20 seconds of it. On that note, see Cory Doctorow’s rant about YouTube’s copyright filters: How copyright filters lead to wage-theft.)

It’s not surprising that different songs have different patterns of popularity. What is surprising, though, is that if we look at the popularity trend for every Spotify hit, a clear pattern emerges.

On that front, check out Figure 3. Here I’ve taken Spotify hits and plotted their daily streams in the same way as above. (So a song’s peak happens at t = 0.) But instead of looking at each hit individually, I’ve looked at all hits simultaneously. Figure 3 shows the popularity trend for every song that made it into the Spotify Top 200 since 2017. (That’s about 5500 songs.) When we look at the popularity trend across all these songs, we get the tent shape shown below.

Spotify hits, it seems, burn fast and bright. They rise to their peak popularity in a month or so, and then fall from grace over another couple of months. And it’s not like they get to bask in glory at the top. Peak popularity lasts no more than a few days. Yes, a few songs have staying power. But most Spotify hits live fast and die young.

A song’s half life

To quantify a song’s rise and fall, I’ll borrow an idea from chemistry. Chemists often measure radioactivity in terms of an isotope’s half life — the time it takes for half of the radioactive substance to decay. Let’s apply this measure to hit songs.

I’ll define a song’s half life (on Spotify) as the time it spends continuously above 50% of its peak popularity. Figure 4 shows this half life for three songs:

1. Eminem’s The Ringer
2. Drake’s Nice For What
3. The Weeknd’s Blinding Lights

We can use these three songs to classify different types of hits. Let’s start with Eminem’s The Ringer. The song peaked on its first day on the charts, and then plummeted in popularity thereafter. Its half life was a mere 8 days. We’ll call this type of short-lived hit a fast burner. Its popularity is burnt up in a flash.

Drake’s hit Nice for What lasted longer, with a half life of 56 days. We’ll call this type of hit a medium burner. Its popularity is neither ephemeral nor long-lasting.

Finally, we have The Weeknd’s hit Blinding Lights. The song took 2 months to reach its peak and another two months to fall. Overall, its half life was 159 days. We’ll call this type of long-lasting hit a slow burner.

Because the payout lasts, every artist wants a slow-burner hit. Yet few musicians achieve this feat. Most hits burn briefly. The median half life of a Spotify Top 200 hit is 59 days — just short of 2 months. Moreover, the ‘half-life distribution’ is highly skewed. As Figure 5 shows, most hit songs have a brief half life, while a lucky few are slow burners.

To get a sense for just how hard it is to make a slow-burner hit, look at the pie chart in Figure 6. Here I’ve grouped hits according to my ‘burn’ categories (using arbitrary cutoffs for a song’s half life). Only 10% of Spotify hits are slow burners, with half lives more than 150 days. A whopping 46% of hit songs have a half life less than 50 days. Think about that. Most artists with a hit song have less than 2 months to make their money before the song falls out of favor.

Half-life income

Next question. If you have a Spotify Top 200 hit, how much can you expect to earn during the song’s half life?

The answer depends on your royalty agreement, which varies by country and by artist. But for example purposes, let’s assume a royalty of 0.2 cents per stream (a typical payout).

If every Spotify artist received this royalty (since 2017), then the median payout during the half life of a Top 200 hit would be about $160,000. But as Figure 7 shows, the payout distribution is far from equal. The bottom fifth of Top 200 hits earn less than $45,000 during their half life. Yet the top 2% of hit songs pay out more than $1 million.

Yes, the payout distribution for hit songs is unequal. Still, the median half-life income seems good enough. Heck, if I earned $160,000 each year I’d be gleeful. The problem, though, is that the payout for a hit song is a one-time deal. So if you want income that lasts, you need to pump out hit after hit.

Unsurprisingly, most artists are not hit-making machines. They’re one-hit wonders. To see this fact, look at Figure 8. I’ve plotted here the number of Spotify Top 200 hits per artist who’s had a hit since 2017. The overwhelming majority of hit-makers have only a few hit tracks. Nearly half of hit-makers (49.7%) have only one hit. And 75% have fewer than 5 hits.

Now we see the hard reality of the music-streaming business. The vast majority of artists will never make a living on Spotify.

Let’s do the math. For starters, the likelihood of having a Top 200 hit is exceedingly small. Spotify is currently home to about 70 million songs. Since 2017, about 5500 of these songs have reached the Top 200. That puts the probability of your song reaching the Top 200 at about 1 in 10,000.

If you happen to beat these odds, you’ll probably do it once. You’ll take your $100K payout, and then wonder how to make money for the rest of your life. Such is the life of a musician.⁴

The Spotify production function

Most musicians do what they do because they love the craftsmanship of their art — the thrill of playing or composing great music. For them, making money is an annoying necessity. It must be done, but is never the main goal. That’s good, because if you get into music for the money, you’ll likely be disappointed.

The corporate side of the music industry, though, doesn’t run on craftsmanship. It runs on mass production. If you want to generate income from music streaming, the surest way to do so is to become a hit-making machine. Figure 9 tells the story. On Spotify, artists’ total number of streams increases reliably as they generate more hits. Yes, there are rare artists who top one billion streams with a single hit. But they are outliers. If you want to guarantee billions of streams, you need dozens of hits.

Economists call this type of income-generating formula a ‘production function’. Let’s have a look at the Spotify production function in a more abstract form. Instead of looking at individual artists, let’s look at the aggregate trend, as shown in Figure 10. Here you can see the production function in action. The blue line shows the number of streams you can expect for the given number of hits. The shaded region shows the uncertainty — the region within which you have a 50% chance of falling. By removing the noise, the Spotify production function becomes crystal clear. To reliably get more streams, you must pump out more hits.

Next question: how do you generate hit songs? Well, it obviously takes talent. But more than that, it takes a team of labor. Behind the scenes of most hit singles is a panoply of songwriters, producers, editors, arrangers and backing musicians. In other words, if you think Drake pumped out 94 Spotify hits by himself, you’re sadly mistaken. He’s just the public face of a hit-making production function.

The biggest production function, though, is not the one for individual artists. It’s the business model of music-streaming oligopolies themselves. The streaming model is built on the steady supply of songs, each with a brief half life by which artists live and die. But the music-streaming oligopolies sit above the fray, confident that there will always be a new hit to replace the last one.

To artists, the streaming oligopolies sell a dream. But you’d have to be asleep to believe it.⁵

---

Support this blog

Economics from the Top Down is where I share my ideas for how to create a better economics. If you liked this post, consider becoming a patron. You’ll help me continue my research, and continue to share it with readers like you.

---

Stay updated

Sign up to get email updates from this blog.

---

This work is licensed under a Creative Commons Attribution 4.0 License. You can use/share it anyway you want, provided you attribute it to me (Blair Fix) and link to Economics from the Top Down.

---

Sources and Methods

I’ve used Spotify data for global daily streams. Note: Spotify only lets you download data for one day at a time, which is annoying if you want the whole database. They also put up a CAPTCHA to make it difficult to scrape the site. Still, there are ways to automate the download. (I won’t reveal them here, in case prying eyes from Spotify are reading. Contact me if you want my code.)

You can download my dataset here.

I’ve excluded Christmas songs from my analysis, since they rise and fall seasonally, rather than just once.

To calculate songs’ half life, I’ve excluded tracks whose streams have not yet declined below 50% of their peak, and I use only the tracks that stayed in the Top 200 for 30 days or more.

Notes

Further reading