Collaborating with a superstar? Don't hold your breath for a big break.

By Peter Sinclair, CEO & Co-Founder at beatBread

A few weeks ago, Sam Smith and Kim Petras won a Grammy for Best Pop Duo/Group Performance for their smash single "Unholy." At beatBread, we've been interested in collaborations like this for a while now, because of persistent requests we receive from artists interested to bring collaborators onto their next project. 

We use data to generate predictions of artists' future streaming performance from their catalog and for unreleased music too. We put money behind those predictions when we fund artists, so we need to be smart about every different set of facts we see. Several times a month, artists seeking funding at beatBread will show us a list of star and even superstar collaborators joining their next project with hope that these collaborators will impact the amount of funding they can get from our algorithm. Artists with significant collaborators lined up often tell our team they believe the collaboration will not only create a sure hit, but will help them grow their fanbase and reach the next level in their careers. 

So what do we do when we encounter this type of situation? It makes sense to think an artist who will have one, two, or even more star collaborators on their next project is about to jump on a career rocketship. After all, there are lots of examples of stars who seemingly had their ‘big debut’ in the form of a collaboration (whether that be Kim Petras, David Guetta, or countless others) 

Kim Petras never cracked 4M monthly listeners on Spotify before the release of Unholy, and her monthly listeners have increased 10x since the song was released. Is this a model for other artists to emulate? Is her experience the outlier or the rule? And if so, just how much is a superstar collaborator worth?  

This question of the precise value of a collaboration is obviously something that interests us because we fund advances based on data, not anecdotes or gut feeling. If we are wrong, it costs our business dearly. But it’s a question that is crucial for artists themselves - even for those who will never get funding from us. Developing artists sometimes are asked to pay a star collaborator to join a track (in cash or in the split), so they should know how to think about what the collaboration might be worth, and the chances that the star will help generate a hit. And major labels often dangle the possibility of collaborations with artists on their roster to induce developing artists to sign. How much stock should a developing artist put in collaboration, given all the value they are asked to give up in exchange? 

Our data science team, led by my co-founder John Haller and his colleague Travis Andersen, have dug deep into the impact of collaborations over the past year. In light of Kim Petras' big win this year, we wanted to share what we found with artists who are looking for collaborations to power their big break and give them a healthy word of caution. Sometimes, what seems like a “big break” opportunity may be a mirage that doesn’t materialize, and often, if that big break does materialize, the data suggests that that success might have happened anyway, without the help of the star collaborator. 

But let’s not get ahead of ourselves, here’s what we did: First, we started with a large dataset containing over 14 million tracks and tens of thousands of artists, each with several years of historical data.  The artists in our data set were from small artists with as few as 10,000 monthly listeners on Spotify, prominent artists with over 10 million monthly listeners, and a lot of artists in between. (By the way, we do look beyond Spotify when we analyze artist data below and in our core business, but Spotify is a convenient screen to define the data set.)  

Within this group of artists, we looked at all the new releases over time and created a metric to gauge any new track's performance. We call it the "release ratio."

The release ratio measures how much better or worse a new song performs in number of total streams than that artist's median first-year performance of previously released songs.  

If the release ratio is bigger than 1, that means a track performed better than the median of that artist's past releases. If the release ratio is smaller than 1, the song performed worse than the artist's median release of the past. A release ratio of 10 would mean that a new song did 10x more streams in its first year than that artist's median release in the past. 

We then compared four types of release:  

1. tracks with no featured artist  
2. tracks where the featured artist(s) are about the same size as the "main artist,"  
3. tracks where the featured artist(s) are significantly smaller than the main artist, and  
4. tracks where the collaborator is very large relative to the main artist and looked at the range of "release ratios" in each case.  

Here's what the numbers show. 

First, it is worth noting "breaking out" is really hard. Working artists know this all too well, but it’s worth pointing out nonetheless. Whether there is a collaborator on a new song or not, it's rare for a new song from an artist to do significantly better than any previous songs released by that artist.  

Only about 1 in 5 releases without a collaborator out-stream an artist's "median" prior releases by more than 3x in the first year of release. An actual "breakout" song, e.g., a career changer, which for most artists would mean a track that performs at 10x, 100x, or even more than the median of past releases, are even rarer. When there is no collaborator, only about 8% (less than 1 in 12) of new songs achieve a release ratio over 10, and only 1.2% (~1 in 85) achieve a release ratio of 100 or more.  

But surely adding a more significant artist with a bigger fanbase changes the game in a big way, right?  

Yes…and no.  

The odds of a new release doing better changes some but not a whole lot when a larger collaborator joins the project. If an artist can get a guest feature from an artist with a larger number of monthly listeners than the main artist, the new track only has a 30% (about 1 in 3) chance of generating a 3x increase in streams from the median. That's only a tiny improvement from the no collaboration case (20% to 30%).  

The story's almost the same for bigger jumps in the release ratio. Adding a larger collaborator (one who is 3-5x larger than the main artist) doubles the chance of a new song achieving a release ratio of 10 or more, but the odds of getting there are still really low, at only around 15%. And while the odds of generating a release ratio of 100+, a true breakout hit, improve by nearly 3x, achieving that level of performance with a larger collaborator is still a longshot, only 3.4%. 

When a "real star" is added into the mix, the odds of breaking out get better, but are still much lower than one would think. When a featured artist has at least 10M monthly listeners on Spotify and that artist is more than 10 times larger than the main artist releasing a new song, the odds of achieving a release ratio of 3 or more go from 20% (no collaborator) to 50% (star collaborator). The chances of achieving a release ratio of 10 increase from 8% (no collaborator) to 30% (star collaborator). Finally, the odds of a real "run-away" hit, e.g., a release ratio of 100 or more, go from 1% to 9.1%.  

I’ll be honest, we were surprised that the impact of a much larger artist is so low. You'd think that if you are an artist with 10,000 monthly listeners, adding a featured artist with 1,000,000 monthly listeners (or if you have 100,000 monthly listeners, adding an artist with 10M monthly listeners to your record) to your next song would be a sure bet-- you'd benefit from the exposure to that larger artists' fans. It turns out that data science shows us that the impact of a major collaborator is not zero, but it’s a lot less than one might expect, maybe because we only remember the successes, and not the failures when we think about songs. When a much bigger artist joins the track, the odds of having a song that does materially better than your last releases are still less than 50/50. The odds of getting a breakout of 100x higher than a past release are less than 10% when an artist with a following at least ten times bigger than you do joins your project.  

We also looked at overall career impact, to see how additional releases from an artist who participated in a collaboration performed in the following year and several years after that.  And here’s where the story told by the data is even more surprising, and more profound. In the year after a collaboration between a small artist and a larger artist (controlling for all of the other proprietary variables we use to predict an artist's performance), the impact that a large artist collaboration has on smaller artists' career is nearly zero, and if anything, it's negative. This is true when the smaller artist lands a bigger artist to join as a collaborator on one of their own songs, and also when the small artist joins the larger artist's track as a featured performer.    

It isn't until four years later that we see a positive relationship between a past collaboration and an artist's career growth. Because the effect takes so long to develop, we think that artists who can convince larger artists to collaborate with them have other characteristics that lead to success over the long run, and possibly, that artists learn from each other when they collaborate in ways that take a while to shine through. In any case, this longer-term impact is relatively small, and the notion that landing a big collaborator will make your career just isn't backed by the data. 

Wait a second, what about Kim Petras? How does her experience jibe with all of the analysis above? Doesn't her recent success call all of this analysis into question? 

We don't think so. As strong as the performance of "Unholy" may be, it is not evident that the releases Petras made after her collab with Sam Smith are doing materially better than some of the tracks she released before Unholy. "brrr" and "If Jesus was a Rockstar," both released after "Unholy," do not appear to be streaming at levels all that much different than we might expect even if she hadn't collaborated with Sam Smith at all. On the other hand, "Treat me Like a Slut”, released recently, does appear to be doing 2x or 3x better (but not 10x or 100x) in its early days than we might have predicted. We'll see how Petras's newest album performs in the coming weeks. And to be fair, we aren’t saying that superstar collaborations never make a difference in an artists long term career, just that they don’t make a difference more often than a whole lot of other factors that might drive success for an artist.   Whatever the case, we wish Kim all the best.   

So what to make of all this? It would be overstating the case to suggest that artists should not seek collaborations with other artists or that there might not be a case for hitching your wagon to a superstar if you are a developing artist. In many cases, beatBread will advance more money for a song or group of songs when a major artist is collaborating because, even though there are no guarantees that the collaboration will create a hit, it certainly improves the chances in a game that is really hard to win. Perhaps most importantly for artists, working with the right collaborator might be the best way to make great music.  

Nonetheless, we think it is essential for aspiring artists to realize that working with a more successful artist will likely have much less impact than they might hope.  And we think that the data clearly shows that while a superstar may be entitled to a greater split for joining a given song, an artist should never give a share of their other music or future career in exchange for a collaboration. In the end, there is no substitute for an artist building their own fan base, and relying on their own efforts to build a career.  

A few days ago, the ever-insightful Dan Runcie, who writes the Trapital newsletter, posted about the best way for artists to leverage TikTok. It turns out that artists’ best strategy on the platform is to create original content rather than riding the latest meme or participating in the latest dance challenge, as enticing as chasing the hot trend may be. We think that the findings of this beatBread analysis rhymes with that astute insight from Trapital. There are no short cuts to success, and the best way for artists to grow is to focus on their own great work rather than pursuing a strategy than relies on the halo of another.   

Did you find this interesting? If you're an artist or manager looking to finance your next project, have a look at beatBread.com. If you're a data scientist who wants to do work like this or think we missed something, drop us a line at company@beatbread.com.