The Numbers Dictate the Next Move

The Numbers Dictate the Next Move

By Michael Raine

If you’re an avid sports fan like I am – the type who turns on sports radio in the car and podcasts when doing chores, and reads detailed analyses of how players are performing – analytics has become just part of the conversation. Hockey fans throw around terms like “Corsi” or “shooting percentage,” or baseball conversations commonly use the acronym WAR (wins above replacement — google it). The point being, in sports, even casual fans understand that analytics are influencing the decisions their favourite teams make, even if the fans often don’t understand exactly how analytics work. For music fans, however, where the sheer amount of data available in the digital age is nearly infinite, it’s less understood how analytics are influencing what we hear, from how a song is marketed to where a band tours, or whether they get a record deal. And yet, there are many decisions, both big and small, that are influenced by data.

The idea that music is subjective is a prevalent one, and somewhat true, obviously. But there is the still-common belief that the great decision makers have some well-refined “gut instinct” for these things — that their eyes and ears can identify an artist’s potential and that elusive X factor that others don’t see. It’s perpetuated by Netflix documentaries about near-mythical industry figures like David Geffen and countless books and magazine profiles. But now more than ever, the truth is, data rules. Data, though, is only as useful as one’s ability to analyze and apply it (i.e. analytics), and there is a lot of data out there.

“Basically, it’s really based on numbers — streaming numbers, radio performance, and all that sort of thing. So, we always let the numbers and the data sort of dictate the next move,” Scott Cooke told me last September when chatting for the Canadian Musician Podcast. Cooke is a chart-topping engineer and producer and now the cofounder of Canadian country indie label Local Hay Records, which is a joint venture with Nashville-based Big Loud Records. Cooke and I were talking about Local Hay’s first signing, Shawn Austin, and the blueprint for his career. I’d been thinking and reading more about the use of analytics in the music industry, so Cooke’s comments stuck with me for months after.

“They have all these interesting predictors to where you can release an EP and you can look at certain ratios of streams to saves and listens and all these things, and go, ‘You know, even two or three weeks later, this song is reacting in a certain way where it’s showing that it’s connecting with people, so this could be a potential hit song.’ Or, ‘We love this song, but it’s just not reacting in the way we thought it was going to react, so let’s release another song in a month or two and see if that gets a reaction,’” Cooke continued. “So, it’s very fluid, and I find you have an idea in your head of how you want to roll it out. But I mean, numbers don’t lie.”

As we ended that chat, I mentioned to Cooke that I’d been thinking of writing an article on streaming analytics and how it’s influencing decisions in the music industry (only took me seven months to get around to it!). The guy I needed to connect with, he said, is Patch Culbertson, the current senior vice president at Big Loud Records (Morgan Wallen, Dallas Smith, MacKenzie Porter), who Cooke said was one of the smartest people in the industry on this topic — and he wasn’t lying. Prior to joining Big Loud in 2017 as its VP of A&R, Culbertson was a talent scout and director of A&R at Republic Records, which had a reputation as the most data-driven major label and “the gold standard for using analytics in scouting and marketing,” as noted in a widely-read 2014 article in The Atlantic called “The Shazam Effect,” which featured Culbertson prominently. (That Atlantic journalist, Derek Thompson, later wrote the book Hit Makers, which also featured Culbertson’s music analytics insights.)


“Discovering actionable insights from streaming data has been crucial to my success as both an A&R and label executive,” Culbertson later tells Canadian Musician. “When I started my career in 2009, the streaming landscape included platforms like MySpace, YouTube, Last.FM, and Pure Volume. Over time, platforms came and went, but the fundamentals in evaluating hit records remained the same. One’s ability to ‘read records’ through consumption data aids both in the signing process (for research-led signings) and single determination of a roster artist’s released music.”

As far as what labels are looking for in the ocean of data they collect from the DSPs (e.g. Spotify, Apple Music, etc.), Culbertson is careful to guard his secrets, saying he “can’t give away the special sauce,” but notes two fundamental principles. The first, he says, is: “Does exposure lead to consumption?” And the second: “Consumption rises with the tide of a hit.”

Expanding on those two thoughts, Culbertson continues, “It is important to remember that in our now-ubiquitous medium of streaming, exposure of a record is often misunderstood to be defined as genuine consumption. If you consider a playlist like you would radio, we didn’t define ‘record sales’ as radio audience plus CDs plus downloads in the 2000s,” he explains. “But today, it’s a grey area when we define ‘consumption’ as music that may be pushed to a listener via programmed playlists and algorithmic means in addition to the genuine consumption of a song or album (double platinum is the new platinum these days for that reason). Are people consuming and engaging with the record and artist beyond the exposure the song is receiving? With regards to number two, hit records most often exhibit consistent growth, sometimes exponential growth, everywhere it exists with respect to the audience of that platform.”

Using an apt baseball analogy, Culbertson says once a single is released, the streaming metrics help categorize that song as a single, double, triple, home run, or the rare grand slam. “Each base carries with it certain marketing and promotion strategies in relation to its consumption strength,” he explains. “If we see those implemented strategies continue to move the needle, it’s time to pour gas on the fire. If there’s an identifiable ceiling in the record’s growth following those campaign strategies, we identify how we can authentically support the record recognizing its level of reactivity while shifting gears to a new focus. It’s simply a disservice to the artist, fans, our team, and our partners to continue pushing a non-reactive record to our format’s audience.”

Speaking of data being used to push a song to the audience based on format, region, genre, etc., radio promoters and programmers are looking at the numbers, too. “I don’t actually have access to that data, but it’s interesting that I just had that conversation this morning, talking about a certain song that a station added because they had seen stats on how often it had been Shazamed in their market,” says Toronto-based radio promoter Andrea Morris, the founder of AM to FM Promotions. “And that’s another thing, too, sometimes with streaming numbers, you’ll approach radio with an artist who has had, say, 15,000 streams, but it’s about, where are they coming from? Then they look at the breakdown of the data and go, ‘Well, there’s none of that in my area, these are all streams in Europe and that doesn’t actually affect us.’”

On the A&R front, how analytics inform decisions about which artists to sign (or not sign) can differ widely across labels. “Some choose to mitigate the risks in the signing process by evaluating consumption metrics of an unsigned artist or band’s repertoire and offering recording contracts at what they deem as actionable checkpoints of their campaign,” says Culbertson. “Once signed, streaming data – whether on DSPs or other non-commercial platforms – is a real-time feedback loop from consumers (and hopefully fans) on release strategy/configurations, focus singles, content strategies, and promotion.”

That said, some A&R executives remain firm believers that, while the data can inform marketing and other decisions to do with established acts, when it comes to spotting unsigned artists and assessing potential, the data can’t predict the future. This isn’t sports, where the purpose of analytics is not only to judge past performance, but to use trends in the data to predict future performance. Music is far too subjective for that, at least according to some.

This topic came up last year when I was interviewing rap pioneer and Universal Music Canada’s Senior Vice President of A&R, Kardinal Offishall, for a Canadian Musician cover story. “We live in a time with music where a lot of the industry is heavy on analytics — less on gut, a lot more on analytics. I think it’s interesting to me because relying on analytics means you’re relying on the past, because analytics are something that happened already. Analytics are not something that can tell you the future,” he told me. “I know a lot of people don’t use that type of language and don’t really consider it that, but when you look at it at its core, that’s what it is. You’re seeing a trend, or part of a trend, that has passed already. So, now it’s up to you to guess — is that something that’s going to be sustainable, or is it going to fizzle next week or next month?”

Knowing Offishall was a basketball and baseball fan, I pointed out the sports comparison; that sports analytics use trends found in the past to predict the future, but he was quick to dismiss the comparison. “Nah,” he said, “When you think about the breakdown of it, it’s like, ‘Okay, if 1,000 guys go to that one basket, this is usually what happens.’ When you go through the analytics, break it down to get field goal percentages, three-point percentages, whatever, here’s what’s most likely to happen. But you have hundreds of guys shooting at the same single goal. In music, the equivalent would be 100 different nets, with a million different people shooting at the same time, from a million different places, in a million different courts, and trying to gather analytics from that and then make that apply to a singular artist!”

Instead, Offishall told me, when he looks at the analytics, it’s about the bigger picture it paints. “I look at the analytics and it shows trends and I’m like, ‘Alright, cool.’ It’s the bigger picture. And I’ve heard it somewhere that analytics are kind of like GPS — like terrible GPS. We’ve all had it on our on our phones or whatever when you’re going somewhere and it says ‘turn right here’ and you’re like, ‘Why would I do that? I know if I go straight, it’s going to be five minutes faster.’ But you follow the GPS anyway and then you’re like, ‘Shit, I should have just listened to my gut.’ So, analytics is kind of like the GPS where it’ll get you there, but if you’ve gone down that road a thousand times, use the GPS for what it is.”


So, if analytics for A&R is more opaque, what other uses in the music business are there for all this data, and where is it all coming from anyway? I checked with the data analysis team at Chartmetric, one of the leading streaming and social data analytics companies serving major music companies and artists. Its main service is an all-in-one dashboard made available to clients in the music industry to collect, combine, and analyze data from more nearly 30 sources, from social media companies, the major DSPs like Spotify and Apple Music, to more regional DSPs like Boomplay in Africa, Line Music in Japan, and Melon in South Korea.

“If you’re a paying client, everyone can see the same thing, effectively,” says Rutger Ansley Rosenborg, digital strategy lead at Chartmetric. “That said, we do have some special arrangements with some clients who maybe need to see more from the data, or they need to see the data in a certain way, or they don’t want to see the data that isn’t relevant to them. So, if they’re tracking a particular roster, but they have hundreds of artists and they can’t do that through the tool, then we can build something custom for them. Or, sometimes they just want reports generated, which we can also do. Sometimes they have their own data analysis team and we just drop some CSVs to them. So, it runs the gamut.”


I ask if the companies getting the bespoke service are the major labels and Rosenborg stays coy, just saying, “I guess I’ll put it this way — they are higher-value clients.”

Of course, a lot of data on an individual level is now made available to artists through the DSPs’ artist portals (Spotify/Apple Music/Amazon Music for Artists), their digital distributor dashboard, or other Chartmetric-type services like Soundcharts. It’s a fantastic development that indie artists can now access their own data, which they never could in the past. But something like Chartmetric aggregates all the data and makes sense of it to inform decisions at a company/industry level.

“In my talks with different clients, from the big indies to the major labels, it frequently revolves around playlists, but I think that conversation is kind of changing as we speak. I think a lot of them are getting past the playlist world and starting to think more about things like social engagement and how to better measure those things,” adds Jason Joven, the music data insights manager at Chartmetric. “So, what a lot of those teams are trying to measure is something like, ‘Okay, artist X is always on the top editorial playlists on Spotify, but there’s a lot of ground to gain still in their social engagement despite the enormous number of followers that they have.’ I think that’s an interesting question that they’re starting to bat around a lot.”

At Culbertson noted, artists, labels and others are wary of playlist-driven data, because passive listening doesn’t create a fanbase who pays for merch and tickets. That said, Joven notes something that Chartmetric terms “Playlist Journeys,” which uses the data to analyse the relationship between playlists and how a song moves between them. “That is something I think an indie artist could really use, because it’s a level of analysis that’s just not really possible unless you’re a data scientist yourself,” says Joven.

In terms of how Chartmetric’s analytics tools are used to inform marketing, Joven adds: “For a lot of the people on the marketing side of things – and not even just a digital marketing team at a label, but sometimes they’re the creative agencies that do the bands-and-brands thing –they’re super keyed in on demographics. And so, we have a pretty good body of demographics that are based out of Instagram followers, TikTok followers, or YouTube subscribers, and based on age, gender, location, languages spoken, and what other brand affinities they have. That’s really their world. And so, for them to be able to cross reference that type of demographic information with emerging artists that they’re trying to place with a certain campaign they’re working with, I think that’s another use case, too.”

Ultimately, there’s no one number or metric that matters most and tells a complete story about growth or popularity (essentially what WAR is in baseball analytics). For example, Shazam data is a useful indicator about a song’s ability to pique curiosity, but you need to see if that’s leading to listens over on the streaming services.

“Shazam is an incredibly valuable tool in our research toolkit, but should be thought as a secondary indicator,” adds Culbertson. “The primary indicator, and ultimate goal of promotion, is consumption following that exposure. Shazam is an indicator of that lean-forward curiosity, but true hit records pull you through that interface to add/download the song to your personal library. I also appreciate [Shazam’s] geographic and genre charts for context.”

That is a good example of why Chartmetric prioritizes seeing crossover and steady growth across the board. It’s that relationship between data sources that paints a fuller picture for decision makers of any kind.


“The important thing, generally, is not necessarily one platform, one metric, but the relationship between those,” adds Rosenborg. “So, for example, are my Spotify followers growing at the same time that my Spotify monthly listeners are growing? Because if only my monthly listeners are growing but my Spotify followers are staying stagnant, then I’m not converting actual fans. It’s just momentary passive listening because maybe I got on a huge playlist and then no one hears from me again,” explains Rosenborg. “So, the relationship between platforms is super important. What is the crossover that’s happening? Is TikTok crossing over to Spotify eventually, or is it just remaining on TikTok? If it remains on TikTok, it’s just going to dissipate really quickly. So, it’s all about the conversions that are happening in the relationship between various metrics and making sure that everything is growing — not necessarily in an exponential way, but in a healthy, steady way across the board. I think, in my opinion, that’s the most important thing.”

So, we’ve seen how labels, radio, and marketers across the music industry are reading and interrupting data to inform decisions, but what about those dealing in music rights, such as performing rights organizations (PROs)? They play a massive role in the industry, not least of which is getting artists and publishers paid for the use of their music. And importantly, music rights companies may have to deal with more data than anybody else.

“We have billions of transactions — the amount of data that we have in SOCAN is unbelievable,” Alec McGlaughlin tells me. He is the manager of data analytics at SOCAN, the PRO representing Canadian songwriters, composers, and publishers. Under the guidance of Alan Triger, who was hired last year to lead the new Strategic Solutions and Analytics department, SOCAN is in the process of standardizing and centralizing its data collection and analyses to better inform its business decisions and advocacy work, and to provide more useful individual data to its members.


“If we’re proposing changes to our distribution rules, and let’s say we’re going to modernize to try and keep up with the changes of how the music industry has developed, we have to keep our distribution rules up to date, right? So, how are we distributing royalties when we get the licensing fee and the DSP data in?” begins McGlaughlin. “And so, from an analytics front, we don’t want to essentially do the equivalent of throw grass up in the air and say, ‘We’re going to change this and this’ and then go and institute it. We’re really looking at modeling out what those different scenarios are for changes to those distribution rules.”

One of SOCAN’s primary tasks is to find and license any business that should be paying performing rights fees or royalties, such as any business with live music. “I know on the general licensing side, we’re constantly looking for different data sources on licensed businesses that are out there. So, whether it be online events that we can collect data from, online platforms that say ‘here’s concerts that are happening here and there,’ to even open-data sources,” explains McGlaughlin. “For example, the City of Toronto has an open-data source for all the licensed businesses that are part of city. So, it’s about how do we collect that data, bring it into our environment, merge it with our data assets, and provide value to our licensing agents to say, ‘Okay, here’s a list of qualified businesses to check in on’ and that kind of stuff. So, we have done it in the past, that type of analysis for digital licensing and general licensing.”

Music rights organizations like SOCAN also do their own A&R, though unlike labels where A&R is about finding, signing, and developing promising artists before another label does, SOCAN is looking for unknown Canadian songwriters and getting them in its membership before their songs takeoff and uncollected royalties accumulate.

“As much as possible, we have a lot of data that comes from the DSPs, in terms of the usage on the digital services. And so, we’re starting the journey of using that for A&R purposes to say, ‘Are there diamonds in the rough here who aren’t a member and how do we identify them and direct our A&R staff to reach out to them?’ McGlaughlin says. “And then even for the members who are newer and maybe are SOCAN members, how can we use that data to help them in their careers, and help them develop their careers based off the data we have? So, we’re starting that journey and we’re really still in the data collection and standardization phase of that and how we look at it — but 100% that’s in our goals and what we’re trying to do.”

McGlaughlin emphasizes that when it comes to analytics, that data collection and standardization phase is the hard and time-consuming part. The action items — licensing businesses, signing up new members, and providing insights to members based on what the data says – are just the tip of the iceberg.

“The challenges that we’re dealing with on an everyday level is the scale of the data that’s coming in, and how do we efficiently collect the data from as many different sources as we possibly can? Then once the data is collected, how do we model that data to support our analytics?” he says. “It all comes back to what is the question that we’re asking of ourselves on the analytics front? We’re always focusing on that question, right? It’s great to collect data and to model the data, but if you’re asking the wrong questions, it’s not going to provide the right value. So, the biggest thing that we focus on is standardizing how we’re collecting data and making sure that all of our analysts have access to the data to be able to do the work that they need to do to provide that insight. Because you could ask me a question and if we haven’t effectively captured the data and stored it and provided it, then the analysts can’t give you the answer. And so, we want to be able to look proactively and have access to that data and get ahead of what those questions are.”

In terms of using its analytics to help songwriting, composers, and publishers, SOCAN’s mission is to do something akin to what Chartmetric does for labels and others —aggregate and analyse data from many sources to provide a clear overall picture of their own career.

“That could be an analytics product that says, ‘You’re doing really well on a certain DSP in this region or territory,’ and how can we provide that information to them so that they can say, ‘Okay, well next time I’m going on tour, I can book a show and try and focus there because I’m doing extremely well.’ So, giving them as much insights into where the royalties are coming from as possible to help them with those decisions. That’s the vision and where we are pushing towards getting to,” McGlaughlin says.

Speaking of the live music side of things, analytics is being used to inform decisions by artists and agents, such as McGlaughlin’s example of where an artist should tour based on where people are listening, but there is still room for growth in that sector.

“I need to put this diplomatically… and this is coming from different folks, but the amount of money that’s being exchanged in a lot of these [festival and concert] deals, and the relative lack of data informing a lot of those agreements, is something where I think we can really serve well, on both sides,” says Joven at Chartmetric. “It could be a festival trying to book a slot a year from now on one of their second stages, and they’re trying to predict what’s going to be really big in whatever city for the best rate. And for all those artists who are vying for that spot, how can they convince that promoter that they deserve to be in that slot, versus whatever competing artists have a similar sound and a similar fan base in that particular locale?”


‍In terms of music analytics’ current shortcomings, there is one metric that Culbertson wishes for that isn’t currently available — what he terms “lean-forward listening charts.”

“Imagine all-genre and by-genre charts at the national and [regional] level that omit programmed listening,” he says, meaning: “Remove any plays that come from editorial/programmed playlists and algorithmic features, and simply base the chart on listening that comes from personal libraries, artist pages, album/single product pages, and search. It would democratize hits, make songs less dependent on editorial favourability/placements, and let genuine consumption lead the charts.”

Overall, though, the answers are out there to be found in the data. Collecting that data, knowing what questions should be, and analysing the data to better and more accurately inform decisions remains a never-ending a mission for folks in many facets of the industry.

“I don’t think in the history of the music industry have artists been able to have access to the data that big companies have had access to. So, I think it can be very revolutionary, but I think that the hurdle is to show artists the value of that data, and to help them understand how to use it, how to get the most out of it, and how to make it actionable,” says Charmetric’s Rosenborg in closing. “The more transparency and ability there is for artists to have the same tools that big companies have, I think the better it’ll be for everyone. Not everyone might think that, but it’s just my own personal opinion. But yeah, the more we can put tools in the hands of artists to actually be able to understand who their audience is – know where to target, know that they’re getting a radio play in some country overseas that they had no clue about, or know that they’re being streamed a lot in some other country that they never would have thought about – I think those things are very important for artists to be able to understand.”