The Record Label Recovery. Part II: Analysis

Tunes and Tales
7 min read4 days ago


By Patrick Clifton. Why record labels need to move on from using data to impact the short term, to embed a culture of analysis for the long term.

Record labels’ relationship with data is ambivalent at best, fearful at worst. To fully harness the rapid change the industry finds itself in, labels will need to conquer this fear. To do this they must embed a culture of analysis across their teams. This will allow labels to move beyond the exploitation of data for short term outcomes to the analysis of data for long term artist development and for the growth in consumption of their repertoire.

Despite prevailing assumptions, record labels have been using data for decades. To get a record on the radio required plugging, in which the label would supply the radio station with a bunch of datapoints to convince them the record was “hot” at specialist level. Pre-streaming, radio airplay was a leading indicator of success and a driver of performance revenue, so was measured by promotions and sales teams. The sales chart measured the transactions everyone was employed to generate. The chart generated a fair amount of data — format and regional splits, day-by-day sales histories — and analysis was done by sales teams or by analysts employed for that purpose, who would share the results with the marketing and A&R teams executing those release campaigns.

Airplay and sales charts provided information about snapshots in time — short term metrics that could be acted on immediately. The primary example of this was the “midweek battle” — midweek charts would tell labels the position of their releases relative to others and teams would deploy tactics to affect these metrics, like short term price drops or additional marketing spend with the goal of securing higher chart positions. Fast forward to the streaming era and label teams are engaged in similar practices even as the paradigm is different and the amount of data that is accessible is a step-change greater in terms of scale and complexity. The midweek chart is still consulted, and actions are taken to affect it, with some DSP’s even creating programmes to alter the chart trajectory of releases. Track download price drops are actioned to increase sales. Label focus is still the use of data to alter the here and now. But these tactics are less effective and less predictable than they were pre-streaming.

This is best described as the marketing use of data and is analogous to how information was used in the transactional era. The use of data in the A&R process is problematic for many in the business. A&R teams look at data from streaming but especially from TikTok to detect viral trends in new music; if a network effect is evident, this may cause the track to be signed.

This has caused a lot of disquiet in the business — a lot of money seems to be being spent on deals for some of these tracks; anecdotally, many of the songs burn bright and then disappear and their consumption half-life is short. The record labels don’t get a jump-start in artist development, as the process of building a genuine, engaged fanbase is painstaking and lengthy. As well as this, the use of data in signing and marketing music is a problem for many artists. As outlined in Midia’s “Sustainability from Chaos” report, “artists and their teams have prioritised playlist[s]…[but] For artists…indicators that [they] have found true fans have become more complex…establishing whether this…adds up to more fans…remains as hard as ever.” In his blog post outlining the study, Keith Jopling explained that “Over the past decade, the music industry’s approach to talent discovery, marketing, and artist careers has become too data obsessed.”

The way people express these misgivings, it sometimes feels as if data is to blame. It’s part of the narrative subtext in music, of a dystopian future, where music generation is the product of machine learning and songwriting is optimised by malign algorithms, fuelled by data, for the means of distribution. But the issue isn’t that data is being used to discover music that’s popular on TikTok — it’s that the same resources aren’t being deployed across label teams to help them develop the long-term careers of artists that have been signed outside of the viral hits process. This isn’t just due to resource constraints but also due to the widespread ambivalence about data inhabiting the label landscape. This may be due to the ineffectiveness of the use of data in current approaches to artist marketing, but is also a product of ambivalence about the data’s role in the creative process.

This ambivalence is an obstacle that needs to be overcome. The kind of music we as an industry are passionate about creating must continue to be made by humans and reflective of our unique human experiences. “Data” has no role in its creation. Once created, however, data is critical to our understanding of that music’s ability to find an audience, which is the thing that is most important to artists (as evidenced in the “Sustainability From Chaos” report referenced above). Many artists are already savvy to this, for example crowd-sourcing fan engagement and/or consumption data to determine which track versions and remixes to subsequently release — a tactic often attributed to Fred Again, so often the poster child of modern artist development.

Considering the value chain from the artist’s invention to the fan’s action of listening to that music, once it’s made, there’s no point along the chain that doesn’t involve a technological platform that provides data to the end user, be it social media, DSP, e-commerce or distribution. These tech firms obsess about user engagement as it drives their own measures of success. Their employees will make decisions based on engagement data they observe; often there will be overlap with specific artists and their music (for example, where decisions are made about artist visibility in-app or about music curation). To put it another way: if you’re not looking at data, your counterparts at those services will be — and will be making decisions based on what they see. If you have the requisite analytical skills, you’ll be empowered to assert your viewpoints or to provide counter-arguments, based on a common language and perspective, which will be more effective.

Earlier in this article the opportunistic transactional-era use of data was described. This framework for data continued into the streaming era in the marketing of tracks and in A&R teams’ detection of viral hits on TikTok. What’s now required is a radical upskilling of record label teams so that they can move beyond this tactical use of short-term retrospective metrics to much deeper and valuable analysis of longer-term trends. Below are some examples I plucked out of the air — these are the questions I have come up with but I’m not in the job of artist development right now and I’m sure there are better and more sophisticated questions to ask.

· Benchmarking of track volume and audience size across multiple dimensions (track vs track, artist vs artist, track on platform a vs track on platform b, etc);

· Examination of repertoire volume growth over time (week on week, month on month etc.) rather than reference to chart positions on DSP’s and the Official Chart; growth benchmarked against total stream growth rate.

· Streaming penetration analysis (effectiveness of artist audience penetration over time; audience size expressed as a percentage of total available audience).

· And end to trendlines with guesswork that pins events to inflection points (a great example of that is this Billboard article about Chapel Roan’s recent success) and in its place, linear regression analysis to determine robust correlative independent variables — with the long-term aim of building predictive models (a pipedream, but worth attempting).

· In-depth fan segmentation and fan behaviour analysis -revenue and platform segmentation, CRM analysis, engagement analysis, ticket purchase trends, e-commerce basked segmentation, etc.

· Fan conversion analysis — what are the actions that change a casual listener to a fan? Is any one action more potent than another or is the spread of actions more important?

· Comparative sources of revenue, their interrelation and their growth over time.

What’s the point of all this? As described in The Record Label Crisis, the old artist development funnel is broken; to break an artist now takes five years at least, where it used to take less than two. Once limited to radio, TV and retail, the place where these artists build an audience is social media and streaming services. Their engagement and “stickiness” as fans of that artist — and their downstream potential as economic contributors to an artist’s career — takes place via CRM, on e-commerce platforms and through ticketing apps. A long term process that takes place across a wide variety of places, each with their own metrics, can only benefit from a joined-up approach to understand how success can be generated over time.

To facilitate this, labels will need to decentralise analysis, out of finance, royalty or streaming teams and into the campaign management teams who are managing artist’s trajectories day-to-day. They’ll also need to re-engage streaming services to improve the visibility and quality of data available. Artist data apps provide information that works at surface level, facilitating the kind of short-term retrospective worldview that is a hallmark of an era that no longer exists.

This article follows both The Record Label Crisis and The Label Recovery, Part I: Purpose. If you’re going to set new measures of success, it follows that you should have the means and the desire to track your progress against these measures.



Tunes and Tales

From Clifton Consults founder, Patrick Clifton. A blog about music and the music industry. Opinions his own.