Understanding music rights data: the challenges of delivering timely royalty payments to artists

By: Julie Knibbe

Published: 2021-04-12

This is the final installment in a five-part series about how music professionals use data across multiple industry sectors, including A&R, artist development, live events and royalty management.

Navigate to the rest of the series here:

One of the music industry’s blessings is that there is no singular “correct” way to build a career as an artist. There are multiple potential paths to success, especially as technology advances and the business becomes more global.

But there’s one common factor that almost every artist around the world has to worry about at some point: Royalty payments.

Recording and publishing royalties are essential for artists not just to make a living, but also to help them gauge the commercial performance of their work to inform future career opportunities. Hypothetically, royalty payments should also be one of the areas of the music industry that is easiest to automate, with streaming services tracking music consumption in real time and sharing consumption reports with stakeholders on a monthly basis.

But for the most part, financial innovation has not caught up to the speed of cultural innovation in the music industry.

We expect to see songs go “viral” and rack up millions of streams overnight, but the artists behind those songs might not get paid until several weeks or even months after the fact. This is especially the case with publishing royalties, where an estimated 25% of music publishing revenue doesn’t make it back to its rightful owners due to lack of accurate metadata.Case in point: The Mechanical Licensing Collective, formed in the U.S. as part of the Music Modernization Act, recently reported $424.4 million in unmatched or “black-box” streaming royalties — i.e. royalties that have not been paid to songwriters and publishers because they could not be connected properly to the correct recordings.

As a whole, music rights holders and tech platforms are trying to move forward with the two following goals in mind:

In this piece, I’ll explain why song metadata matters, and who should care about it. Then, I’ll explain in more detail the challenges of collecting and distributing royalties to their rightful owners, and highlight a few measures and improvements currently underway that would ease the pain of managing copyright data and increase the reliability of royalty payments for artists in the long term.

Why is song metadata important?

In short, metadata is data about other data (hence the “meta-” prefix). In our case, music metadata consists of both qualitative and quantitative data about a given song.

I usually distinguish among four different data types, in no particular order:

For this piece, we’ll focus on ownership metadata, which is meant to identify who owns the rights for a given recording or work, in a given territory, over a specific time frame.

To put it simply: Most businesses that monetize or license musical recordings and works at scale — DSPs, social media platforms, online music stores, physical music venues, etc. — rely on song ownership metadata to distribute royalties to major rights holders and licensing partners (i.e. collection societies, major labels and publishers, self-serve distributors, royalty funds like Hipgnosis, indie licensing bodies like Merlin). Those partners are then responsible for wiring the royalties to individual artists, producers, songwriters and performers.

The record industry generally does direct deals with digital streaming platforms, which then report sales to and pay said rights holders on a monthly basis. Digital distributors like Stem, TuneCore and Amuse as well as third-party startups like Paperchain, Hifi and Revelator are building tools to expedite and advance recording royalty payments more quickly to independent artists, on a weekly or even daily basis (more on that later). In response, major labels have been forced to catch up, with the likes of Sony Music introducing “real-time” royalty payments and advances for their roster artists in recent years. Even though complaints still regularly surface around labels failing to pay artists and producers on time, these tend to be more specific to the label’s own bureaucratic processes and contracts, rather than to the technical limitations of DSPs.

Publishing, though, is a whole other ordeal. Delays are usually quite long between the time writers get their music out and the time they get their first payment — anywhere from six to nine months after a song is first streamed, if they’re lucky.

Not only are direct deals rare between publishers and DSPs rare, but the landscape of publishing rights holders also varies widely by individual geographic markets. Depending on the country, you might have one single society handling both mechanical and performing rights (e.g. SACEM in France and APRA/AMCOS in Australia), or you might have multiple different societies that each are responsible for a different kind of right (e.g. in the US, the MLC initiative is newly appointed to collect digital mechanical rights, while ASCAP and BMI focus on performing rights). In addition, the exact splits between mechanical and performing rights, and between writer and publisher shares, vary from country to country, often based on government regulations.

Long story short, all these entities need to know who is involved in a given work and recording, to establish splits correctly and distribute their earnings. Importantly, these platforms rely largely on song metadata that artists, rights holders and licensing partners input themselves. If platforms receive incomplete or inaccurate information from these partners — or if rights owners don’t submit their works, period — they can’t trace revenue back to its rightful owners. Those unclaimed royalties end up in a “black box,” waiting for more information in an attempt to trace revenue back retroactively. If they don’t succeed, they either allocate money to a general fund or redistribute those royalties to all authors, songwriters and publishers based on their market share — an approach that is controversial because it arguably just makes the rich even richer.

The challenges of collecting and distributing publishing royalties to their rightful owners

In short, song metadata matters because it makes sure the right people get paid at the right time. Yet, especially on the publishing side, it’s notoriously difficult to nail down, for the following reasons:

1. Responsibility — and blame — is decentralized

Who is even responsible for managing music metadata in the first place? The answer isn’t so clear-cut.

Some may argue that the burden is ultimately on the artist (and/or their team) to register their songs properly with accurate metadata across a wide variety of platforms (e.g. the likes of SoundExchange, ASCAP and BMI in the U.S.). Yet, as shown in my diagram above, publishing royalty chains are quite complex and fragmented, with every link in the chain potentially introducing delays, errors or loss of information.

To mitigate this complexity, trade organizations have spent decades establishing metadata standards for the music industry at the creation, delivery and reporting levels — like ISRCs for recordings (developed by the IFPI in the 1980s), ISWCs and CWR for works (developed by CISAC in the 1990s and 2010s, respectively) and a whole alphabet soup of more recent standards like ERN and RDR as introduced by DDEX.

Hypothetically, these standards should act just like barcodes on physical products: Being able to provide a fixed point of reference when a given recording or work is used across different services, across geographic borders or under different licensing deals. Yet, there remains no standardization at the database level for how these recordings and works are ultimately input and stored — let alone used by third-party businesses like Spotify to facilitate payments (more on this in the next section). As an artist, once you publish your song on streaming platforms and register your work with a collection society, the process of editing this ownership metadata after the fact is difficult, if not impossible, due to this lack of standardization and interoperability.

This means that there is no universal, authoritative source of “truth” for song ownership information.

To date, all attempts to build this source of truth in the music industry have failed. Perhaps the most notorious example is the fiasco around the Global Repertoire Database (GRD) — an industry-wide initiative founded in 2008 and backed by millions of dollars in investment from a coalition of publishers and collection societies, with the aim to provide “a single, comprehensive and authoritative representation of the global ownership, administration and control of musical works,” in the words of PRS for Music at the time. But after six years of back-and-forth and millions of dollars in funding, the GRD was ultimately shelved, reportedly due to inactivity and reluctance on the part of collection societies to share data so openly. The major lesson from the GRD fallout is that historical delays in financial innovation in the music industry have been equally as political and contractual as they have been technical.

[Pictured above: Example of Blokur’s graph-based matching algorithm, mapping relationships among various music data entities.]

2. ISRCs and ISWCs are not always accurately documented

Now let’s dive specifically into the issue of matching sound recordings with their corresponding works — which, in most cases today, involves matching ISRCs with ISWCs, respectively.

Even though these identifiers are meant to be unique, they’re often not. Several ISRCs are often attached to the same recording, which can happen when labels submit recordings several times to distributors to edit performers’ information, change the name of the label as a whole, etc. In extreme cases, recordings can end up with up to 1,000 ISRCs each.

To reiterate, such a mess is not causing royalties to get lost on the recording side, as a given recording is always attached to a record label. Therefore, there is no incentive for the recording side of the industry to improve their practices. However, this inconsistency of ISRCs is messing up collection societies’ task of matching recordings with their underlying works. If they do find a work match for a recording, it’s likely that they have to do that matching process again when a new ISRC for the same recording pops up.

ISWCs are not perfect, either. Two different works can end up with the same ISWC by mistake, or a single song can have two or more conflicting ISWCs. These sorts of errors can happen when one or more publishers submit the same work several times to the same collection society. Or, the works registration info might be missing altogether. According to music rights startup Blokur — which provides an automated metadata conflict detection and reconciliation service to publishers — only about 30% of songs in publisher databases even have an ISWC in the first place, and that’s an optimistic average.

As Blokur’s founder and CEO Phil Barry told me:

“We don’t use the [ISWC] identifier as the definitive answer. It’s just one piece of metadata that might help you to identify the song, along with other pieces of information like the title, alternative titles, songwriters and their alternative names, publishers’ internal IDs, etc. We build a web of relationships surrounding the song and we use all of that context to identify what is a unique song, or what is a match. In the end, I’m not sure that the idea of a single true definitive identifier practically could exist.”

Like Blokur, each intermediary in the publishing royalty chain has developed its own proprietary algorithms to match recordings with works or compositions. While this investment should be a significant competitive advantage for any company in the music industry, there are no commonly agreed-upon rules for managing data conflicts — meaning that one song match for one CMO may not be the same for another.

As Julien Lefebvre, Head of Architecture and Innovation at Sacem, told me in an interview, matching ISRCs to ISWCs efficiently depends on several factors, including the “quality and reliability of the data provided by the DSP, which is highly variable; the quality and consistency of work registrations provided by the various and sometimes numerous rightsholders; and of course, the ability of the [collection society] to build a matching engine combining performance (with billions of usages processed every month), reliability and cost efficiency.”

This leads to a difficult irony in the current environment: The faster processing goes, the more mistakes are likely to occur. The consequence of these data discrepancies is that royalties don’t make it back to their rightful owners, and significant time and resources are spent in reconciling information and managing claims. CISAC recently announced a new ISWC system that assigns codes centrally to avoid duplication and inaccuracies — but this was only announced in September 2020, and took 15 years to develop.

3. Creation- and contract-level trends in songwriting increase the odds of inaccurate, conflicting, or delayed song deposits

Several evolving trends in songwriting as a whole are further complicating publishing royalty payments — even as they might be making the creative, collaborative and fundraising processes easier for songwriters as well.

For one, the number of credited writers on an average Billboard top-10 song has consistently hovered at around five writers since 2015, with the extremes getting higher (e.g. “Sicko Mode” crediting over 30 writers, as pictured earlier in this piece). The proliferation of larger collaborations usually results in late work registrations — sometimes several months after the song has been online on streaming services — as most collection societies require all composers’ and writers’ signatures to register a given work.

Licensing agreements and publishing deals are also becoming more complicated, as they evolve over time to meet songwriters’ needs. Songwriters usually enter into a main publishing deal in a given territory, which encompasses sub-publishing agreements with other regional publishers around the world to manage other territories. Collecting societies then need to know not only about these ownership shares, but also about all the licensing and sub-publishing agreements that took place as part of that contract, in order to be able to pay the right amount of royalties to each rights holder. The rise of rapid-fire copyright buyouts from the likes of Hipgnosis, Round Hill and Shamrock Capital is not making this task any easier.

What are the solutions — and who’s building them?

Despite these challenges, startups and trade organizations are working on several measures that will ease the pain of managing copyright data in the long term.

To revisit our argument at the beginning of this piece: Making royalty payments easier for the music industry is about getting it right, and getting it fast.

Get it fast: Strive for simpler, more global licensing deals to decrease database complexity

Complexity in rights databases only model and reflect what is happening in the real world. In turn, more simplified music licensing models would make data management much easier.

For instance, even though most streaming services operate at a global scale, a lot of music licensing deals that have different terms on a territory-by-territory basis have yet to reflect that globalized reality. As the MMF puts it in their song royalties guide:

“If you were starting from scratch, no one would invent the current song licensing framework for streaming services. The current territorial approach and the employment of multiple royalty chains for single streams of single songs is the byproduct of systems, institutions and reciprocal partnerships that were created for an analogue era.”

There is a belief among a few in the music industry that long-established music companies thrive off this complexity as a form of protection from disruption by external players. But soon that benefit will no longer be relevant, as multi-territory mandates and copyright hubs from the likes of Armonia or ICE — who simplify deals with a multi-territory approach and provide a “one-stop shop” to license music for digital players — are getting more and more in fashion.

Get it fast: Identify — and pay out — music consumption more proactively

Rarely do licensing partners proactively monitor usage on digital services before receiving consumption reports, nor do they thoroughly audit what is reported. The infrastructure and investments required to conduct such an operation is indeed quite costly for now.

However, being able to know what’s consumed at scale and in a timely fashion would enable right holders to have more predictability regarding what they’re expected to earn in a given period of time — or, on a higher level, what their copyrights are worth.

Pex, a tech company specializing in identifying audiovisual content across the web, recently raised $57 million in funding to further their vision of creating a better and more proactive licensing and payments system for digital media, with a focus on servicing major rights holders with larger catalogs. On the individual artist level, services like Paperchain and Revelator’s Original Works app allow artists and creators to cash out on anticipated royalty payments on a daily basis, instead of waiting a month or more for statements from their distributors to come in.

Get it right: Embrace transparency and interoperability

Today, it is standard practice for streaming services to make crucial data like ISRCs or their own content identifiers freely available in their APIs. This openness, which is commonplace in the tech industry, has benefited the recording industry by enabling more interoperability among different software entities, in spite of the shortcomings of the ISRC format.

The publishing industry, however, has not been as open on a technical standpoint. First and foremost, this poses a significant disadvantage to the artist, from the very moment they need to register their catalog. As the artist and engineer Kimberly Hou pointed out in a recent Water & Music article: “Every time I want to release a new song, I need to log in to each company’s individual site and fill out much of the same information about this song across each platform in a repetitive way.” This lack of standardization can easily lead to registration and matching errors, which then trickles down to the inability to pay rights holders properly. The fragmentation of the registration process also means that streaming services usually don’t know what songs are connected to which recordings.

Since metadata challenges will hardly be solved in the blink of an eye, it’s best to stop trying to hide them. Solutions that assume imperfect metadata from the start have the best chance of short-term success. Namely, being more transparent about the source of data clashes, duplicates and mismatches enables stakeholders to participate in solving these issues. Nowadays, more databases are emerging that make publishing metadata more open and searchable, such as Rightsholder.io, CISAC’s CIS-Net and ASCAP/BMI’s Songview. As mentioned earlier, Blokur also provides automated conflict detection and reconciliation tools to publishers.

In today’s digital world, where we are aiming for interoperability, “value is not really in the raw data itself — it’s in the way that it’s matched and the insights you gain from it,” argues Barry. “It’s better to aim towards an ecosystem where data can be shared efficientlyso that it’s possible to take the best bits from different places and combine them together. If we could open the borders between different organizations, then we could compare and enrich data more easily, and problems like the availability of ISWCs could be solved.”

Get it right: Bring publishing and recording datasets closer together

Labels usually don’t provide publishing/works information about their recordings to their distributors — meaning that information is not passed along to the streaming service, who then can’t identify songwriters and publishers automatically.

That said, recent developments like Spotify’s implementation of public credits for songs and publisher-facing analytics has prompted distributors, labels and other licensing partners to provide more complete metadata for all their songs. If this becomes a new standard, recording rights holders may have to wait for underlying works to be registered before distributing their recordings to streaming services, which is not currently the case. While that would result in increased delays early on to get a song out, it would favor faster and more accurate royalty collection and distribution after the fact.

Singapore-based startup Fairphonic is experimenting with this approach in emerging markets by managing all data from beginning to end. Fairphonic gathers all recording and publishing metadata for a given label’s catalog, in addition to handling the distribution process themselves. They also retrieve consumption data on both works and recordings directly from DSPs on a daily basis, via an API, and then pay royalties on a monthly basis.

Similarly, Bluebox, the suite of blockchain-based tools launched by distribution and services company Ditto Music, is working on a system that allows artists to instantly register copyright, publishing and mechanical splits at the point of creation — i.e. in the studio — using decentralized and legally-binding smart contracts. As a proof of concept on the recording side, Bluebox recently held NFT auctions of royalty shares in songs by Taylor Bennett and Big Zuu. Importantly, it is because the artists own their copyrights that they are able to pioneer these kinds of auctions.

For those who cannot access this more integrated approach at the distribution level, New York-based HIFI works on aggregating and analyzing artists’ and labels’ royalty statements from across multiple sources (i.e. DSPs, PROs, publishing administrators like Songtrust) after the fact, in one central hub. Similar to what Paperchain and Revelator offer, HIFI’s “Cash Flow” service also pays clients twice a month based on their anticipated earnings across both recording and publishing — mirroring a wider ecosystem of tools that allow artists to borrow against future income, but in a way that captures a more holistic picture of artists’ entire copyright businesses.

As you can see, there is no silver bullet for solving the problem of music royalty payments. There are literally dozens of moving parts involved, each with their own technical, legal and political challenges to navigate.

That said, everyone from the most established PROs to the latest emerging startups are working together to embrace more proactivity, interoperability, transparency and simplicity moving forward, which will make sure artists and rights holders will get paid in a more prompt and precise manner.

My own forecast is that the divide between publishing and recording datasets in particular will continue to get blurrier — up to the point where matching recordings with works will simply become a commodity, just like with catalog size for DSPs. Once this gets achieved, music rights companies, who currently differentiate themselves primarily by how quickly they can collect and distribute revenue, can shift to focusing on more holistic services for rights holders, like better data analytics and more predictable cash flows — without old-school silos getting in the way.