How Metadata Errors Quietly Cost Artists Real Money

Royalties do not find people. They are matched to identifiers, and where those identifiers are wrong, missing or inconsistent, the money does not fail to exist. It fails to arrive. It sits unmatched in accounts held by the organisations that collected it, waiting for a claim that in many cases never comes.

This is the least dramatic problem in the music business and one of the most expensive. It has no villain. It is caused by typing errors, inconsistent name spellings, absent codes, and two parallel data systems that have to agree with each other and frequently do not.

What Actually Travels With a Release

A release is not just audio. It is audio accompanied by a structured set of claims about what the audio is and who is owed for it.

On the recording side, the essential piece is a unique identifier assigned to each individual recording. That code is what reporting systems use to say this specific master was played this many times in this territory. A different mix, a live version, a shorter edit and a remaster are different recordings and require different identifiers. Reusing one code across versions makes reporting ambiguous in ways that are hard to unpick later.

Around that sit release-level details: the release title, the label or distributor of record, the release date, and a separate identifier for the product itself as distinct from the tracks on it.

On the composition side there is a second, entirely separate identifier for the underlying song, along with the writer names, each writer’s affiliation with the organisation that collects on their behalf, the publishers representing them, and the percentage each party holds. Those percentages have to total correctly, and they have to match what every other party has registered.

Then there are contributor credits: performers, producer, engineer, mixer, mastering engineer, featured artists. These have historically been treated as presentation detail rather than as data, which is why they are so frequently missing or inconsistent, and why income associated with performance participation is so difficult to route.

Two Databases That Have to Agree

The structural cause of most lost income is that recording data and composition data live in different systems, maintained by different parties, with no automatic link between them.

The recording side is populated by whoever distributes the release. The composition side is populated by publishers and by the writers’ collecting organisations. Nobody sits in the middle holding both. The two are supposed to be connected by matching a recording to the song it is a recording of, and that match is frequently made by comparing text: song title, writer names, sometimes duration.

Text matching fails in entirely mundane ways. A song registered with an apostrophe and delivered without one. A writer registered under a full legal name and credited under a shortened one. A title delivered with a parenthetical the registration does not have. A featured artist folded into the track title on one side and held in a separate field on the other. Two different songs sharing a common title. A name that can be transliterated more than one way, spelled differently in two systems.

Each of these is trivial in isolation. Each of them can prevent a match. An unmatched recording generates income that cannot be attributed to a composition, which means the writers’ share of it has nowhere to go.

How a Royalty Becomes Unmatched

Following a single play through the system shows where the breakages happen.

A platform reports usage against a recording identifier. It pays the master side, which is comparatively straightforward, because the identifier came from the distributor and points at a known owner.

It also owes for the reproduction and performance of the composition. To pay that, the composition has to be identified, and its owners and their shares have to be known. If the recording cannot be matched to a registered composition, if the composition was never registered at all, if the registered shares do not total correctly, or if two parties have registered conflicting claims to the same share, the payment cannot be made. It is set aside.

Public performance follows a parallel path with the same failure points, complicated further by reporting that is sometimes based on samples rather than complete logs, and by broadcast documentation that may describe a piece of music imprecisely or omit it.

The result in each case is a pool of collected but unattributed money. Where it goes next depends on the rules of the organisation holding it. Some of it is held for a defined period and then distributed to other rights holders by market share, which means it is not lost to the industry. It is redistributed away from whoever failed to register correctly and toward whoever is already earning most. The mechanism is not malicious. It is simply what happens to money nobody claims.

Why the Problem Compounds Across Territories

A registration error that would be one problem in a single market becomes many problems internationally, because collection is territorial.

Performance and mechanical income are administered country by country by separate organisations, each with its own database, its own registration format, its own deadlines and its own tolerance for imperfect data. A composition earning in many countries is represented in many systems, usually through reciprocal arrangements between a writer’s home organisation and its counterparts abroad, or through a publisher’s local sub-publishers.

Errors propagate through that structure in both directions. A wrong share registered at home is replicated everywhere the home organisation reports it. A missing sub-publisher in one country means income earned there has no local claimant even when the registration at home is perfect. Conflicting claims filed in different territories cause income to be frozen in each of them separately, each requiring its own resolution.

Time makes it worse. Organisations operate claim windows, after which unclaimed income is distributed away and generally cannot be recovered. Corrections are usually slow and manual, requiring documentation of the correct position, which is where the absence of a signed split sheet stops being an administrative inconvenience and becomes an insurmountable practical problem. And because statements arrive long after the usage they describe, the discovery that something was misregistered typically happens well after the point at which fixing it was cheap.

The compounding is the important part. A single-character discrepancy, replicated across many national databases and left through a few statement cycles, becomes a permanent reduction in a song’s lifetime earnings.

What Registering Properly Actually Requires

The remedy is not sophisticated. It is doing a short, boring list of things before a release goes out rather than after.

The composition needs registering with the writers’ collecting organisations and, where applicable, by their publishers, with shares that total correctly and identical writer names across every filing. Each writer’s affiliation has to be stated, because that determines which organisation collects for them.

The recording needs its own identifier, issued once and never reused for a different version. Titles need to be written identically on the recording side and the composition side, including punctuation, capitalisation conventions and how featured artists are handled.

Contributor credits need delivering as structured data rather than left in a free-text field or omitted entirely, because credits that do not arrive as data cannot be used to route anything.

Version discipline matters as much as any of it. Every remix, live take, edit and instrumental is a separate recording of the same composition, and each needs its own recording identifier linked back to the one composition registration.

And someone has to read the statements. Unmatched or missing income is only visible to whoever compares what the statements show against what the release actually did, and that comparison is the only reliable way to find a registration problem while there is still time to correct it.

What the Unclaimed Pool Really Represents

Metadata is the least interesting subject in music and it determines who gets paid. The money at stake is not extracted by anyone. It is simply never delivered, because a system that pays by identifier cannot pay a claim it cannot match.

The uncomfortable implication is that a portion of the money sitting unclaimed belongs to people who did the creative work and then spent no time on the paperwork attached to it. Registration is the cheapest task in the entire release process, and it is the only one that determines whether the rest of the process results in payment at all.