MEDIA & ENTERTAINMENT

Music metadata and royalties: how identifiers, splits and matching decide what you get paid

Why missing or wrong ISRCs, ISWCs, IPIs and splits leave royalties unpaid, and a practical plan to clean your catalog metadata and recover money.

By Akash Soni, Founder & CEO, Zuroth · Updated 9 Nov 2026 · 9 min read

The short answer

Royalties are paid on matches. A streaming service reports a play; collecting societies and administrators try to match it to a recording (ISRC), a song (ISWC), the people who wrote it (IPI) and the agreed split. If any link is missing or wrong, the money waits in an unmatched pool or goes to the wrong place. Clean, connected metadata is the cheapest way to raise royalty income.

How music metadata becomes money

The identifiers that matter

Identifier What it identifies Who usually assigns it
ISRC A specific sound recording or music video The label or distributor
ISWC A musical work, the composition Collecting societies, via the publisher
IPI / CAE A writer or publisher Collecting societies
UPC / EAN A release, such as an album or single The label or distributor
ISNI A public identity, such as an artist or band ISNI registration agencies

Two chains, one song

Every stream pays two sides:

  1. Recording side: the master owner (label) and performers, linked through the ISRC.
  2. Publishing side: songwriters and publishers, linked through the ISWC, IPI numbers and splits. In the US, the Mechanical Licensing Collective (MLC) handles streaming mechanicals; performing rights organizations handle performance royalties.

If an ISRC isn't linked to its ISWC, or the writer split adds up to 90% instead of 100%, the publishing money can't be paid correctly.

Where it goes wrong

  • Recordings released without the songwriter data attached.
  • Different spellings of the same writer across systems.
  • Splits agreed by email but never registered.
  • Duplicate works registered by co-publishers with conflicting shares.
  • Old catalogs digitized without any identifiers.
  • Acquired catalogs with data in incompatible formats.

How to fix it: a 6-step plan

1. Measure before you clean

Take a sample of 2,000 to 5,000 tracks across eras and labels. For each, check whether the ISRC, ISWC, writers, IPIs and splits are present and agree across your systems and society registrations. This gives you an error rate and a money estimate.

2. Create one golden record

Bring recordings, works, contributors, splits and agreements into one rights data model, with a single ID per entity and links between them. Keep the source of every field so you know which system to trust.

3. Match with machine learning, approve with people

Use fuzzy matching on titles and names, audio fingerprints to link recordings to works, and rules to flag splits that don't add up to 100%. Queue suggestions for a rights analyst to approve; never auto-register.

4. Fix registrations at the source

Send corrected data to societies, the MLC and administrators in their formats, such as DDEX messages and CWR files, and track each submission until it is accepted.

5. Reconcile income every quarter

Compare expected royalties, from usage and your splits, with what was paid. Big gaps point to new matching problems or underpayments to claim.

6. Stop new errors at release

Make metadata complete before release: no ISRC without linked works, no work without 100% splits and IPIs. Add these checks to your release workflow.

How Zuroth helps

We build rights data platforms for labels, publishers and music-tech companies:

  • Data engineering: a single rights data model that brings together catalog, contract and usage data.
  • Machine learning: matching and de-duplication of recordings, works and contributors, with analyst approval queues.
  • Business intelligence: royalty reconciliation dashboards showing expected versus paid income by society, territory and platform.

Our vocal intelligence case study shows the same data foundations at catalog scale: 2.4 million recordings indexed and searchable.

Next step: book a free 30-minute call to plan a sample metadata audit for your catalog.

Frequently asked questions

What is the difference between an ISRC and an ISWC?

An ISRC identifies a specific sound recording, so a remaster or live version gets its own ISRC. An ISWC identifies the underlying musical work, the composition, which can have many recordings. Royalties for recordings and for songs flow through different chains, so you need both linked correctly.

Why do royalties go unmatched?

Royalties go unmatched when a usage report can't be linked to a registered work or rights holder, usually because of missing or inconsistent identifiers, misspelled titles or writer names, missing splits or late registrations. Unmatched money is held for a period and may eventually be distributed by market share.

How much revenue can metadata clean-up recover?

It depends on catalog age and data quality. Older catalogs with incomplete writer and publisher data usually gain the most. The safest way to size it is a sample audit: match a few thousand tracks against usage reports and registrations, then extrapolate before investing in a full clean-up.

Can AI help clean music metadata?

Yes. Machine learning can fuzzy-match titles and names across spellings and languages, suggest missing identifiers, spot duplicate works and flag conflicting splits. Audio fingerprints can also link recordings to works. People should still approve changes before anything is re-registered.

What is DDEX?

DDEX is a set of standard message formats used across the music industry to exchange release, sales and usage data between labels, distributors, streaming services and collecting societies. Using DDEX consistently reduces manual re-keying and the errors that cause unmatched royalties.

Akash Soni

Akash Soni

Founder & CEO, Zuroth

LinkedIn ↗

FREE CONSULTATION

Talk it through with someone who does this work.

30 minutes with a senior expert. You leave with your top 3 risks and clear next steps, whether or not we work together. NDA before we talk.

Book a free consultation →