Preparing Your Catalog Metadata Before You Register Anything

There is an unglamorous truth under every rights system, every royalty pipeline, and every registry including this one: they can only record what you bring to them. Register a catalog with wrong titles, missing writers, and mystery versions, and you now own a beautifully timestamped record of a mess. The registry did its job. The inputs were the problem.

This article is about fixing the inputs — once, systematically — before you file anything anywhere. The work is tedious in exactly one way and valuable in about six, because the same clean dataset feeds your distributor, your PRO registrations, your sync pitches, your split agreements, and your provenance records. Musicians call it admin. Every other industry calls it inventory, and treats it as survival.

Why metadata quietly controls your income

Royalty systems match money to works by metadata. When the metadata is wrong or inconsistent, the match fails — and the money does not vanish dramatically, it just parks. Unmatched royalties from streaming services have accumulated into the hundreds of millions of dollars, much of it stuck for exactly the reasons this article covers: a writer credited as “Jon” here and “Jonathan” there, a track delivered with two different titles, a missing identifier. Nobody is stealing it. It simply cannot find you.

The same defect that misroutes royalties also poisons evidence. If your live version, radio edit, and album master all float around under one title with no version labels, then even a perfect provenance record is ambiguous — which “Midnight” did you register? Precision downstream requires precision upstream.

Build one source of truth

The system does not need software. A spreadsheet is fine; what matters is that there is exactly one of it, it has a row per version of each work, and every other document defers to it. For each row, you want four clusters of information.

Identity. The canonical title, spelled one way, forever. The version label — album master, radio edit, instrumental, demo, live at wherever. The primary artist name, also spelled one way (decide once whether you are “DJ Karma” or “DJ KARMA”; systems treat those as strangers).

People. Every contributor, with legal name, role, and share — separately for the composition and the recording, because those are different works with potentially different participants. A drummer on the session owns none of the composition; a topline writer who never entered the studio owns none of the master. Record their PRO affiliations and IPI numbers where you have them, and — if you intend to register on-chain — the wallet address each contributor confirms in writing.

Identifiers. The ISRC for each recording (your distributor assigns these if you don’t), the ISWC for each composition where one exists, and the UPC for each release. These codes are how machines know two databases are discussing the same work. Collect them as they are issued; retrieving them years later is archaeology.

Files and dates. Where the definitive final file for this exact version lives, its creation and release dates, and — once registered — its asset hash and transaction link. The file column matters more than it looks: the whole exercise fails if “final” is ambiguous, so pick the file, mark it read-only, and back it up somewhere that is not your project drive.

The audit, in five passes

Building the sheet for an existing catalog goes fastest as separate passes rather than song-by-song perfectionism:

1. Inventory. List every released track and every meaningful unreleased version. No details yet — just get every row to exist.
2. Deduplicate. Hunt down title variants, re-exports, and “FINAL_v3_actually” files. Decide which file is canonical for each version and record its location.
3. Credit. Fill in people and shares. Where a split was never agreed, you have found a liability, not a data-entry problem — settle it with the collaborator in writing now, while the song is worth nothing in particular. This pass is where the audit pays for itself.
4. Identify. Pull ISRCs and UPCs from your distributor dashboard, ISWCs from your PRO, and fill the gaps.
5. Verify. Check that what your PRO, your distributor, and your sheet say about each work actually match. Mismatches between these systems are precisely where royalties park.

Then registration becomes trivial

With the sheet done, filing provenance records stops being a project and becomes a batch task: each row already contains the exact file, the title, the contributors with confirmed wallets, and the splits — which is, not coincidentally, the entire pre-registration checklist. Register the versions that matter, paste each asset hash and transaction link back into its row, and your source of truth now points both ways: from your catalog to the public record, and back.

Two cautions as you file. Register from the sheet, not from memory — the point of the system is that memory is no longer load-bearing. And remember that a registry records your declarations without validating them; an error in your sheet becomes a permanent, signed, public error. Clean first, then file.

Keeping it alive

A metadata system decays the moment it becomes optional, so make updating it part of finishing. New song gets a row when it is bounced. Splits get agreed and logged before anyone leaves the session, matching a signed split sheet. Codes get logged when the distributor issues them. Fifteen minutes per release, forever — versus the alternative, which every artist who has done a rights cleanup on a decade-old catalog can describe to you in the tone people use for root canals.

None of this is the fun part of music. But there is a particular calm available to the artist who can answer any question about any track — who owns it, who played on it, which file is final, where it is registered — in under a minute, with receipts. That calm is what metadata buys. The registry just makes it public and permanent.