Most artists spend years perfecting their sound and minutes filling in their metadata. That is backwards. In a catalog of thousands of tracks, your metadata is the difference between your music getting found and your music sitting unseen. Here is how to do it right.

Why Metadata Matters More Than You Think

When a music supervisor is searching for "melancholy piano instrumental, 90 BPM, for a grief scene," they are not listening to tracks one by one. They are running a search. The tracks that surface are the ones whose metadata accurately described exactly that. Every field you fill in is a potential match.

The same is true for AI agents browsing a catalog programmatically. An AI agent building a video does not browse by feel. It queries a database. If your metadata is thin or inaccurate, your track does not exist to that agent. If it is detailed and accurate, your track is a candidate every time someone searches for what you actually made.

Track Title

Your track title should be the actual name of the song. Do not use the filename, do not abbreviate, and do not add version suffixes in the title field. If you have an instrumental version, it should be a separate track with "Instrumental" added to the title. Titles like "Track 04 Final v3" do not appear as professional.

If your song has a subtitle or version name, use the version field if your platform provides one, not the title field. Keep the title clean.

Description

The description is the most powerful metadata field and the most commonly neglected. A strong description does three things: it tells a story about the emotional and sonic character of the track, it uses the natural language a music supervisor or AI agent would actually search for, and it describes what the track is good for.

A weak description: "Upbeat track with guitar and drums."

A strong description: "A driving, optimistic rock track built around a punchy electric guitar riff and a propulsive drum groove. Energetic and forward-moving, with a sense of momentum and confidence. Works well for product launches, athletic montages, travel content, and scenes involving determination or ambition."

Notice the strong description includes genre feel, instrumentation, emotional quality, pace, and use cases. A music supervisor searching for "confident driving rock for a car commercial" will find the second track. They will not find the first.

Genre and Mood

Choose the most accurate primary genre, not the most flattering one. If your track is background jazz, label it jazz. If it is lofi hip hop, label it lofi hip hop. Music supervisors filter by genre constantly. A misclassified track is an invisible track.

Mood is where many artists undersell themselves. Most platforms let you add multiple moods. Use them. A single track can be melancholy, intimate, and cinematic at the same time. All three moods are valid search terms that different supervisors will use. Tag all the moods that genuinely apply rather than picking just one.

Keywords and Tags

Keywords are your opportunity to add context that does not fit neatly into other fields. Good keyword choices include instrumentation details, production style, tempo feel, era or decade references, and specific use cases.

Examples of useful keywords: acoustic guitar, finger-picked, warm, coffee shop, studying, background, cinematic build, orchestral swell, tension, 80s synth, retro, driving, workout, heroic, underscore, emotional, bittersweet.

Do not repeat your genre or artist name as keywords. Those fields exist elsewhere. Use keywords to add information that is not captured anywhere else.

BPM and Key Signature

BPM and key signature are used heavily by music supervisors who need to match music to a specific edit or scene. A supervisor cutting to a 92 BPM heartbeat will filter by BPM. A supervisor looking for something in a minor key for a tense scene will filter by key.

Measure your BPM accurately rather than estimating. If your track has a tempo change, use the dominant tempo. For key signature, use standard notation: C major, A minor, F# minor, and so on. Do not leave these fields blank.

Similar Artists

The similar artists field is one of the most effective search tools available to music supervisors. When a supervisor is building a scene for a show and wants something "in the vein of" a well-known artist, they search for that artist's name. If your music sounds like that artist and you have listed them, you surface. If you have not, you do not.

Be honest and specific. List artists whose sound genuinely compares to yours, not artists whose success you admire. Three accurate comparisons are more valuable than ten aspirational ones.

A practical rule: After filling in your metadata, read it back as if you are a music supervisor who has never heard the track. Does it tell you exactly what the music sounds like, what emotion it carries, and what it could be used for? If you could not make a confident placement decision from the metadata alone, add more detail.

What to Avoid

  • Leaving description, keywords, or mood fields blank
  • Using vague adjectives like "great," "unique," or "professional" that tell a supervisor nothing
  • Copying the same description across multiple tracks that sound different
  • Listing similar artists whose sound is nothing like yours
  • Using filenames as track titles
  • Estimating BPM instead of measuring it

Metadata for AI Discovery

As AI agents become a larger part of how music gets discovered and licensed, metadata becomes even more important. An AI agent does not listen to your track before deciding whether to license it. It reads your metadata, evaluates whether it matches the task at hand, and either selects your track or moves on.

Platforms like OnChain Music expose catalog metadata directly through a REST API and MCP server. Every field you fill in becomes a queryable attribute that agents can filter, rank, and match against. A track with complete, accurate metadata is always more discoverable than one without it, whether the searcher is human or machine.