Methodology

The part most outlets do not publish.

A number without its method is an opinion with a decimal point. This page is the standard every Overtone study is held to, and it is public so you can hold us to it.

1Every study states its sample size in the first screen. If n is small, that is the headline, not a footnote.
2Raw data publishes with the study under CC-BY 4.0. If we cannot release the raw data, we say exactly why.
3Corrections are dated, kept in place, and never quietly rewritten. The original claim stays visible.
4Any commercial relationship touching a subject is disclosed on that piece, in the piece.

Sourcing

We prefer primary documents to reporting about documents. A distributor statement beats a press release. A filing beats a quote about a filing. Where we rely on a secondary source, it is named inline and linked, so you can go and disagree with it directly.

Where a figure is supplied by a person rather than a document, it is labelled as such. "An artist told us" and "we read the settlement" are different classes of evidence, and collapsing them is how numbers get laundered into facts.

Sampling

Sample construction is published with every study: how the population was defined, what was excluded and on what rule, and where the boundary is arbitrary. Most disagreements about music data are actually disagreements about who got counted, so we put that first rather than burying it.

The rule we will not bend: if a study cannot survive its own limitations section being read first, it does not publish. We would rather ship four studies a year that hold than forty that get quietly walked back.

What we refuse to publish

Hard floors

  • A figure we cannot reproduce from our own stored data
  • A ranking position that any party has paid to influence, at any price
  • A lyric corpus republished as text, rather than computed on and released as derived metrics
  • An artist's private financial information obtained without their knowledge
  • A claim about a named person or company we would not send them in advance

Corrections

When we get something wrong we date the correction, leave the original claim visible with a strike, and say what changed and why. A publication that silently edits is a publication whose archive means nothing, and the archive is the entire asset here.

Conflicts

Overtone is independently operated and carries commercial relationships like any publication: sponsorship, affiliate links on gear and software, and partnerships. None of them touch a ranking position. Where a piece covers a subject we have any relationship with, that relationship is disclosed on the piece itself rather than buried on a policy page, because a disclosure nobody encounters is not a disclosure.

Can I use your data commercially?

Yes. CC-BY 4.0 permits commercial use. Attribute Overtone and link back to the study, and you are done. You do not need to ask.

Will you share the raw lyrics corpus?

No. Lyrics are copyrighted works. We compute on them and release only derived metrics, which is both lawful and the reason our datasets can be openly licensed at all.

Do you take pitches?

Yes for data and documents. No for coverage of a release. If your pitch is that we should write about something, the answer is that we write about what the numbers support.

Who funds this?

Sponsorship and affiliate revenue, disclosed per piece. No investor, label or distributor has editorial input, and no commercial party sees a study before publication.