BRENNEN

Industries / Media, arts and recreation / Media and entertainment / Music recording

Promotion

Promotion is the work of getting recorded music heard and bought: winning radio play, video views, playlist spots and short-video trends, and placing high on the charts that count them.

7

Milestones achieved, from 1960 to 2026.

3

To watch for: announced, not yet confirmed outside.

3

Open deliverables. Nobody has shipped them.

The road

AchievedWatch forOpen

Achieved

  1. 1960

    Hidden airplay payments outlawed

    Payola is a hidden payment to get a record played on the air. After hearings that, in History.com's words, "destroyed" the disc jockey Alan Freed, Congress amended the US Communications Act in September 1960: anyone who pays a station's staff to play a record, or takes such pay, must tell the station before the broadcast. Today's text sets a fine of up to US$10,000 or up to a year in prison for each violation.

    Lessonthe law moved payola rather than ending it; in 2005 New York Attorney General Eliot Spitzer found Sony BMG paying radio stations through middlemen called "independent promoters", and to settle, Sony BMG paid US$10 million and Warner Music Group US$5 million.

    law.cornell.eduhistory.commail-archive.comnpr.org

    Credit

    the US Congress; the Federal Communications Commission (FCC), which proposed making payola a crime in February 1960

  2. 1991

    Charts count scanned sales

    Billboard's album chart dated 25 May 1991 was the first built on SoundScan data, which counted the bar codes scanned at store tills instead of relying on rankings that stores reported. The Hot 100 singles chart adopted SoundScan data that autumn.

    LessonJim Caparro, a former head of distribution at PolyGram, told Billboard that counted sales moved charts "from a subjective methodology to an objective methodology", so hype counted for less.

    billboard.com

    Credit

    SoundScan, co-founded by Mike Shalett, and Billboard

  3. 2014

    Charts start counting streams

    In February 2013 Billboard and Nielsen added US YouTube views to the Hot 100, and Baauer's "Harlem Shake", carried by fan videos that used the track, debuted at No. 1 with 103 million streams in a week. From late 2014 the Billboard 200 album chart has ranked albums by "album-equivalent units": 10 track downloads or 1,500 streams from an album count as one album sale.

    billboard.combillboard.com

    Credit

    Billboard (Silvio Pietroluongo); Nielsen; YouTube

  4. 2019

    A TikTok meme tops charts

    "Old Town Road" by Lil Nas X spread through a TikTok trend in which users turned into cowboys, and Columbia Records signed him. Billboard removed the song from its country chart, and a remix with Billy Ray Cyrus reached No. 1 on the Hot 100 in April 2019 and stayed 19 weeks, a total no song beat until December 2025, by Guinness World Records' account.

    Lessonby Axios's account, the row over the country chart helped push the song from No. 15 to No. 1 in weeks; a genre chart could turn it away, but the all-genre chart counted what people played.

    newsroom.tiktok.comaxios.comguinnessworldrecords.com

    Credit

    Lil Nas X; TikTok users; Columbia Records

  5. 2020

    Royalties traded for reach

    Spotify's Discovery Mode, announced in November 2020, lets artists and labels mark songs for extra play in its Radio and Autoplay recommendations, in return for a lower royalty on those plays. Spotify now calls this a commission, and the Guardian put it at a 30% cut. In June 2021 US Representatives Jerrold Nadler and Hank Johnson warned Spotify of a "race to the bottom" in royalties, and Billboard now calls Discovery Mode "a popular and widely accepted industry marketing tool".

    Lessonthe 1960 law is written around radio stations; a listener sued in 2025 calling Discovery Mode payola, and in 2026 a federal judge ruled that the claim could not go ahead as a federal class action, so no court has decided whether it is payola.

    newsroom.spotify.comartists.spotify.comtheguardian.comhankjohnson.house.govbillboard.com

    Credit

    Spotify

  6. 2024

    Fake streams start costing money

    Artificial streams are plays made by bots, scripts or paid "farms" instead of real listeners, and they fake popularity on playlists and charts. A January 2023 study by France's national music centre (CNM) counted 1 to 3 billion fake streams in France in 2021, 1% to 3% of the market. Since 1 April 2024 Spotify has charged labels and distributors per track when it finds flagrant artificial streaming; the distributor TuneCore says the charge is €10 per track per month and passes it on to the uploader.

    LessonSpotify, Amazon, Deezer and the major labels signed an industry code against stream manipulation in June 2019, but Rolling Stone noted it was "not legally binding"; the fee puts a direct cost on whoever delivers the track.

    artists.spotify.comsupport.tunecore.comloudwire.comrollingstone.com

    Credit

    Spotify; CNM

  7. 2026

    Charts begin barring AI tracks

    By Billboard's count, at least six AI or AI-assisted acts debuted on its charts in late 2025, among them Xania Monet, created by the Mississippi songwriter Telisha "Nikki" Jones with the AI platform Suno, the first known AI act on a Billboard radio chart. In January 2026 IFPI Sweden kept an AI-made song with over 5 million Spotify streams off the official Swedish chart, and from the chart dated 31 August 2026 Australia's ARIA charts exclude wholly AI-generated tracks.

    LessonSweden caught its case because a journalist, Emanuel Karlsten, traced the song to a publisher's AI department, not by analysing the audio.

    billboard.substack.commusicradar.comaria.com.au

    Credit

    IFPI Sweden; ARIA

Watch for

  1. AI labels on chart data

    Luminate, which supplies Billboard's chart data, says its CONNECT platform can flag AI-generated songs and artists worldwide, that flagged items will carry an "AI Generated" label, and that artists and labels will have a review process for wrong tags. Billboard says it will identify songs on its charts that use AI.

    Achieved whenthe labels appear on chart entries, and an outside party, such as an artist, label or reporter, publishes how many tags were disputed or wrong.

    luminatedata.comca.billboard.com

    Claimed by

    Luminate and Billboard, August 2026

  2. A human-made test for charts

    IFPI says official charts will count a recording made with AI only if the AI service is authorised and lawful, the track is "substantially human made", and it raises no stream-manipulation concerns. It will apply this to the charts it runs in Latin America, the Middle East, Africa and Southeast Asia, and work to extend it to more than 20 other chart programmes.

    Achieved whena chart beyond Sweden's and Australia's excludes a named track under these rules and an outside outlet reports it, or Billboard adopts the test; Billboard's August 2026 statement commits only to identifying AI use.

    ifpi.org

    Claimed by

    IFPI, July 2026

  3. Most AI streams faked

    Deezer says up to 85% of streams on fully AI-generated tracks were fraudulent in 2025, against 8% across its whole catalogue, and that it removes those streams from royalty payments and keeps fully AI-generated tracks out of its recommendations and editorial playlists.

    Achieved whenan independent party, such as another service, an auditor or a public body such as France's CNM, publishes a comparable measure from its own data.

    newsroom-deezer.com

    Claimed by

    Deezer, January 2026

Open

  1. 01

    Test the playlist bot checkers

    An open, repeatable test of whether public "bot checker" ratings predict Spotify's later corrections. Recruit consenting independent artists and take each playlist that added their songs, as listed in their own Spotify for Artists data. On day 0, record two checkers' ratings and the song's public play count; on day 60, record whether the playlist still exists and whether the count was cut. Spotify says public counts "are always adjusted to remove confirmed artificial streaming".

    The brief
    Credit · unclaimed
  2. 02

    Open airplay monitor, measured

    An open-source monitor that listens to public internet radio streams, matches what it hears against audio fingerprints (compact summaries of a recording's sound) of a set of independent releases, and reports plays, with its detection rate measured against stations that publish what they play.

    The brief
    Credit · unclaimed
  3. 03

    Tally of paid-placement labels

    A dated public table showing, for each music service and short-video app, whether paid promotion is labelled at the moment a listener hears it, covering boosts such as Discovery Mode, sponsored recommendations, paid playlist slots and promoted clips, with the exact words used.

    The brief
    Credit · unclaimed

Open 01 · the brief

Test the playlist bot checkers

Credit · unclaimed

The first to publish is credited on this line, and the deliverable moves from Open to Achieved.

What to build

An open, repeatable test of whether public "bot checker" ratings predict Spotify's later corrections. Recruit consenting independent artists and take each playlist that added their songs, as listed in their own Spotify for Artists data. On day 0, record two checkers' ratings and the song's public play count; on day 60, record whether the playlist still exists and whether the count was cut. Spotify says public counts "are always adjusted to remove confirmed artificial streaming".

Why it is still open

Artist.Tools' 2026 report covered 7.3 million playlists, but it says its "exact thresholds, weights, and evasion-sensitive logic remain private", that it cannot tell whether a playlist committed fraud, and that it has none of Spotify's enforcement data. No published test of any checker against Spotify's own corrections was found.

artist.toolssupport.spotify.com

Built with what exists

Spotify for Artists (each artist's own data); the free Artist.Tools Bot Checker and the IsItAGoodPlaylist analyzer, used by hand; the public Spotify app for play counts; a spreadsheet. Spotify's rules limit automation: since February 2026 new developer apps in development mode are capped at 5 users and cannot read the tracks on playlists they do not own, and the Developer Terms forbid building "compilations or databases of Spotify Content".

developer.spotify.comdeveloper.spotify.com

  1. First step

    Recruit five artists and log every playlist add in their Spotify for Artists, with checker ratings and today's play counts.

  2. Done looks like

    At least 200 playlist placements from at least 20 artists, each with two day-0 ratings and a day-60 outcome; published hit rates (precision and recall) for each checker with error bars; an anonymised table others can re-run. No named playlist is called fraudulent.

Open 02 · the brief

Open airplay monitor, measured

Credit · unclaimed

The first to publish is credited on this line, and the deliverable moves from Open to Achieved.

What to build

An open-source monitor that listens to public internet radio streams, matches what it hears against audio fingerprints (compact summaries of a recording's sound) of a set of independent releases, and reports plays, with its detection rate measured against stations that publish what they play.

Why it is still open

WARM, a paid service, says it monitors "21,000+ radio stations" in "160+ countries", but its site publishes no detection rate. The nearest open project, a broadcast advertisement detector, reports perfect scores from synthetic tests and warns they are "not a real-world accuracy guarantee". No open music airplay monitor with a real-world detection rate was found.

warmmusic.netgithub.com

Built with what exists

The Radio Browser API (free and open source; it listed 53,032 stations in 224 countries on 19 November 2026); FFmpeg to decode streams; the open fingerprinting libraries Olaf or Panako. Olaf runs even on small embedded chips, which makes a cheap always-on listener possible, but its authors warn of two US patents, so check them before any public release. Listen as any listener does, keep fingerprints rather than audio, and check each station's terms.

api.radio-browser.infogithub.com

  1. First step

    Pick 20 stations from Radio Browser that publish now-playing lists, fingerprint 50 songs they play often, run for 48 hours, and tally matches against their lists.

  2. Done looks like

    20 consenting releases, 200 stations, 14 days; a published share of station-listed plays the monitor caught, and false matches per 1,000 station-hours. If an artist already pays for WARM, compare the same plays.

Open 03 · the brief

Tally of paid-placement labels

Credit · unclaimed

The first to publish is credited on this line, and the deliverable moves from Open to Achieved.

What to build

A dated public table showing, for each music service and short-video app, whether paid promotion is labelled at the moment a listener hears it, covering boosts such as Discovery Mode, sponsored recommendations, paid playlist slots and promoted clips, with the exact words used.

Why it is still open

Spotify says only in general terms, on its "Understanding recommendations" page, that "commercial considerations" may influence recommendations. The 2025 lawsuit argued that "users cannot distinguish between genuine personalization and covert advertising", and the court did not decide that point. No regulator ruling and no tally across services was found.

spotify.combillboard.com

Built with what exists

Two test listener accounts per service; screen recordings; each service's help and terms pages; Claude to pull disclosure clauses out of long terms pages (see the Note "Seven more flagged"); Gemini Decision 1 to sort hundreds of screenshots by whether a label is shown (see the Note "A thousand questions, one pass"). A person checks every row.

  1. First step

    Start with Spotify. List the paid tools on its artist pages, such as Discovery Mode and Marquee, then log 100 Radio and Autoplay recommendations in a listener account and note any label.

  2. Done looks like

    At least 8 services and 20 paid products, each row recording yes or no for "labelled at point of play" and "explained in help pages", the exact wording, the date and a screenshot; a second person checks every row, and the tally is repeated after three months.