Streams count plays. We count who can name it.
Every day, Muzify players hear a clip and race a five-second clock to pick the title. quizzes later, this is what people actually know about the music they stream.
All of it in the 44 days to , across 545K songs played — of which 4,539 clear 200 answers and get published individually. Four titles per round, so 25% is chance; no published song scores at or below that line. A quiz counts as verified only when all ten rounds were answered. The lifetime figure above is Google Analytics' count of round-ten answers since December 2024 — everything else here is measured from our own event log.
% of players picked “” correctly — but only % did it inside four seconds. A song streamed this much should manage %. answers · 4 options, chance = 25%
Every song, ranked by its own fans
Pick an artist. Every measured song they have, ranked by the share of that artist's own players who named it correctly in about four seconds. The thin line is the 95% range.
Streams and memory disagree
Left: how many players should name a song quickly, judging by how much it is streamed. Right: how many actually did. Lines that climb are songs people know better than their streams suggest.
Known beyond its streams
Fans carry these in their heads. Cult favourites, deep cuts and scene anthems.
Bladee, “Amygdala”
Streamed beyond recognition
Big streaming numbers, yet players can't place the song when it plays.
Michael Jackson, “You Are Not Alone”
Every country has its own obsession
How many times more a country plays an artist than that artist's global share predicts. At least 20 different players stand behind every number, so one superfan can't move the map.
Who people are racing to play
Week-on-week change in quiz demand. Growth is damped and has a volume floor, so a handful of plays can't top the board.
If you play this, you play that
Artists whose players overlap the most, learned from who plays whom. Useful for support slots, playlist seeding and ad audiences.
Ten rounds. How fast are you?
Every quiz is ten rounds on one artist. This is how quickly players get all ten right — and where your own run would sit.
The full index, every song
- Song scoresRecognition and speed with intervals for every measured song
- Country gridEvery country × artist cell, refreshed nightly
- Early risersArtists moving below the public volume floor
- Taste graphNearest artists for any act in the catalogue
Faster than the fans?
Pick an artist. Hear the clip. Beat the clock. Every answer goes into next month's index.
Play on muzify.comWho plays, and how
Players choose the artist inside the Muzify app, so most are existing fans, not a random public sample. Acquisition is organic installs and repeat play — no paid audience-buying shapes who answers.
Does replaying teach the song?
Yes, and it isn't corrected for yet. A returning player can meet the same song again in a later session and score higher the second time. First-exposure recognition isn't isolated in this measurement — a named limitation, not hidden.
Where do the clips come from?
Apple Music 30-second previews, the same clip length across the whole catalogue. Never used as report artwork.
Is this scored on chance?
Four options per round, so 25% is what a coin flip would score. Every published figure sits above that line — none of Muzify's reported numbers score at or below chance.
Does timing include load time?
Yes. The ~4-second "quick" threshold is measured from the start of the round, clip loading included — not from when audio starts.
Where's the streaming baseline from?
Spotify popularity (via song_meta) fit with a logistic model against recognition. "Expected" is that fit's prediction; the residual (z-score) is what ranks the streams-vs-memory chapter.
What drives weekly spikes?
Not yet attributed. We can see week-to-week demand swings in the rising chapter but don't yet know whether a release, playlist placement or promotion caused any one spike — an open limitation, not glossed over.
Privacy
Sessions are salted hashes. No one is identifiable, and every published cell has a minimum player count.