A month after release the typical album does 39% of its first week, a year in 11%, ten years in 2.4%. How rare the exceptions are says a lot about quality.
Research, October 29, 2026, 9 min read
A month after release, the typical album is down to 39% of its first week. A year in, 11%. Ten years in, 2.4%.
We tracked 126,342 albums on Spotify every day for nine months. Almost all of them follow that curve. The few that break it, and how rare they are, are the closest thing streaming has to a measure of quality.
The short version
How much of its first week a typical album still does, and how many albums break the rule at each age.
39% of its first week's daily streams. Each dot is a 2026 album: its streams on day 30 against its own first week.
287 albums
11% of its first week's daily streams, estimated. Each dot is an album about a year old in January: how its streams changed by October.
409 albums
2.4% of its first week's daily streams, estimated. Each dot is an album about ten years old in January: how its streams changed by October.
2,696 albums
The rule is steep at first and then flattens. A month in, the typical album has lost three-fifths of its first week. By a year it's down to about a tenth, and still losing a fifth of what's left each year. By ten years it's at 2.4% and barely moving: the people still playing it are the ones who chose it.
Breaking the rule is rare, and it gets rarer the bigger the break. A month in, 1 in 26 albums is still at its first-week level. Among albums a year old, about 1 in 4.5 grew over the nine months we watched, but only 1 in 68 doubled. At ten years growing is more common, partly because Spotify itself keeps growing, yet doubling stays at about 1 in 100.
The first six months of every big 2026 launch, and how long each halving takes after that.
Here is the rule in detail. We took every album released in 2026 that opened with at least 3 million streams in its first week (355 of them) and compared every later day with its own first week.
100% is the album's own average day in its first week. Day 1 is the first day Spotify's count moves, usually two days after release.
The first week holds steady. Then the new-music playlists move on and the typical album is down to half by day 21. After that it keeps losing listeners, but more and more slowly: each halving takes longer than the one before, until after about fifteen years it stops falling at all.
Each bar is how long a typical album takes to lose half of what it has left, drawn to scale. Lighter bars are estimated from older albums.
That slowing is the exceptions showing up in the average. The people who were only there for the launch leave fast. The ones who are left chose the album, and they leave slowly.
Six 2026 albums: daily streams since release, each line on its own scale.
If launch size measured quality, the biggest launches would hold up best. Drake's ICEMAN opened at 65 million streams a day, the second-biggest launch of the year. By day 165 it was down to 8% of that. BTS opened even bigger and kept twice as much. Sienna Spiro's Visitor opened at a sixteenth of Drake's size and was doing almost twice its own first week three months later, as people found its songs on their own.
Is that the rule? We checked all 241 big launches of 2026.
Each dot is one 2026 album. Across: how big it opened, in streams a day. Up: how much of that it still did on day 120. The dashed line follows the middle.
No. If size predicted staying power, the dots would slope. They're a cloud. The biggest launches start with a sharper first day and fall harder in the first month, then catch up. By day 120, albums of every size keep between 16% and 27% of their opening.
A launch measures how many people were told about an album. Marketing, playlist slots and a loyal fanbase all show up here. What it can't tell you is whether people will want to hear it again, and that's the part that looks like quality.
Six albums that came back in 2026: daily streams, a week's average. The dotted level is each album's January pace.
Old albums come back all the time, and something outside the music always starts it. What happens after is the useful part. Look at where each line ends.
A moment borrows attention. Bad Bunny's Super Bowl halftime show took DeBÍ TiRAR MáS FOToS from 17 million streams a day to 64 million. By October it was 42% below where it started the year. Millions of people pressed play once, for the occasion, and went back to what they were listening to before.
A discovery brings in people who stay. The Michael biopic took Thriller from 3.5 million streams a day to 19 million, and six months later it still runs at twice its January level. A Netflix film put The Chemical Brothers' Go in front of a new audience, and it's still at 11 times its January level. Some of the people who heard these for the first time liked them enough to keep playing them.
That's replay value working as a test. A spike is attention, and attention can be bought or scheduled. What's left after the spike is choice: the new listeners who decided the album was worth hearing again.
Each album scored against albums of the same age. 100 is a typical album its age; 110 means it held on to 10% more of its audience.
Latin albums hold on to their listeners best, followed by country and rock. Jazz and classical lose theirs fastest, probably because a lot of what Spotify files under them is background music on mood playlists, and those playlists rotate. Greatest-hits albums hold up best of all: they're made of songs people already chose. And the bigger an album already is, the better it holds. On streaming, popularity compounds.
This covers albums from two to fifteen years old with at least 50 million streams.
So far this has been about how albums age. Here's why the exceptions matter. Ask ten people whether an album is good and you get ten answers. Critics disagree, awards follow fashion, and the numbers everyone quotes (first-week sales, debut chart position, launch-day streams) mostly measure how much was spent telling people the album exists.
A number that tracks quality would need to pass three tests.
Replay value passes all three. Long after the marketing stops and the new-music playlists move on, people either keep pressing play or they don't. Spotify counts every one of those plays, for every album, every day, and we record those counts.
Here's the measure we use. Take an album's average day over its whole life so far, and compare it with an average day now. An album at 200% is being played twice as much today as it was on an average day of its life, including the launch. One at 20% has lost four-fifths of its audience.
Daily streams. Both albums have more than a billion streams in all.
One correction matters. Spotify has grown a lot since 2014, which flatters older albums: the typical billion-stream album from 2014 runs at 153% of its lifetime average today, and the typical one from 2021 at 57%. So every album is ranked only against albums released the same year.
Billion-stream albums from 2014 to 2021: today's daily streams against the album's own average day since release, ranked within each release year.
The most replayed albums are mostly slow builders: Adrianne Lenker, Mac DeMarco, Videoclub, AURORA, the Texas country band Treaty Oak Revival, a set of Dominic Fike demos. Few of them were big launches. People found them over years and kept them.
The least replayed are mostly rap albums from 2015 to 2018 that were huge when they came out. Chance the Rapper's Coloring Book won the Grammy for Best Rap Album. It now streams at 25% of its lifetime average. Lil Pump is at 17%.
That's the first sign the number is measuring something real and not just popularity: it disagrees with the charts and the awards, in a direction you can explain.
Replay value isn't quality itself, and it fails in ways worth knowing:
A launch measures how many people were told. Replay value measures how many people cared.
Even with those holes, nothing else comes as close. It costs nothing to collect, it covers every album on the largest music service, it can't be bought the way a launch can, and it gets harder to fool the longer you wait. If you want one number for whether an album is good, the best one available is whether people are still choosing it years later.
For artists and labels, that changes what to look at. The first three weeks are half the story of a launch, but they aren't the asset. The asset is the tail, and the tail depends on whether people come back on their own once the playlists move on.