Korean pop culture, in numbers

How long a K-pop group stays looked up after its peak

Every Korean act has one month when more people looked it up than in any other. We found that month for 774 of them and counted what was left afterwards. The median act fell below half of its best month within 2 months.

The whole set

Korean acts with an English Wikipedia article847
Measured here (at least 24 months of data)774
Dropped for too few months — not counted as zero73
Have fallen below half their peak month731
Median months from peak to half2
Have not fallen below half yet43
…of which peaked this month, so had no time to fall8

That last line matters. An act whose best month is the most recent month cannot have declined yet, and counting it as durable would be wrong. Only 12 acts have gone twelve months or more past their peak and still hold half of it.

Held on longest — acts whose peak month drew 50,000 or more

Restricted to the 100 acts with a peak of at least 50,000 reads, so that a small article with noisy months cannot lead the table.

ActMonths from peak to halfPeak monthPeak readsLeft today
Stray Kids652020-09155,81951.5%
Infinite262015-0785,0794%
VIXX212015-1163,3506%
Mamamoo192019-12103,45022.1%
Seventeen142023-05156,28730.6%
Red Velvet142020-07125,02427.4%
Day6142017-0655,08017.9%
BTS132021-061,010,08325.7%
Exo132023-06120,73421.4%
Monsta X132017-0495,49415.3%
Girl's Day122015-0763,89413.5%
NCT 127112020-0366,27046.4%

The gap between first and second place is the finding here, not the ranking itself.

The largest peaks, and what is left of them

ActPeak monthPeak readsLatest monthLeftPeak to half
SuperM2019-101,083,6884,8160.4%1 mo
BTS2021-061,010,083260,08525.7%13 mo
Blackpink2020-10889,189140,88115.8%1 mo
Shinee2017-12694,34917,6262.5%1 mo
Treasure2020-12481,4858,5131.8%1 mo
Astro2023-04438,43110,4032.4%1 mo
Enhypen2026-03395,70454,31913.7%1 mo
Girls' Generation2018-01318,40045,76414.4%1 mo
BigBang2025-01288,93670,13424.3%1 mo
Wanna One2017-08280,28815,7015.6%6 mo
Got72021-01271,42213,8645.1%1 mo
Super Junior2017-11267,83616,6966.2%2 mo

A very large peak does not buy a slow decline. Several of the biggest months in this table were an act's debut or a single event, and the month after was already half.

Past their peak by a year or more, and still above half

ActPeak monthMonths sinceLeft
Sinawe2020-057553.2%
Can2022-104652.3%
TRPP2024-032959.6%
Mot2024-032953.7%
Target2024-062662.3%
Achime2024-102266.7%
The Freaks2025-011955.7%
Broccoli, You Too?2025-031762.3%
Seaweed Mustache2025-031761.9%
Rolling Quartz2025-031757.5%
BOL42025-081280.9%
3rd Line Butterfly2025-081260.2%

Most of this list is not idol pop. Several are rock and indie acts whose readership was never large enough to spike, and a flat line never has a peak to fall from.

When the peak arrives

Our reads begin in July 2015, so for an act that debuted earlier we cannot know whether its real peak came before we were looking. This table therefore uses only the 183 acts that debuted in 2016 or later, whose whole career sits inside the window. The other 350 acts with a known debut year are left out of this question rather than given a peak we cannot verify.

Gap between debut year and peak monthActs
Peaked in the debut year itself35
One to two years later72
Three to five years later56
Six years or more20

The median gap is two years. Fewer than one act in six had its biggest month in the year it debuted, and the tail runs long — some acts do not reach their best month for six years or more.

The fall has not changed in thirty years

Debut yearsActsMedian months from peak to half
1990–1994132
1995–1999442
2000–2004422
2005–2009672
2010–20141583
2015–20191452
2020–2024642

For acts that debuted before 2016 this is the fall from their highest monthinside our window, which need not be their career high — but the descent is measured the same way for every act, and it does not vary by generation.

What this is not

This is not a popularity ranking. It counts what people looked up, not what they listened to.

Acts we could not measure are counted separately, never filled in as zero.

A Wikipedia article grows more detailed after an act becomes known, so early months read low partly for that reason.

The current month is excluded because it is not finished.

(G)I-DLE has no country property on its Wikidata item and so is absent from this table. That is a miss on our side, not a zero.

One rupture, eight groups — was the drop only theirs?

A Korean desk called it one group’s lost two years. The same twelve-month before-and-after can be run on every fourth-generation girl group, and it is the only way to know whether a fall belongs to the group or to the generation.

GroupReads, 12 months beforeReads, 12 months afterAfter ÷ before
aespa664,332604,1150.91×
NMIXX282,600256,7390.91×
LE SSERAFIM1,218,745845,3400.69×
IVE608,289403,2380.66×
STAYC147,69892,2350.62×
NewJeans (the one the desk wrote about)1,443,607888,3680.62×
ITZY514,603258,8870.50×
(G)I-DLE727,616342,3490.47×
Groups counted8

The median of the other seven is 0.66×, and 2 groups fell further than the one the desk named. ⚠ The window moves as months are added: when we first published this the named group read 0.61× and it now reads 0.62×. Neither is a correction of the other — each is the same measurement on a window one month apart, which is the whole reason we print the window with the number. The full piece is at the lost two years happened to everyone.

Where else this leads

Korean titles lose half their reading in about the same two months — we measured that separately, on a different unit, and the timescale came out the same. Comics run the other way: a Korean webtoon is read 14 times more once a drama exists, so an adaptation builds attention where a comeback mostly spends it.

What we wrote from this data

Sources

Acts and debut years — Wikidata SPARQL (P31/P279* Q215380 musical group, P495 country of origin Q884). Reads — Wikimedia Pageviews REST API, English Wikipedia, all-access, agent=user (bots excluded), monthly, 2015-07 ~ 2026-08. The current month is excluded because it is not finished. Collected 7 September 2026, 10:15 KST.