New K-pop groups look 69% down since 2017. So do American, Japanese and British ones.

Here is a question we get asked, and it looks easy: how many new K-pop groups debut each year?

Wikidata answers it in one query. Korean musical groups, with a recorded formation date, counted by year. It returns 691 groups and a clean shape: a rise through the 2010s, a peak, then a fall.

Read alone, that is a story. New Korean groups down by more than two-thirds in six years, right as the industry’s global profile was supposedly at its highest. We could have published it this morning.

Then we ran the same query for three other countries

Every one of them falls. American group formation does not collapse by 78% in the same six years that Korean group formation collapses by 69% and British by 78%, for reasons internal to four separate music industries. What fell is not the number of groups. It is how quickly this source records them.

A group formed in 2024 does not get a Wikidata item the week it debuts. Someone writes an article, someone else creates the item, someone else adds a formation date. The lag is invisible in any single country’s chart, and it looks exactly like a decline.

Why we are publishing the failure instead of the finding

We nearly ran the first table on its own. The reason we did not is that the control took eleven minutes and the article would have been wrong for years.

That is the whole method here, and it is not specific to K-pop: when a count comes from a source that people maintain by hand, ask what the same count does somewhere the story should not apply. If the shape survives, it is about the subject. If the shape appears everywhere, it is about the source.

We are keeping the query and the numbers in our data files, marked unusable as a trend, because someone will run it again — possibly us — and get the same clean-looking line.

It is the same discipline behind our K-pop attention page, where the counts come from Wikipedia traffic that is recorded the day it happens rather than whenever an editor gets to it, and behind how we build every number here.

The other thing this source cannot tell you

While we were in there we tried a second question: how long does a Korean group last? There is a widely repeated idea that seven years is the natural span, because that is the length of a standard Korean artist contract.

Of the 691 Korean groups with a recorded formation date, 67 have a recorded dissolution date — 9.7%. Among those the median span is 3 years, and there is no bump at seven: 7 groups ended at seven years against 6 at six.

We are not reporting that as evidence against the seven-year idea, because a 9.7% sample of endings is not a sample of endings at all. Groups that quietly stop are far less likely to have someone record a dissolution date than groups that announce a split. The number we can produce is therefore biased in a direction we cannot size, and the honest output is this paragraph rather than a chart.

What would answer either question

For debuts, a source that records the event at the time it happens — a national registry, a chart body, or a broadcaster’s own listings. For endings, the same. Wikidata is excellent at telling you what exists and poor at telling you when you are looking at all of it.

If a reader knows of a Korean source that publishes group debuts by year with a fixed collection date, we would like to hear about it: parkintaek2@gmail.com. We would rather run the count properly than not run it.