Netflix reaches 32% of Korean series, 15% of the films

Every figure we publish about Korean titles on Netflix starts from the same place: 421 titles that reached a Top 10 in six Southeast Asian markets. Until today we had never said what that is 421 out of.

A share without a denominator is not a share. So we counted one.

What the charts touch

Roughly three in ten Korean series made since the charts began have appeared on one in Southeast Asia. For films it is about one and a half in ten.

Against the whole recorded catalogue the shares are 5.2% and 1.7%, but those numbers describe an archive rather than a market: most of what Wikidata holds was made long before Netflix published a weekly list, and a 1979 film had no chance to appear on one.

The same asymmetry, again

A Korean series is twice as likely to reach a chart as a Korean film. That is not a new finding here so much as the same finding arriving from a new direction. We have already measured that the median Korean series charts in 35 countries and the median film in 11, and that films are far more likely to carry a name we cannot verify.

Now we can add the step before both: fewer films get onto the chart at all. Whatever is happening to Korean film abroad is happening at the door, not only after it.

We are not going to tell you why. Volume of Korean film output includes a long tail of small releases that were never licensed to a streaming service in Southeast Asia, and nothing in these counts separates that tail from a film that was available and simply not watched. Settling it needs a distribution record we do not have.

How we found our own mistake

The first version of this count asked Wikidata for a publication date (P577) and got 84 Korean series from the chart era.

Our panel holds 237.

A coverage share cannot exceed 100%, so the error was in the denominator, not the panel: Korean television series overwhelmingly carry start time (P580) rather than publication date. Counting both fields raises the chart-era series from 84 to 782 and the ones with any date at all from 210 to 3,792.

The collector now refuses to emit a share above 100% and writes an empty cell instead. The impossible number was the only reason we looked, and if it had been merely implausible rather than impossible we would probably have published it.

What this does not mean

Wikidata is a record of what somebody wrote down, not a census of what was made. Its 10,233 films are a floor. If Korean cinema produced more than that — and it almost certainly did, once you count releases nobody catalogued — then our 14.3% is an upper bound that gets smaller, never larger.

We would rather hand you a ceiling and say so than hand you a number and let you assume it is a count.

What changed on 3 September 2026

The panel is 420 titles, not 421. This piece said 421. We removed Dangerous Liaisons — the 2022 French film — on 25 August after confirming against Wikidata that no Korean work carries that name, and one of the three titles we removed that day sat inside this panel.

We also re-ran the Wikidata denominator query the same day. It now records 785 Korean series and 1,126 Korean films from the chart era, against 784 and 1,122 before — Wikidata gains entries continuously, so this will keep moving. Films that reached a Top 10 went from 173 to 172 and the film share from 15.4% to 15.3%.

The finding did not change. Roughly three in ten Korean series made since the charts began have appeared on one in Southeast Asia, and about one and a half in ten films. Series are still about twice as likely to be seen at all.

What changed on 3 September 2026

We added the chart week of 23 August 2026, which had been in our archive while this panel still stopped at 16 August. One series entered the panel with it, and re-querying Wikidata on the same day returned six more recorded Korean series.

The point of the piece is unaffected and so is its size: series are still about twice as likely to be seen at all as films, 31.7% against 15.3%. And the ceiling argument still holds — these are upper bounds that shrink as Wikidata records more, not measurements of a market.