You cannot ask what a Korean hit leaves behind. The article is born with the show

The article is born with the show

A percentage change from nothing is not a small number; it is not a number. These titles have no floor from before the wave because they had no article before the wave.

This is a limit of the question, not of the window. Widening the window does not create a floor that never existed. Anything that arrives and creates its own article is unmeasurable here permanently.

The 9 titles that already had a floor

Reads are per million reads of that Wikipedia, summed across Indonesia, Vietnam, Thailand, Malaysia. The floor is the 6 months before the peak and the 6 after, skipping 2 months either side because those are still the wave. Window — 2020-08 through 2026-06, 71 months.

Median peak 2.2 times the surrounding months; median floor afterwards −4.5%. 3 of 9 ended higher. We report the median rather than the mean because with 9 titles one of them moves the mean on its own — the mean reads −9.4%.

These are not hits arriving. They are back catalogue rising again, which is the only kind of title that can answer this question at all.

⚠ Do not lean on that median. Remove any single title from the 9 above and it moves to somewhere between −15.1% and −2.3% — a swing of 2.86 times its own size. removing a single item moves the answer by a large fraction of itself, which means the sample is not yet large enough for this median to be reported as a finding — not that the figure is wrong. We ran this check on this page after correcting it twice in one day; the method is here.

Everything we could not use

The largest peak-to-floor ratio in the whole dataset belongs to Physical: 100 at 15.2× in 2024-03, and we cannot report it. Its floor before the wave was 1.3 per million — a rise between two quantities that are both, in practice, nobody.

Each title rose in its own month

Each title peaked in its own month, which is what measuring titles rather than months requires.

The peaks — 2021-08 · 2021-10 · 2022-09 · 2024-01 · 2024-05 · 2025-02 · 2025-04 · 2025-07. An earlier version of this page, built on a shorter window, had 9 peaks landing in three clusters of neighbouring months, which meant we might have been measuring those months rather than those titles. That objection is answered.

What this page cannot tell you

We measured 9 titles. That is a small number and it is the number we have: of the twenty we fetched, some have no article in these editions, some peaked too near the edge of the window to have a floor on both sides, one had a second wave sitting where its floor should be, and one had a floor too thin to take a percentage. Every one of them is listed with its reason.

The titles are ones we expected to have had a wave, not the most-read Korean titles and not a random sample. A set chosen this way leans toward titles that travelled.

A read is not a viewer. Someone opening an encyclopaedia article about a series may have watched it, may be deciding whether to, or may have seen the name somewhere and been curious. What this dates is when the looking happened.

We are not saying the wave caused the floor to move, in either direction. Twelve months either side of a peak contain everything else that happened in those months too.

⚠ 2026-07 is excluded. It arrived at 3919 against a mean of 119640.7 over the twelve months before it — Wikimedia's monthly totals were still filling when we fetched, and reading that month as a collapse in attention would have been wrong.

Reads — Wikimedia Pageviews API, human traffic only, monthly.