Korean pop culture, in numbers

Two shows from one Korean company on the same chart cost each other half a place

A country's Netflix chart has ten places on it. A company with two shows out at once is using two of them. Whether that is a problem is a scheduling question with money attached, and it is answerable from the weekly lists.

When two series from the same Korean company are on a country's chart in the same week, do they sit lower than when either is there alone?

The answer is half a place

How this is counted. One pair = one series in one country at one age in weeks, with its average rank on the weeks it was alone set against its average rank on the weeks a series from the same company was also on that chart. Rank 1 is the top of the chart, so a higher number is worse.

A Korean series on a country's chartAverage rank
On the weeks it was there alone5.21
On the weeks a series from the same company was also there5.71
Difference0.5

0.5 of one place, on a chart with ten of them. It is in the direction a scheduler would fear — a title sits lower when its stablemate is beside it — and it is small. A company weighing whether to hold a title back for a month is weighing about half a rank.

Why every comparison is inside one title, at one age

A title's rank falls as its run goes on, and a company's second title usually arrives later. Comparing without holding age fixed would credit the decline to the sibling instead of to time. Every comparison here is between weeks at the same "weeks in top 10" value.

Age of the titlePairsAloneWith a siblingDifference
Weeks 1–22725.15.570.47
Weeks 3–41905.045.170.13
Weeks 5–82605.316.070.76
Week 9 and later815.696.290.6

All four bands point the same way, and the last column never reaches a whole place. The comparison never leaves a single series in a single country, so the title's own strength, its company and that country's appetite for Korean shows are all fixed and cannot produce the gap.

What we threw away, and the version of this that was wrong

Weeks of charts265
Chart rows read493,600
Korean series rows with a company attached25,598
Of those, rows with a sibling on the same chart5,431
Weeks discarded — no week of the same age to compare against22,529
Pairs that survived803

Far more weeks were discarded than kept. A week only counts if the same series, in the same country, at a comparable age, also spent a week without a sibling. Most runs never overlap a stablemate at all, and those contribute nothing.

The first version of this measurement was wrong, and it is worth saying how. We began by matching weeks at exactly the same age — week 4 against week 4. That left 205 pairs, because a title never reaches the same week number twice in one country: the only matches were runs where Netflix's week counter had reset. The sample was not small, it was strange. Matching inside age bands instead gives 803 pairs across 648 runs, at the cost of allowing up to a week of slack inside a band. That cost is stated here rather than hidden.

What we wrote from this data

Where these numbers come from, and what they cannot answer

Does an earlier hit help the next one · How many companies fill half the shelf · How we count