flow. Concentration Index

The share of Korean top 100 album sales taken by the ten biggest records of the year, with every edition of a record counted as one record.

Computed and maintained by Richard Dufaur, who writes flow., a weekly letter on the business of the Korean music industry. Short name for this index: FCI.

What the number says

When the index goes up, the same chart is selling more of fewer records. When it goes down, the middle of the chart is holding more of the volume.

The top gave up 6 points from 2023 to 2025, then took 5 of them back by 2026. It did that in the most crowded year the window has ever measured: 352 acts, more than any year before it.

How it is built

Four steps. Anyone with the Circle album chart can repeat them.

1. The window. Weeks 10 to 41 of each year, the same weeks every year. Week 41 is the last complete week available, and the window extends by one week each time Circle publishes another. Week 10 is where my sales figures begin: weeks 1 to 9 are present in my base as chart positions without sales attached, in every year alike, so they sit outside the count rather than being excluded from it. Nothing in the method depends on those particular weeks, only on their being identical from one year to the next.
2. The base. Every album row on the Circle weekly top 100 inside the window, weekly sales, not cumulative totals. Summed over the window, this is the denominator.
3. The consolidation. This is the step that changes the answer, and it is the reason the index exists. Circle gives a chart line to each edition of a record. A record released as a standard version, a Weverse version, an LP and a POCA occupies four chart lines, each carrying only its own share of the sales, so read straight off the chart that record looks four times smaller than it is. The effect is not stable over time either: the sales sitting on a secondary edition of a record went from 4% of the window's top 100 in 2022 to 17% in 2026, so the raw chart understates the top by a little then and by a lot now. I merge editions back into the record they came from before counting. A parenthetical suffix on a title is treated as an edition marker, applied twice so that a title like ARIRANG (Deluxe Ver.) (LP) returns to ARIRANG. Records are keyed by artist and consolidated title.
4. The count. Rank the consolidated records by sales over the window, take the top ten, divide by the window total.

What it is not

• Not a measure of popularity. It measures physical album sales in Korea and nothing else. No streaming, no downloads, no export figures.
• Not the share of chart ranks 1 to 10. That number is a different thing, it sits between 84% and 86%, and it barely moves. This index counts the ten biggest records of the period, not the ten highest rows of a week.
• It does not go back before 2022. My Circle sales figures start there. Any claim of the form "the highest since" stops at 2022.
• It says nothing about revenue. A record that sells twice as many copies at half the price is invisible here.
• The consolidation rule is a judgment. A record whose editions carry no parenthetical suffix will not merge. I publish the rule so that anyone can see where my number and theirs would part.

Source

Underlying chart: Circle Chart, weekly album chart, circlechart.kr. Circle publishes the chart; the index is my computation on top of it, and Circle bears no responsibility for it.

I publish the source and the calculation, never my collection. A scraper is not a source. The Circle rows themselves are not republished here: what is published is the derived index.

Citing it

flow. Concentration Index (FCI), weeks 10 to 41, 2026: 38.8% — Richard Dufaur, flow., 2026-10-19, flow.databyric.fr/method

The index is recomputed and republished as the window extends. Corrections are stated in the issue that carries them, and this page carries the current value. Link back to flow.databyric.fr/method rather than this domain directly: the shortlink always resolves to the current page even if the address behind it changes.