Netflix's "weeks in top 10" is not weeks in a row

Netflix’s weekly top 10 files carry a field on every row: how many weeks that title has been in that country’s top 10. It appears in press write-ups as a streak — twelve straight weeks in the top 10.

It is not a streak. It is a cumulative count, and the difference is not small.

What it does on consecutive weeks

We took every row of the per-country files — 500,340 of them, across 269 weeks and 93 markets — and grouped them into cells: one title, with its season label, on one chart, in one country. That gives 191,537 cells. Then we looked at every pair of rows in a cell that are one week apart.

There are 287,291 such pairs, and in every single one the counter goes up by exactly one. Not 99.8%. All of them.

So on consecutive weeks the field behaves exactly as you would expect, which is why the streak reading survives.

What it does after a gap

Then we looked at the pairs that are not one week apart — a title left the chart and came back. There are 21,512 of those.

Ninety-eight per cent of the time it keeps counting. A title that charted for six weeks, vanished for four months and came back reappears at week seven — not week one.

Among Korean titles specifically, 1,250 of 1,314 post-gap pairs continue (95.1%) and 64 restart.

Why this matters if you quote the number

Three things follow, and each one changes a sentence people write.

• “25 weeks in the top 10” can be 25 weeks spread over two years. The title may have been absent for most of the period between its first week and its last. The largest counter we found anywhere in the file is 133; the largest on a Korean title is 72.
• It is per country and per chart. All 191,537 cells begin at 1, so the number is not a world tally that arrives in each market — Vietnam’s counter for a title and Malaysia’s are independent.
• Seasons are counted separately, because Netflix lists them separately. A show in its second season starts again at 1 even though the show never left.

This is why every run length on this site is computed from the weeks themselves rather than read off the field. When we say a Korean series holds a chart for a median of three weeks, that is three consecutive weeks, measured by us. The two numbers are not interchangeable, and a comparison that mixes them will make Korean titles look either much stickier or much less sticky than they are, depending on which side of the comparison used which.

The 424 we cannot explain

Two per cent of post-gap pairs do restart at 1. We do not know why.

Netflix publishes no note about this field, and nothing else in those rows — the season label, the position, the gap length — separates them from the 21,072 that continue. It may be a different ingestion path, a re-release, or a change in how a title is identified upstream. All three would look identical here.

We are reporting the split rather than picking the explanation that reads best, and we are not rounding the 2% away: if you are building anything on this field, 424 cells behave differently and you cannot tell which ones in advance.

Every count above, with the row totals behind it, is on the weeks-counter page.