Methodology — The Restaurant Index

Source: Commercial District Information (Small Enterprise and Market Service) — Small Enterprise and Market Service (소상공인시장진흥공단), taken from its open API on Korea’s public data portal (dataset 15012005). Licence: no restriction on use, as stated on the dataset page. Period: Q2 2026 (as of 30 June 2026). Retrieved: 2026-09-10. Updated quarterly.
We did not choose the categories: The items are every mid-level category under the registry’s top-level class I2 (food), read from the source’s own classification endpoint. That it comes to 12 is a result, not a target. No judgement about what belongs on the list enters this index.
Formula: The count of premises, as published. No weighting and no adjustment. The percentage beside each row is its share of the total. As a check, the category counts must sum exactly to the published class total (840,255), and the build fails if they do not.
Zeros stay in the ranking: Where a category has no premises, the source API answers “no data”. That is a zero, not an error, so we publish it as zero. Keeping only the rows that happen to have numbers would be choosing the ranking after seeing it, which is the thing this site exists not to do.
The English category names are our translation: The source publishes Korean category names only. The English labels here are ours; the codes (I201 and so on) are what you check against the original. The Korean name is the original.

Limits worth knowing before you cite this

This is a registry, not a census. The data is a commercial-district registry, and the source API describes itself as covering premises within major commercial districts. Anything the registry misses is missing here too.
A count of premises is not a measure of size. The category at the top is not necessarily the largest by revenue or by seats. What is counted is signboards.
The category boundaries are not the ones you would draw. That is why a broad bucket like “other quick-service” ranks high. Regrouping the categories would change the order, and the moment we regrouped them the ranking would be our opinion, so we do not.
It is a single snapshot. The registry is refreshed quarterly, and we do not place figures from different quarters side by side.