Not a list of links. Every city below is counted with the same instrument, on a stated date, from a source anyone can open. Where a city has not been measured yet, it is absent rather than guessed.
| City | Shops mapped | Stating a street | Share that does | Boundary used | Measured |
|---|---|---|---|---|---|
| Dubai | 667 | 518 | 78% | administrative boundary, no admin_level | 2026-09-26 |
| Taipei | 413 | 256 | 62% | 台北市, admin_level 4 | 2026-09-26 |
| New Taipei | 276 | 116 | 42% | 新北市, admin_level 4 | 2026-09-26 |
| Shenzhen | 226 | 25 | 11% | 深圳市, admin_level 5 | 2026-09-26 |
| Taichung | 223 | 90 | 40% | 台中市, admin_level 4 | 2026-09-26 |
| Hong Kong | 154 | 74 | 48% | 香港, admin_level 4 | 2026-09-26 |
| Taoyuan | 151 | 64 | 42% | 桃园市, admin_level 4 | 2026-09-26 |
| Tainan | 140 | 34 | 24% | 台南市, admin_level 4 | 2026-09-26 |
| Hsinchu | 43 | 27 | 63% | 新竹市, admin_level 4 | 2026-09-26 |
| Country | Shops mapped | Stating a street | Share that does | In an enclosure | Share inside | Boundary used | Measured |
|---|---|---|---|---|---|---|---|
| Malaysia | 1440 | 327 | 23% | 161 | 11% | ISO3166-2 of each state polygon | 2026-09-26 |
| Singapore | 300 | 104 | 35% | 139 | 46% | country polygon, admin_level 2 | 2026-09-26 |
| Brunei | 91 | 12 | 13% | 10 | 11% | country polygon, admin_level 2 | 2026-09-26 |
A further 62 shops in the same extract carry no country code. The state polygon that would name them is cut by the edge of the extract, so nothing on the map says where they are. They are counted here and assigned nowhere: an edge of an extract is not filled in by judgement.
This table does not measure how big a market is. It measures what OpenStreetMap holds. Look at the street column: in Dubai 78% of shops state their own street, in Shenzhen 11%. Dubai is mapped shop by shop, Shenzhen building by building — which is why a city with hundreds of counters inside a single tower can show a smaller number than a district of shopfronts. One instrument, one day, and what it reveals is a difference in recording, not in trade.
City boundaries are not numbered the same way from one country to the next. The boundary used is named for each city, so that nothing is compared across levels without saying so.
An enclosure is a named polygon — building, mall or marketplace — containing at least two of the counted shops; where polygons overlap, the smallest one wins, so no shop belongs to more than one enclosure.
Singapore stands apart there: 46% of its shops sit inside a named enclosure, against 8 to 11% elsewhere. Two explanations produce that figure and this data cannot separate them — Singaporean electronics retail may genuinely happen inside malls, or Singaporean malls may simply be mapped shop by shop. It would take a source outside OpenStreetMap to say which.
Every row here depends on a boundary that closes. Where a boundary's members run past the edge of the data extract, the shape does not assemble and the place cannot be counted — so a missing row means a cut boundary, never an absence of shops.
Measured or looked for, and not in the table above:
Places measured so far: 12. The others are being counted with the same method.
Huaqiangbei: 61 named markets in 1.45 square kilometres
The duty-free gateway: Golden Computer Centre, Apliu Street, Sim City
Guanghua Digital Plaza, Syntrend and the NOVA malls across six cities
Greater Bay Area retail, alongside Hong Kong and Shenzhen
Deira and Bur Dubai, street by street
The island’s electronics retail, mall by mall
Kuala Lumpur, Penang and the rest of the peninsula