Last month we published a breakdown of every hotel in China that will actually let a foreigner check in. The number our readers kept quoting was not the total. It was the hole in it.
Of the 359 cities in our database, 355 have at least eight hotels we would put in front of a stranger. Four do not: Alashankou, Beitun, Kunyu and Sansha.
That reads like a data quality problem. It is not. It is a design decision that happens to produce a visible number, and the interesting part is the decision, not the four cities.
What we actually measured
For every property we track three things: whether the listing states it accepts foreign passports, whether it has a verified street address in Chinese, and whether there is a public review history behind it. A hotel with no Chinese address is not usable by the person we are writing for — a traveller standing at a front desk with a phone.
So "eight hotels" does not mean eight rows. It means eight entries that survive all three filters. That is why the count per city is lower than any booking site will tell you.
Where the four sit
Alashankou and Beitun are border towns on the Kazakhstan and Mongolia crossings. They are transport stops, not destinations. Kunyu is the old name for a county seat in Xinjiang. Sansha is an administrative district of about 20,000 people served by a single road.
None of them is missing from our data. Each has hotels. What they lack is a supply of places a foreign traveller would plausibly book — which is a different and more honest statement than "we have no data here".
Why eight is the floor
This is the part worth arguing about.
Below eight properties, a city page stops answering a question and starts being a list. With three options you cannot choose on district, on price, or on whether the desk will take your passport — you take whatever exists. The page becomes a dead end that looks complete.
So we made a rule: cities under the floor get the nearest larger city instead of an empty page. The traveller lands somewhere useful rather than somewhere technically correct.
The alternative — publish all 359 and let four of them be three-hotel pages — would have made our coverage look better and made the product worse. This is the trade every directory-style product has to make, and it is almost always made in the direction of the bigger number.
The other 355
Those are the ones worth planning around. Shanghai, Beijing and Guangzhou each carry hundreds of properties with verified addresses. The long tail is smaller cities where a handful of well-chosen hotels is still enough to decide where to base yourself for three nights.
You can browse the full list of 359 cities and their hotel counts, or start from a city and see the map, the Chinese name card and the taxi instructions for each property. Everything else we track — month-by-month climate, which cities are comfortable right now, the taxi cards — lives at sinostay.
The general lesson
If you are building anything that aggregates messy real-world data, the threshold is the product. A count that includes everything is easy to produce and useless to act on. A count that excludes things forces you to write down what "usable" means — and once you have written it down, the number that comes out is defensible even when it is smaller than the one your competitor is advertising.
We publish the rule alongside the number for exactly that reason. When the rule changes, the number will change, and you will be able to see why.
Top comments (0)