Methodology & data sources

Three layers, one dataset

The guide deliberately separates what is measured from what is judged, because the two carry different weight in a decision.

1. Official statistics. Population, GDP, GDP growth, disposable income, urbanization and internet users are taken from the National Bureau of Statistics and from provincial or municipal statistical bulletins for the reference year printed on each profile. The policy designations (free trade zone, international consumption centre, services opening-up pilot, offshore duty-free) are also official facts, taken from the State Council or ministry notice that created them — they are recorded, not assessed.
2. Derived indicators. USD conversions, per-capita figures and platform penetration percentages are calculated from the statistics above plus published platform disclosures. These are reproducible from the stated inputs.
3. Judgement layer. Key industries, business districts, hub level and brand entry notes are YIVA's editorial assessment from client work and market observation. They are opinions, clearly labelled as such, and they are the part of this dataset you cannot get from a statistics bureau.

City profile fields

Province profile fields

Units and conversions

• Population and internet users are stored in 万人 (ten-thousands) and displayed in millions or billions. A stored value of 1,868 means 18.68 million people.
• GDP is stored in 亿元 (hundred-million CNY) and displayed in USD billions at an indicative rate of 7.1 CNY per USD.
• Income is annual per-capita disposable income of urban residents in CNY. It is not household income and it is not adjusted for purchasing power.
• Percentages for platform penetration are shares of local internet users, not shares of the total population. A city with 60% WeChat penetration has 60% of its internet users active on WeChat.

How platform penetration is estimated

Platform penetration is the only field in this dataset that is modelled rather than reported. It is built from platform-level user disclosures and third-party industry data, combined with each city's internet-user base and tier classification. It is published as a directional estimate, and it is labelled as one everywhere it appears.

The City Entry Index

The YIVA China City Entry Index (YCEI) is a 0–100 score for each scored city. It is built only from fields documented above, so anyone holding the dataset can reproduce it exactly.

Why “entry”. The other metrics in this guide describe a city; an entry score describes a decision. Reach, spending power, market size and policy permission are the four things that decide whether a first store, a flagship or a cross-border launch survives its first year — and they can disagree sharply. A city can be rich and unreachable, or reachable and too small to carry the fixed cost. Naming the index after the decision rather than after one of its inputs keeps the score from being read as another GDP ranking.

• Digital access (30%) — Xiaohongshu 30%, WeChat 25%, Douyin 25%, internet penetration 20%.
• Consumer depth (30%) — disposable income 45%, GDP per resident 35%, real GRP growth 20%.
• Market scale (25%) — gross regional product 60%, population 40%.
• Policy access (15%) — four published designations, scored absolutely rather than ranked (see below).

Each input becomes a percentile rank within the scored set, inputs are averaged inside a pillar, and pillars are averaged into the overall score. Rank scaling is used because GDP, population and income are heavily skewed; a min–max scale would turn the index into a GDP ranking. Two consequences follow and are stated on the index page: scores are relative to this release, and they are not a recommendation.

Policy access

Policy is the one pillar that is not ranked against peers. Each city is checked against four designations taken from published State Council and ministry notices. The points sum to exactly 100, so they need no rescaling:

Why absolute and not ranked: a mostly binary variable has almost no variance. Ranking it would turn one missing designation into a twenty-point gap and make a yes-or-no question look like a gradient. The pillar is capped at 15% for the same reason — policy changes what you are allowed to do in a market, it does not create the market.

What is deliberately not tracked: cross-border e-commerce comprehensive pilot zones, because by 2022 they covered 165 cities including almost every city here, so the flag carries no information. Nor does the pillar distinguish a thriving free trade zone from a nameplate, which is why the city pages show the designations themselves rather than only the score.

Hong Kong and Macao are profiled but not scored: their social reach runs through WhatsApp, Instagram and Facebook rather than mainland platforms, so mainland penetration is not a comparable measure there. Full weights, bands and the ranked table are on the City Entry Index page.

Downloading the dataset

The city and province releases are served from the live database, so a corrected figure is in the download the moment it is in the profile — no release lag, no stale snapshot.

Access needs a free API key from your YIVA account, sent as Authorization: Bearer <key>. Licence is CC BY 4.0. Excluded from the release: brand entry tips, business-district lists, consumer highlights, policy-zone notes and provincial narrative — those remain in the profiles.

Refresh cadence and versioning

• Full refresh once a year, after provincial statistical bulletins and the national yearbook are published.
• Individual profiles are re-verified when a city publishes a mid-year revision or a major administrative change.
• Pages revalidate hourly, so a correction propagates without a redeploy; the date on each profile reflects its last edit, not the last deploy.
• Material changes are logged below rather than silently overwritten.

Known limitations

• Official figures are revised. A city may restate GDP a year later; we update when it does, but historical pages are not retroactively rewritten.
• Administrative boundaries change. Municipal mergers and new districts can make year-on-year comparison misleading.
• Tier labels are a market convention, not an official classification. We use them because media buyers do, and we say so here.
• Disposable income covers urban residents only. It overstates rural purchasing power and understates wealth concentration at the top.
• No value is imputed. Where a figure is unavailable, the profile shows the year actually used.
• This is a market-entry dataset, not an investment prospectus. Verify before committing capital.

How to cite

Cite the specific profile rather than the index page, and name the reference year. For the dataset as a whole, use the dataset citation below.

Corrections

If a figure here contradicts the publishing agency's own release, we want to know. Send the city, the field and a link to the source to [email protected] and we will correct the profile and note the change in the log above. Factual corrections take priority over new features.

Questions this page answers