Editor’s Note — DCC.
The Chinese data-element market is one of the most distinctive features of the country’s data regime — yet most overseas analyses treat it as a black box. Wang Qinglan’s plain-language primer answers the precondition question: what is actually for sale? The bakery metaphor she uses is more useful than any of the formal definitions and worth internalizing before approaching the data property rights registration regime, the Data 20 Articles policy framework, or the Shenzhen Data Exchange’s listing practice.
The four trading objects
The legal anchor is the Shenzhen Provisional Measures for Data Trading Administration (《深圳市数据交易管理暂行办法》). Article 6 lists four categories of object that can be traded on a Chinese data exchange:
- Data products (数据产品)
- Data services (数据服务)
- Data tools (数据工具)
- Other trading objects approved by the competent authority (其他经主管部门同意的交易标的)
The first three are well-defined; the fourth is a catch-all that has, in practice, expanded the regime’s flexibility — particularly to accommodate data resources (数据资源) as a tradable object even though they don’t fit cleanly into “data products.”
The bakery metaphor
Wang’s mental model: imagine a flour mill that becomes a bakery. The metaphor maps cleanly onto the legal categories.
| Bakery element | Chinese data concept | English translation |
|---|---|---|
| Wheat from the farmer | 原始数据 | Raw data / primary data |
| Flour (after milling) | 数据资源 | Data resources |
| Cake or cake base (after baking) | 数据产品 | Data products |
| The baker who turns flour into cake | 数据服务 | Data services |
| The oven, mixer, frosting spreader | 数据工具 | Data tools |
The full chain: a farmer harvests wheat (raw data); the mill turns it into flour (data resources); the bakery turns flour into cakes (data products). When a flour mill wants to enter the bakery business but lacks the skill, it hires a baker (data service). The baker needs equipment — oven, mixer, frosting tools (data tools).
The metaphor solves the conceptual puzzle. Raw data, data resources, and data products are all data in different states of processing. Data services are skills applied to data. Data tools are instruments for processing data.
The narrow vs. broad data product distinction
Wang highlights a frequently overlooked distinction:
- Narrow data products — data products in the strict sense. The data resource is the input; algorithmic processing yields the output. Examples: data sets, data analytics reports, data visualization products, data indices, API data products, encrypted data products. The flour-becomes-cake pattern.
- Broad data products — narrow data products plus data services and data tools. The broader category captures everything traded on a data exchange.
The key conceptual divide:
- Narrow data products contain data — they are the cake.
- Data services and data tools are methods or instruments for processing data — they’re the baker and the oven, not the cake.
Wang treats “data products” in the strict sense throughout her piece — the cake, formed by substantive processing of data resources, yielding derived data or data-derivative products. This narrow usage tracks the NDA’s Common Data Terms (First Batch) definition: “data processing products and data services that are formed on the basis of data processing and can meet specific needs.”
What raw data, data resources, and data products mean operationally
The bakery analogy maps to a concrete example Wang gives:
- You want to open a milk-tea shop. You hire counters to stand at major shopping-district entrances and record foot traffic in notebooks. Each notebook entry is raw data — an electronic-or-otherwise recording of an observable phenomenon (here, foot count at a location).
- At end of each day, the counters consolidate notebook entries into a single Excel spreadsheet. The spreadsheet is a data resource — raw data, primarily processed, with potential for value creation.
- A tourism-data company buys the spreadsheet and processes it into a shopping-district heat-map analytical report. They sell the heat map to an advertising agency for targeted ad placement. The heat-map report is a data product — the result of substantive processing of data resources.
This three-level distinction — raw / resource / product — is foundational. The DSL, the Network Data Security Regulation, and the NDA’s Data Property Rights Registration Work Guide (Trial) all rely on it. The legal consequences of mishandling each tier differ.
Can data resources be traded?
Wang flags a useful operational point. The Shenzhen Provisional Measures’ three-category enumeration (products / services / tools) seems to exclude data resources. But Article 6’s fourth category — “other trading objects approved by the competent authority” — accommodates them.
In practice, data resources can be traded on Chinese exchanges as a fourth-category object — both on-exchange and off-exchange. The Shenzhen Data Exchange and other regional exchanges have listed both data products and data resources.
The one substantive exception: public data resources (公共数据资源). Per the Implementation Specifications for Authorized Operation of Public Data Resources, public data must pass through authorized operation (授权运营) and be converted into public data products before it can be traded. Public data, in its raw resource form, is not a tradable object — only the products built on top of it.
The current trading-object landscape
Wang’s summary of the practical scope of tradable objects on Chinese data exchanges:
- Data resources — tradable under Article 6’s catch-all, both on-exchange and off-exchange (with the public-data exception).
- Data products (narrow) — the core tradable object. Includes analytic reports, indices, data sets, API products, encrypted data products.
- Data services / data tools (broad data products) — methods or instruments for processing data, tradable as their own category.
The exchange ecosystem is still maturing. New trading object types may emerge as the regime develops, and the trading rules will continue to refine.
Why this matters for overseas teams
Three operational takeaways for overseas counsel and compliance leads engaging with the Chinese data exchange ecosystem:
- Categorize before you transact. Whether you’re a buyer or a seller, the first question is what kind of object you’re trading. A data set is a different category from an analytical report (both data products, but with different compliance profiles). A SaaS analytics platform sold to a Chinese counterparty may sit in data tools, not data products. The categorization determines licensing path, classification obligations, and (for cross-border transactions) export-compliance obligations.
- Public data has a different transactional path. Public data resources cannot be acquired by foreign entities directly. They must be turned into public data products via the authorized-operation regime first. Foreign entities partnering on public-data-derived products should structure the partnership through an authorized operating institution (运营机构) recognized under the Public Data Resources Registration Interim Measures.
- The narrow vs. broad distinction matters for IP and product structure. A data service business model (algorithm-as-a-service, classification-as-a-service) operates under different rules than a data product business model (selling the resulting data set). Where the business is sold matters too — services trade differently from products on most exchanges.
The underlying point in Wang’s piece is that the Chinese data-element market has a much richer trading object taxonomy than the Western “data licensing” framing. A multinational treating data trading as a single category will miss the operational handles that the four-category framework provides.
— Wang Qinglan (王青兰), 数据交易，到底在交易什么？ (Data Trading — What Is Actually Being Traded?), 青兰数据观察 WeChat Official Account, May 28, 2024. Original article (Chinese).
Not legal advice. The above is DCC’s structured summary of Wang’s commentary; not a verbatim translation. The author’s views are her own and do not represent her employer.