On November 5, the government website of Nanshan District, Shenzhen, posted a news item. It was short on words but substantial in information: the National Data Property Rights Registration Service System has entered trial operation, and the Shenzhen Data Exchange has been selected for the national-level catalog of data property rights registration institutions.
The sentence in parentheses is even more crucial—only three data exchanges nationwide were selected: Beijing, Shanghai, and Shenzhen。
For Guangdong enterprises, this means that in the future they will no longer need to travel to Beijing; in Shenzhen they can obtain a nationally mutually recognized “ID card” for their own data, and the processing results will be integrated into the national system—one registration, valid nationwide.
What is written on this certificate
The credential states three things: the data holding entity, the scope of use, and the authorization boundary.
Put it in terms businesses understand: whose data this is, what it can be used for, and who can use it.
Moreover, what can be registered is not one right, but three mutually independent rights: the right to hold data, the right to use data, and the right to operate data, and enterprises apply according to their own actual circumstances.
The separation of these three is the most thought-provoking aspect of this institutional design.
A company that makes components has more than a decade of process parameters and quality inspection data. It may not want to sell them, but is willing to license them to equipment manufacturers to train models—this is the right of use. It may also not act itself and instead package the data into an industry dataset for trading—this is the right of operation. It may also simply want to prove that “this batch of data is mine,” so that there is a basis in future litigation or mergers and acquisitions—this is the right of holding.
Three kinds of demands, three kinds of registration—there's no need to bundle them together.
A misunderstanding that must be clarified
Many people treat data property rights registration and the inclusion of data assets in accounting as the same thing, believing that registration must come first before they can be included in accounting.
No.
The official position is clear: registration and inclusion in financial statements are mutually independent, and registration is not a prerequisite for inclusion. Conversely, registration certificates can support subsequent business such as inclusion in financial statements, financing, and equity investment.
This sequence needs to be clarified: registration is the confirmation of rights, while inclusion in the financial statements is accounting treatment. Confirmed rights do not necessarily mean inclusion in the statements (because the asset recognition criteria may not be met); anything included in the statements must have a basis for ownership.
The practical significance for enterprises is: the threshold for registration is much lower than for inclusion on the balance sheet. Getting the certificate first is a sure-win, no-loss deal.
The Four Uses of a Voucher
The message listed four points; I've reordered them by importance to the company.
Financing. This is the most direct monetization path. Data asset pledge financing has been tried in various places over the past two years; whether banks approve loans hinges first on “is this thing actually yours?” With a national-level certificate, this hurdle is much easier to clear.
Transactions. To sell or buy data, unclear ownership is the biggest obstacle. The buyer has to conduct due diligence, the seller has to provide proof, and both sides investigate separately, making transaction costs absurdly high. A single certificate can save most of the process.
Presenting evidence. When data is taken away by employees or misappropriated by partners, the hardest part in litigation is proving “this is mine.” The role of proof here is equivalent to a property deed.
Policy recognition. All localities are promoting data enterprise cultivation and data element demonstrations; during review, they look at how much data the enterprise actually has and of what quality. Certificates are the easiest proof to provide.
But here's the real question: what can you register?
A system launch is a good thing, but don't assume you can get certified just by filling out a form.
Data that can be registered must have at least four characteristics.
Bounded. Which fields, how many records, what time period is covered, how often it's updated. Saying "all of our company's business data" won't work — that's unbounded.
Has a source. How the data was collected, who authorized it, and whether it infringes on personal information or third-party rights. This item is the most prone to problems—the customer list contains individuals' mobile phone numbers, and supplier quotation sheets contain confidentiality clauses; both can get stuck.
Quality. Completeness rate, accuracy rate, consistency. A dataset missing 30% of fields and with five names for the same customer won't pass the registration authority.
There are applications. Who is using it, what it is used for, and what value it creates. Data that simply lies on a hard drive is meaningless even if it is registered.
What exactly do manufacturing enterprises have in their hands?
Among the clients I have encountered, most people's first reaction is, "We don't have much data."
Actually, quite a few. They're just scattered.
Supplier master data, material master data, BOM structures, process parameter libraries, equipment maintenance records, quality inspection data, energy consumption data, and after-sales failure databases—these are all real assets in manufacturing. Process parameters and failure databases in particular are things accumulated over more than a decade, which a new factory cannot buy even with money.
The problem is that the way this data looks in the system is often not presentable.
The same material has three codes—procurement uses one, the warehouse uses another, and finance created yet another; the same customer has five records because a salesperson left and took the original records; the BOM has been revised seven times, and no one can say clearly which version is the mass-production version; quality inspection data is in the inspector's Excel and has not entered the system.
Under these circumstances, the registration materials simply cannot be fabricated. Even if they are barely fabricated, they will be exposed as soon as they are checked.
So the order should be reversed
It is not 'register first, then govern data'; it is 'govern data first, only then can registration be discussed.'
Specifically, do three things; none are difficult, but someone has to keep an eye on all of them.
Unified coding. Define one coding rule for each of the four types of master data — materials, customers, suppliers, and equipment — and use only one set company-wide. Once this is done, it will solve half the problem.
Keep a trail. There must be a record of who changed which piece of data and when. This is not to investigate people; it is to prove that the source of the data and the process of changing it are trustworthy.
Manage by batch. Much of the value of manufacturing data is reflected in the batch dimension. Which batch of material, which line, which shift, what parameters, and how many qualified products were produced. If batches cannot be linked together, the data is just a pile of isolated orders.
These three things happen to be exactly where ERP is strongest. Master data management, change trails, and batch traceability are inherently part of the system's job. Data governance does not require starting from scratch; it is enough to build the fundamentals in the system solidly.
One last honest word
During the "14th Five-Year Plan" period, pilot locations across the country cumulatively issued more than 40,000 data intellectual property registration certificates, with financing credit enhancement and licensing transaction amounts approaching 15 billion. That sounds like a lot, but spread across the tens of millions of enterprises nationwide, the proportion is so low it can be ignored.
This shows two things: first, the road has just opened, and those who move early have the advantage; second, the vast majority of enterprises are not yet ready.
Ready with what? Not ready with materials, but ready with the data itself.
Certificates are issued to people who have data. If you have a pile of Excel files you cannot sort out, you cannot get in even with the registration window open.
