Important data isn't about headcount but about process—what's stored in the BOM is exactly that kind of thing.

Six national standards will solicit public comments starting November 16; this industrial one lists R&D and design separately from production and manufacturing

A client in semiconductor packaging did data classification and grading the year before last, hired a consulting company, and produced a booklet of more than 100 pages. Looking at the grading results, the most sensitive was labeled “customer list,” followed by “financial data.”

I took a look at their process routing: more than twenty operations, with three to five parameters attached to each operation—temperature, time, concentration, and pressure. The process cards are PDFs, and the attachments are directly attached to the operation records; anyone with production module permissions can download them.

In that booklet, the process parameters were categorized as “general data”.

Six national standards are open for public comment together; the industrial one ranks fourth

On November 16, the Secretariat of the National Data Standardization Technical Committee issued a notice, Document No. [2026] 60 of the National Data Standardization Technical Committee Secretariat, soliciting public comments on six national standards: "Important Data Identification Catalog: General Provisions," as well as five industry catalogs for civil aviation, telecommunications, industry, aerospace, and geographic information. The deadline for feedback is January 16, 2027.

The project initiation date of this batch of standards should be traced back to May 22 this year, based on National Standards Committee Document [2026] No. 33, the 2026 fifth batch of recommended national standards plan.

That is to say, it took nearly half a year from project initiation to the formation of the draft for public comment. From public consultation to official release, under the normal process it still has to go through consolidation of comments, submission for review, and submission for approval; whether it can be released next year is impossible to say right now. Let me make this clear first: as of the writing of this article, all six documents are drafts for public comment, not standards in effect, and cannot be used as a basis for compliance.

The applicable subject stated in this industrial document is “industrial data processors.” The classification method consists of six categories: R&D and design, production and manufacturing, business management, operation and maintenance, platform operation, and application services. At the same time, it leaves an opening—as a macro-level general rule, subdivided industries may separately formulate standards.

The six industries are being issued together, and there is a deliberate rhythm to the timing. The general provisions set a common baseline, while the industry catalog sets specific entries. In the past, when enterprises carried out classification and grading, they relied on the 2024 "Guidelines for Identifying Important Data in the Industrial Field" and GB/T 43697, making judgments and setting grades themselves; after grading, no one verified them. Now that there is a cross-reference table within the industry, the criterion for judgment has changed from "what you think" to "checking against the catalog."

In the threshold column, most factories can't reach it; what they can reach is another column

The identification dimensions of the industrial catalogue are roughly five: related to the generation of state secrets, related to national security, related to industry development security, related to export-controlled items, and related to personal information.

The threshold for this personal information clause is written specifically: one million or more people's personal information, or ten thousand or more people's sensitive personal information—once this is reached, it constitutes important data. The threshold for core data is higher: ten million or more people's personal information, or one hundred thousand or more people's sensitive personal information.

For the vast majority of manufacturing enterprises, this threshold is out of reach. An auto parts factory with an annual output value of 200 million, 400 employees, and several thousand customer contact records is two orders of magnitude away from one million. Seeing this, they feel relieved and then put the whole matter aside—this is the most common misjudgment.

What is genuinely most likely to be triggered is the third item, which relates to industry development security. The current Guidelines for Identifying Important Data in the Industrial Sector are very explicit under the dimension of 'related to national science and technology security': they cover technical data unique to China's industrial sector that involves technologies in the Catalogue of Technologies Prohibited or Restricted from Export in China, such as unique formulas, manufacturing processes, production processes, coding and encryption technologies, preparation technologies, and decoration techniques.

Formulations, processes, and preparation technologies, translated into system language, are: the hierarchical structure of the BOM, the operation sequence of the process route, the parameter ranges for each operation, and the substitution relationships of key materials.

Where do these things exist in Odoo? BOM is in mrp.bom and mrp.bom.line, routing is in mrp.routing.workcenter, parameters, if extended, are in a custom table, and the process card PDF is attached on ir.attachment. Four tables, a product's complete process knowledge.

The worst situation I have seen at a customer site was these four tables being open to everyone. Sales could see BOM costs, procurement could see process parameters, and when someone who left took away a PDF, they took away a formula. Before classification, none of this was a big deal; after classification, every item is a hole that needs to be plugged.

There is one more point that is easily overlooked: supply chain data. The guidelines mention that "related to industry development security" includes supply chain data. For companies engaged in foreign trade, data such as customer structure, order distribution, and capacity utilization, when aggregated to a certain extent, can be used to infer industry prosperity. This follows the same logic as the previously mentioned "undisclosed excavator index"—a single company alone doesn't amount to much, but when aggregated, it becomes macroeconomic information.

This is also where this catalogue differs from the Data Security Law. The Data Security Law addresses the obligation of tiered protection, whereas the catalogue addresses "which category counts as important." The former is a principle, the latter is a list. Only once the list is out can law enforcement have a basis, and can enterprises have a reference for self-inspection.

After the grading, three things need to be changed

Suppose that after self-inspection, certain categories of process data are designated as important data. At the system level, what needs to be changed is not "adding a password," but three areas.

The first is storage location. Important data cannot be mixed with ordinary data in one database and one permission system. The easiest approach in Odoo is to create a separate attachment directory for this type of record and configure an independent access group, rather than using the default public and portal permissions. Attachments stored in the filestore are hashed by SHA, so the path does not reveal the content, but the ir_attachment table in the database records the associated object—who attached it, which document it was attached to, and when it was attached, all of which can be queried.

Second is access logs. Currently, most Odoo projects only enable message logging through mail.thread, not field-level read logs. If you really need to trace “who downloaded this process card and when,” default logs cannot answer that. Either add an audit module, or add a layer of logging on the attachment download path. This must be done after classification and before inspection; it cannot be retroactively added afterward.

Third is export behavior. In the compliance requirements for important data, export and cross-border transfer are two high-risk actions. Foreign trade enterprises should pay particular attention: sending quotations, drawings, and process descriptions to overseas customers is a routine business action. Once data is classified, this action must go through approval and must leave an audit trail. What the system can do is separate “external sending” and “internal circulation” in terms of permissions, so that salespeople do not use the same button to complete two things.

There is a practical issue that needs to be laid out clearly. Grading is an obligation for enterprises, but the remediation after grading costs money, requires downtime, and requires changing processes. For SMEs, the motivation to do this is often not compliance but orders—when clients audit the factory, they want to see your data management, and large overseas clients especially look at this area. So what really drives this may not be the Cyberspace Administration, but the purchasers.

Three things you can do this week

1. Export the access permissions for BOM and process routes and take a look. List clearly in two tables who has read permission and who has export permission. This step does not need to wait for the standard release; it can be done now, and once it is done, you will have a clear picture.

2. Pick three products and first check for yourself whether their process parameters fall under the "R&D design" or "production manufacturing" category according to the six categories in the catalog. Note down any mismatches. Before January 16, you can submit comments through the official website of the National Data Standardization Technical Committee—this is a window for enterprises to influence the rules, and once it passes, it's gone.

3. In foreign trade business, make a list of "drawings and technical documents sent to overseas customers" and mark which ones contain process parameters. The list does not need to be long; it only needs to cover the main customers. This list will be useful later, whether for responding to customer factory audits or for responding to the approval workflow after grading.

The standard is still a draft for public comment. But the matter of enterprise self-inspection never waits.

关于我们

​我们致力于帮助中小企业实现数字化转型,我们的团队由一群充满激情和创新思维的专业人士组成,他们具备丰富的行业经验和技术专长。

扫一扫获取顾问以及手册

Sign in to leave a comment
72% of the equipment must be able to produce data, but the factory's ledger only records the original value.
The "Digital Guangdong Construction 15th Five-Year Plan" includes the equipment penetration rate in the indicator table, leaving the 2025 column blank.