Can You Check a Chinese Company's Employee Count?
A Chinese annual report may show employee and social-insurance figures, but they are not a live factory headcount. Learn how to compare the entity, date, and worker scope behind each number.
Published by ChinaValidatePublished August 24, 2026Last updated August 24, 2026
Start with the Chinese legal name or USCC. Confirm the matching company before opening its profile.
A sourcing manager is preparing to approve a new electronics supplier. The sales deck describes a 300-person factory. During a video tour, the buyer sees perhaps 90 people across one workshop. The supplier's latest public annual report shows a much smaller employee figure, while its social-insurance section gives several counts that do not quite match one another. Which number describes the company the buyer is about to pay?
None of the four figures is automatically false, but they are not interchangeable. One may describe a corporate group, another a shift at one site, another the employees of one legal entity at the end of a past reporting year, and another the people enrolled by that entity in a particular insurance scheme on a specified date. A public company record can expose a claim that needs explanation. It cannot, by itself, count everyone currently inside a factory.
- Sales deck300Claimed factory workforce; date and legal-entity scope not stated
- Video tourabout 90People visible at one site, during one shift, at one moment
- Annual report68Company-reported employees for one entity at the report-year end
- Social insurance61-66Participants reported by insurance type on the reporting date
Can the public record tell you how many people work there?
It can give you a defined historical figure for a defined reporting entity, if that field is publicly visible. That is useful. It is also much narrower than the question buyers usually have in mind. “How many people work there?” may mean the seller's direct employees, everyone working at a factory today, the workforce across a corporate group, or the people available at peak production. A Chinese annual report does not promise to answer all of those questions.
Use the public number to test the supplier's story, not to declare an exact live headcount.The record is strongest when you can attach four labels to the number: reporting company, measurement date, worker category, and source.
Under China's current Interim Regulation on Enterprise Information Disclosure, companies file the previous year's annual report between January 1 and June 30. Employee count is among the report items, but the enterprise may choose whether to make that field public. The company is responsible for the truthfulness and timeliness of what it reports. This creates two important limits: the number is self-reported rather than an independent workforce audit, and a missing public value may mean “not disclosed” rather than zero.
The official English-language Shanghai market-access guide likewise describes annual reporting for the preceding year and lists the number of employees among the report contents. If you need to find the filing year, corrections, shareholders, assets, websites, and other sections first, start from the ChinaValidate Resource library. The narrower task here begins after you have found the workforce figures.
Why can one supplier show four different headcounts?
The four figures in the opening case answer different questions. The sales deck may use “our factory” to describe a group, a manufacturing affiliate, multiple sites, direct and temporary workers together, or a peak-season workforce. The video shows who happened to be visible within the camera's field of view. The annual report belongs to one named legal entity and a past period. The insurance figures count enrolments reported by that entity under separate schemes.
- Who reported it?
- A supplier's marketing team, the registered company, an insurance-reporting unit, or the buyer's own observer may each use a different method.
- Which entity or site?
- The contracting company, a branch, a manufacturing subsidiary, a labour provider, and an entire group are different subjects even when the same brand name is used.
- When was it measured?
- A year-end filing, today's first shift, a peak-season plan, and an undated website claim cannot be compared as if they were simultaneous.
- Who is included?
- Direct employees, dispatched workers, contractors, retirees, office staff, and everyone physically onsite are not necessarily included in the same measure.
Before judging the gap, rewrite each figure as a complete sentence. For example: “The contracting company reported 68 employees at the end of 2025,” or “The sales presentation claimed that 300 people work across the supplier's factory operations, but gave no date or entity.” Once the nouns and dates are visible, some apparent contradictions disappear. Others become sharper because the supplier cannot say what its larger number covers.
Do not average the figures or select the largest one. A headcount is not more reliable merely because it is precise, and a visual estimate is not more current merely because it came from a live call. The task is to make like-for-like comparisons. If the supplier says the contracting entity itself employs 300 people today, a prior-year figure of 68 does not disprove the claim, but it makes the growth, entity scope, and current support worth asking about.
What do the public workforce figures actually cover?
The annual-report employee field and the social-insurance fields sit close together in company data, but they are not one repeated measure. Read each section under its own reporting rule before comparing either of them with a supplier's current claim.
What does the annual-report employee number cover?
Current official filing guidance from the Hainan Administration for Market Regulation describes employee count as a report-year-end figure for employees who have labour contracts with the reporting entity. Most annual-report values reflect the position at December 31. This is a useful reading rule, but you should still check the year and any field notes shown in the actual record rather than assuming a current staffing level.
What the field can support
“This legal entity reported this many employees at the relevant year end.”
What it cannot support alone
“This many people work at the supplier's factory now.”
The legal entity matters as much as the date. A sales company may contract with the buyer while a related manufacturing company operates the plant. The annual report opened under the seller's name then describes the seller, not automatically its affiliate. Conversely, a buyer may open the factory entity's record while the supplier's “300 employees” refers to the wider group. Company names, Unified Social Credit Codes, and parent-child relationships must be matched before the counts are interpreted.
Branch reporting creates another trap. The same Hainan guidance says that a non-independently accounted branch may report zero employees when the head office reports those workers centrally. Zero can therefore have an accounting explanation in a specific branch record. That possibility is not a universal excuse: the buyer still needs to confirm that the record is a branch, identify its head company, and understand where the relevant workers are reported.
Annual reports can also be corrected, and the regulation makes correction information visible. A corrected number should not be silently combined with an earlier screenshot. Record the report year, the date you viewed it, and whether the screen identifies a correction. If the staffing explanation depends on a newly formed entity or recent restructuring, compare it with the company's registered change history instead of treating the filing as a real-time snapshot.
Why do the social-insurance numbers not always match?
Social-insurance data was added to enterprise annual reporting through a joint notice from the market-regulation, human-resources, and statistics authorities. The official notice distinguishes the enterprise's employee-related statistical fields from the number of people participating in each insurance type. Seeing several numbers in one annual report does not make them alternative versions of one audited headcount.
The Jinan Social Insurance Service Center's current filing guidance defines an insurance participant count as the reporting enterprise's employees enrolled in that insurance type on December 31 of the report year and excludes retired personnel. Counts may differ by scheme because eligibility, enrolment status, timing, or reporting treatment differs. The public interface may also allow some insurance information to remain non-public.
- 61 pension participants
- people reported by this entity in that scheme on the stated date
- 64 unemployment-insurance participants
- a separate scheme count, not an invitation to choose the larger number
- 66 workplace-injury participants
- another defined enrolment count, still not everyone seen at the site
A modest difference between schemes is therefore a question to understand, not instant proof of false reporting. A much larger gap between a claimed factory workforce and every company-level figure deserves a clearer explanation. Ask which people are direct employees of the named company, which belong to another group entity, and whether any workers are supplied under a documented labour arrangement.
Chinese labour law provides one reason why people working onsite may not all appear under the site's company record. Under the Labour Contract Law, a labour-dispatch agency is the employer of dispatched workers and signs labour contracts with them; the dispatch agreement addresses the accepting work unit, positions, worker quantity, remuneration, and social-insurance arrangements. Independent ILO research on non-standard work arrangements in China also illustrates that the worker, intermediary, and place where work is performed do not always align as one entity.
That does not mean “dispatched labour” should be accepted whenever numbers conflict. It is one possible, documentable relationship. A supplier relying on it should be able to describe the labour provider, the approximate people and roles covered, the sites involved, and the period, without sending unnecessary personal records. An unexplained label does not reconcile the figures.
Which company do the people belong to?
Supplier conversations often use one English brand name for several Chinese entities. The buyer, however, signs with and pays a particular legal person. The annual report and insurance data also belong to particular reporting units. If the workforce claim is attached to the brand while the public figures attach to one company, the entity structure must be made explicit before the comparison is meaningful.
People working on the orderWho employs them, and where?
Contract companySigns the purchase contract, invoices, and receives payment
Factory companyOperates the production site and may be a subsidiary or affiliate
BranchMay share centralized accounting or workforce reporting with its head office
Labour providerMay employ dispatched workers who perform work at the factory site
The relationship may be commercially normal, but it changes the transaction. If a manufacturing affiliate employs most workers while a trading company takes the order, the buyer should know which company controls production, which one is responsible for quality and warranty obligations, and whether the contract accurately describes subcontracting. Where the activity is regulated, also match the correct company's business license and operating permission. A group headcount should not be presented as the contracting company's headcount unless the scope is stated.
The same distinction matters during a site visit. People wearing one uniform or using one badge do not prove a common employer. The premises may contain shared services, temporary labour, multiple subsidiaries, or tenants. Ask for an aggregated explanation and match the legal names to the company records. Avoid collecting employee names, identity numbers, individual social-insurance histories, or identifiable face images when they are not necessary for the supplier decision.
Headcount should also remain one input to the wider factory question. A small manufacturer can be capable in a specialized process, while a large trader can coordinate substantial outsourced production. If the real concern is whether the supplier owns and operates the factory it claims, continue with the broader supplier-verification checklist rather than using employee count as a shortcut.
How should a headcount gap change your supplier decision?
The answer depends on how specifically the supplier described its workforce, how material capacity is to the order, and whether the company can support a current explanation without exposing personal employee data.
When does a smaller number need an explanation?
A difference becomes more important as the supplier's claim becomes more specific. “Our group has about 300 people during peak season” may coexist with 68 year-end employees in one company. “The company named on your contract directly employs 300 permanent workers at this factory today” is much harder to reconcile with the same public history without evidence of rapid hiring, restructuring, a changed reporting entity, or an error.
The marketing number covers two manufacturing subsidiaries and seasonal dispatched workers; the annual-report figure covers only the contracting company at year end. Company records and current aggregated documents support the explanation.
The supplier identifies when a new line opened, which entity hired the team, and what current operational records can show. The explanation is plausible, but the old filing alone cannot confirm it.
No one can say whether the figure covers a group, one factory, direct employees, temporary workers, or a past peak. The contract company and factory company also keep changing across documents.
Other context can raise the need for verification: the supplier describes multiple production lines but cannot identify the operating company; the factory address differs from the company record without an explanation; the video tour avoids ordinary production areas; or the claimed workforce changes substantially across recent documents. None of those details proves misconduct in isolation. Together, they determine how much reliance you can place on an unsupported scale claim.
The opposite error is treating a large public number as a capability certificate. Employee count does not show whether the company owns the factory, whether the workers have the relevant skills, whether the machinery is available, whether the production line is allocated to your order, or whether quality controls work. It also does not prove financial strength or on-time delivery. Those questions require their own evidence.
What can support the supplier's current scale claim?
Ask for information proportionate to the order and the mismatch. The aim is not to obtain a list of employees. It is to connect the current workforce claim to the contracting entity, manufacturing entity, operating site, and relevant production period. For a modest trial order, a clear entity explanation and a live site discussion may be enough. For a large deposit, time-sensitive launch, or regulated product, the current operating evidence should be stronger.
- Current organization and entity mapShows which Chinese companies employ sales, engineering, production, and quality teams, and which company operates each site.
- Aggregated staffing supportA recent payroll summary, insurance declaration summary, or other company-level total can support scale without exposing personal employee data.
- Documented labour arrangementIf dispatched or temporary workers are material to the claim, identify the provider, approximate roles and numbers, period, and work site.
- Recent production recordsShift plans, anonymized capacity records, equipment logs, and order schedules can connect people and lines to the products actually being offered.
- Independent site verificationA targeted visit or live inspection can confirm the operating company, active lines, staffing pattern, and whether the observed activity fits the proposed order.
Each item has limits. An organization chart can be invented; a payroll total can belong to the wrong entity; a busy workshop can be staged or shared; and one site visit captures one period. Cross-checking works because the sources fail in different ways. The legal names should stay consistent, the dates should make sense, and the operational picture should fit the company's public history and the production promised in the contract.
Your presentation states that the factory has about 300 employees. The latest public figures for the contracting company are lower. Please clarify which Chinese legal entities and sites are included in the 300-person figure, the date or period it describes, and whether it includes dispatched or temporary workers. Please also provide current aggregated support for the staffing relevant to our order. We do not need employee names, identity documents, or individual insurance records.
A focused request is easier for a genuine supplier to answer than a demand for “proof of all employees.” It also prevents the buyer from receiving a folder of unrelated personal data. If the supplier identifies a separate manufacturer, update the company checks and contract analysis around that entity. For a material order, the broader company, site, production, and capability work still has to be completed; headcount alone cannot replace it.
What should change before you place the order?
Return to the four figures in the opening case. The buyer does not need to force them into one number. The buyer needs to decide whether the supplier has given a coherent, current account of the people and entities that will perform the order.
-
Supported for the stated scope
The supplier defines the 300-person claim as a group or site total, identifies the participating entities, explains the year-end company figure and insurance counts, and provides current aggregated operating support. The scale claim can be used within those stated limits.
-
Plausible, but not yet current
The difference can be explained by report timing, a new line, branch reporting, or a documented labour model, but current support is incomplete. Keep the claim provisional and size the order, deposit, or inspection requirement accordingly.
-
Unresolved before relying on capacity
The supplier cannot identify which company employs the workers, which site the number covers, or when it was measured. Do not approve a production plan or payment on the strength of the headcount claim.
The disciplined conclusion is intentionally limited. A low annual-report or insurance count does not prove fraud, and a high number does not prove capacity. Public workforce data becomes valuable when it is tied to the correct company and date, compared with a precisely stated supplier claim, and followed by current evidence that is proportionate to the transaction. That turns a confusing number into a useful supplier question without pretending that the company register is a factory turnstile.