01 Layer 1 · Fit

Does your business need a Chinese entity at all?

Most founders register a company before they can say what the company is for. Start with the job the entity has to do — invoice, hire, hold a licence, or nothing yet.

Every year, thousands of foreign founders register a Chinese company before they have a customer, a product licence or a reason. The registration is not the hard part — it is fast, increasingly cheap, and can be done in weeks. The hard part is the year after, when the company has an address, a bookkeeper, a bank account, a tax filing calendar and no revenue.

This layer exists to prevent that. It asks one question: what will the entity do that cannot be done without it?

The five jobs an entity actually does

A Chinese company is a legal person. In practice, it performs five jobs, and each one has an alternative that does not require registering a company.

  1. Invoice a Chinese customer. A domestic customer usually wants a fapiao, the official tax invoice. Without a Chinese entity you cannot issue one. The alternative is to sell to an importer or distributor who issues the fapiao and bears the margin.
  2. Employ people in China. Payroll, social insurance and work permits all require a registered employer. The alternative is a staffing agency or a professional employer organisation, which costs a margin but no entity.
  3. Hold a licence. Restaurants, clinics, schools, travel agencies, factories and telecom services need a licence held by a Chinese entity. There is no meaningful alternative except a licensed partner.
  4. Sign a lease or a contract locally. A Chinese company can hold property, vehicles and equipment. A foreign company can usually sign contracts, but enforcement and registration are harder.
  5. Receive RMB and hold it. A domestic bank account lets you receive RMB, pay local suppliers and keep working capital in the country. The alternative is receiving foreign currency offshore and converting at the border.

If none of these five jobs is on your critical path in the next twelve months, the entity is premature.

Note

"Premature" is not "wrong". Some founders register early to reserve a name, qualify for a subsidy or satisfy an investor. That is a legitimate choice — but price it as an option, not as a business expense, and know what the annual compliance will cost.

The cheaper alternatives, in order

Before registering, work down this list. Each step is cheaper and faster than the one below it.

  • Sell through a Chinese importer or distributor. They hold the licence, issue the fapiao, carry the inventory and take a margin. You keep the brand and the product. The risk is dependence and grey-market leakage.
  • Sell cross-border through a platform. Cross-border e-commerce allows a foreign company to sell into China through bonded warehousing and a platform, with the platform handling customs and tax. It works for consumer products, not for services or B2B contracts.
  • License the product or the brand. A Chinese licensee manufactures or operates under your IP. You need a strong trademark registered in China before you sign, or you are licensing something you may not own there.
  • Use a professional employer organisation. If the only job is hiring one or two people, a PEO can employ them and second them to you. You lose direct control and pay a service fee.
  • Register a representative office. If the job is presence and liaison — market research, supplier coordination, quality control — an RO is cheaper than a WFOE and cannot trade. It is a narrow tool, but a real one.

Only when none of these does the job should you move to Layer 2.

What the entity really costs

The registration fee is the smallest number. A service WFOE in a tier-one city typically costs ¥20,000–¥60,000 in the first year once the address, agency, seals, bank account and bookkeeping are counted, and that is before a salary for yourself. From year two, the floor is the address plus bookkeeping plus the audit and filings, commonly ¥15,000–¥40,000 — a fixed cost that arrives whether or not you have revenue.

Then there is the cost that does not appear on an invoice: the founder's time. Bank onboarding, tax registration, the first fapiao, the first payroll and the first annual report each take days, in Chinese, with a queue.

Rule of thumb

Divide the first-year cost by the gross margin on your expected first-year revenue. If the entity consumes more than a fifth of it, the entity is too early or the revenue plan is too optimistic.

The test this layer applies

Before moving on, you should be able to complete these three sentences without hedging:

  • "The company exists to ____________________, which cannot be done from abroad because ____________________."
  • "The first customer who needs that is ____________________, and they will pay approximately ____________________."
  • "If the company earns nothing for twelve months, I can fund ____________________."

If any sentence has a blank, the next layer will not fix it. Structure cannot rescue a business without a customer, and capital cannot rescue a business without a job for the entity.

When the answer is "not yet"

Choosing not to register is a legitimate outcome of this layer, and the site is designed to support it. A founder who sells through a distributor for two years, learns the market, registers the trademark and then sets up a WFOE has usually made a better decision than one who registered in month one and spent the next two years paying for an empty company.

What "not yet" requires is a trigger: the event that will make the entity necessary. A signed local contract, a hire, a licence application, a lease. Write the trigger down. Without one, "not yet" becomes "never", which is also a decision — just a less honest one.

  • Name the job the entity will do, in one sentence.
  • Name the first customer who needs that job done.
  • Price the first year, including your own salary and the buffer.
  • Write down the trigger that would make registration necessary.
  • If the trigger has not happened, stop here and revisit in six months.

The next layer asks the question that kills more projects than money does: whether you are allowed to do this at all.