A founder showed me a quote once that he was very proud of. Twenty-two dollars a unit, landed, on a consumer electronics product he expected to retail at ninety-nine. He had spent four months getting there: trade show in Shenzhen, three shortlisted factories, a hard negotiation that took the price down from twenty-six. He wanted me to tell him it was a good number.

It was a fine number. The problem was that he had no idea who his supplier actually was. The company he had been negotiating with was a trading house with a showroom and a WeChat account, and the factory that would build his product was somewhere two hours away, had never heard his company’s name, and would learn about his order as a line item. He had negotiated hard with an intermediary who had every incentive to let him win on price, because the intermediary was going to make its money on change orders, tooling amortisation and the second production run, when switching cost would be his problem and not theirs.

He had not solved his China problem. He had bought a quote.

The procurement frame is the mistake

Almost every Western founder I have met approaches China as a task with a finish line. Find a manufacturer, agree the minimum order quantity, get the samples approved, ship. The mental model is procurement: you are a buyer, there is a market of sellers, and your job is to run a competitive process and extract the best terms.

That model works beautifully for commodities you will never modify, where the specification is fixed, the supply is deep and you can walk away on a Tuesday. It describes almost nothing about building a physical product.

What you are actually doing is setting up a long production relationship in which the specification will change at least five times, the tooling will be modified, the certifications will be your problem, and you will be dependent on one or two people in a city you do not live in to tell you the truth about whether a batch is going out on time. The leverage in that relationship does not come from the price you negotiated. It comes from who controls the information and who owns the three connections that matter: the factory floor, the legal entity on the ground, and the end customer.

Across the consumer hardware work I did with the Jean-Michel Jarre venture, we went from one product to eight. The single most expensive lesson in that progression was not about cost per unit. It was about what happens when somebody else owns the relationship with the people who actually build your product. When you need an engineering change in week three of a production run, you find out very quickly whether you are a client or a line item on someone else’s schedule.

MOQ cash squeezes are a symptom, not a condition

Founders talk about minimum order quantities as if they are weather. The factory wants 5,000 units, I can only sell 1,200 in the first six months, so I need to raise more money or find a factory with a lower MOQ.

MOQ is not weather. It is a negotiated position, and the position you get offered is a direct function of how much pricing power you gave away before you ever asked the question.

Here is the mechanic. A factory’s MOQ exists because setup, tooling changeover and line time carry a fixed cost that has to be spread across units. That is real. But the number quoted to you is not purely engineering. It is a risk premium. The factory does not know you, has no view of your reorder probability, cannot assess your funding, and is being asked to allocate line capacity to a customer who may vanish. If the conversation is being run through an intermediary who also does not want you talking to the plant directly, nobody in the chain has an incentive to lower that premium. The intermediary is actually better off with a high MOQ: it increases their commission base and locks you in.

I have seen the same product, same specification, same region, quoted at MOQs with a factor of four between them. The difference was never the engineering. It was whether the founder had a relationship with somebody inside the factory who could vouch for them, and whether that person believed there would be a second order and a third.

So when a founder tells me the MOQ is crushing their cash flow, I do not start by looking for a cheaper factory. I ask who they are actually talking to, how many layers sit between them and the production line, and what the factory knows about their business. Nine times out of ten the cash problem is downstream of a structural problem, and swapping suppliers just resets the clock on the same mistake.

The WFOE question is asked too early and backwards

At some point in this process, usually around month eight, the founder asks the question everyone asks: should I set up a WFOE or work through a local partner?

It is asked too early because it is being asked as a legal question when it is a control question. And it is asked backwards because the founder starts from the structure and works towards the business, instead of starting from the business and working towards the structure.

Setting up a wholly foreign-owned enterprise gives you a legal entity you control, the ability to hire locally, to contract directly, to hold IP in-country, to invoice in RMB, to move money within a framework you understand. It also gives you registered capital commitments, a legal representative with personal exposure, annual audits, tax filings, a chop that somebody physically holds, and an ongoing administrative load that someone has to own. It is not expensive relative to what people fear, but it is never free and it is never passive.

A local partner gives you speed, existing relationships, a Chinese-language interface with the supply chain, and someone who can walk into a factory and be taken seriously. It also means that the relationships being built belong to them, not to you, and that every year you operate this way the cost of unwinding it rises.

Neither is right in the abstract. The question that produces an answer is narrower: which part of the value chain am I refusing to outsource?

If the answer is manufacturing quality and engineering change control, you need direct presence at or near the factory relationship, and you probably need an entity sooner than you want one, because you will be hiring a quality engineer and paying them locally. If the answer is the end customer, because you are selling into China and the brand relationship is the asset, then the entity is not optional at all and the partner conversation is about distribution, not about structure. If the answer is honestly neither, if China is purely a cost centre and your value sits entirely in design, software and Western distribution, then a well-structured partner relationship with clean IP terms may serve you for years and the WFOE is a distraction.

Make that decision before you send a single RFQ. Not because the structure needs to exist on day one, but because it determines who you talk to, what you tell them, and what you are willing to concede. A founder who has decided that engineering change control is non-negotiable negotiates a completely different contract from one who has not thought about it.

How deals actually close over there

The other thing that changes when you stop running a procurement process is the behaviour of the people across the table.

I have built and sold in China in three different contexts: a location-based consumer platform that reached over three hundred thousand users, digital transformation for a private healthcare chain opening sites across the mainland, and hardware production. The pattern was consistent in all three. Deals did not close because I presented the strongest commercial case. They closed when the counterparty concluded that I would still be there in three years.

That is not sentimentality. It is rational. The Chinese side is assessing counterparty risk with very little to go on, and the signals they read are different from the ones a Western buyer thinks they are sending. Showing up repeatedly matters. Bringing the same people each time matters enormously, because rotating your contacts resets the relationship to zero. Giving them visibility into your roadmap, including the parts that are uncertain, reads as trust rather than weakness. Paying on time, early if you can, buys more goodwill than any clause. Understanding that the first agreement is a framework and the real terms get settled in the operating rhythm is not naivety, it is accuracy.

Behaving like a buyer gets you treated like a buyer: transactional pricing, generic lead times, your order deprioritised when a larger customer pushes. Behaving like a partner, which mostly means accepting some shared risk and demonstrating durability, changes your position in the queue. When capacity got tight, and it always eventually does, the orders that shipped on time belonged to the customers the plant manager personally wanted to keep.

What to do differently

Before you look for a factory, write down the one or two parts of your value chain you will not let anybody else own. Be honest. For most hardware companies it is quality control and the customer relationship, and almost never the assembly itself.

Then design your sourcing around protecting those. That means knowing the name and location of the actual factory, not just your contact. It means meeting the people who will run your line. It means putting someone on the ground, yours or trusted, before you need them. It means treating the entity question as the consequence of a control decision, not the starting point.

And when the MOQ comes back higher than you can fund, resist the instinct to go shopping for a lower one. Ask instead what the number is telling you about how little the person quoting it knows about your business, and whether that is a gap you can close by building the relationship rather than by changing the supplier.

Finding a factory in China is the easy part. There are thousands. Owning what matters about the relationship is the part that decides whether you keep your margin.

I write from twenty years of building businesses between Europe and Asia. If your company is facing this, start a conversation.