A founder once showed me a slide he was rightly proud of. It had his new WFOE registration number, the bank account, the local director, the VAT-equivalent registrations, a clean org chart with two boxes filled and four boxes dotted. Eleven weeks from decision to legal existence in China. He asked me what I thought of the structure.

The structure was fine. I asked him a different question: who in that market would take his call on a Tuesday afternoon and put his product in front of a buyer this quarter? He named a law firm, a payroll provider, and a man he had met twice at a conference in Shenzhen.

That is not a distribution answer. That is a services answer wearing a distribution costume. Eight months later the entity was still there, still costing money, still filing, still employing one person whose job description had quietly drifted into “attend meetings and report back.” He had built a very expensive filing cabinet.

The pattern: founders solve the question that has an answer

I have watched this happen enough times to stop calling it bad judgement. It is a rational response to information asymmetry.

If you type “opening a company in China” into a search bar, you get answers. Clean ones. WFOE versus joint venture versus representative office. Registered capital expectations. Timeline. The names of six firms who will do it for you, with published fee ranges. Same for a merchant of record if you are selling software across borders. Same for a Gulf-friendly payment processor, same for an EU-only data stack. These questions have documented, comparable, purchasable answers, and answering them feels like progress because something concrete comes out the other end: a certificate, an account number, a contract.

Now type “who will sell my B2B software in Guangdong.” Nothing useful comes back. There is no directory. There is no comparison table. There is no fee schedule. The answer exists only inside a small number of heads, most of which are not writing blog posts, and the only way to find those heads is to spend months talking to people who mostly cannot help you until one of them can.

So founders do the tractable thing first and tell themselves the hard thing will follow. It rarely does. Structure does not attract distribution. Distribution, once you have it, tells you exactly what structure you need, often something simpler and cheaper than what you would have built speculatively.

What I actually learned putting entities on the ground in Asia

I have opened companies in China and shipped consumer hardware out of Shenzhen. Let me be honest about the relative difficulty, because the folklore gets it backwards.

The legal and administrative work was not the hard part. It took weeks, not quarters. It cost less than we had budgeted, because we had budgeted using fear rather than quotes. The paperwork was tedious, the bank onboarding was slower than promised, there were two or three moments where a document had to be re-notarised for reasons nobody could fully explain. Annoying. Not existential. If you hire competent local counsel and accept that you will not enjoy the process, structure is a solved problem with a known price.

Finding the first partner who could genuinely move volume took roughly eighteen months.

Eighteen months of introductions that led to coffee, coffee that led to a factory visit, a factory visit that led to a cousin who knew a regional distributor, and a regional distributor who turned out to serve the wrong channel entirely. Eighteen months of learning that the person with the best English is almost never the person with the best relationships, and that the person who says yes fastest is usually selling you access rather than volume. Eighteen months of discovering that in that market the real gatekeepers sat one layer further back than the org charts suggested.

When we finally found the right counterpart, everything accelerated. The product line went from one to eight over the following period. Not because the structure changed. The structure had been sitting there the whole time, doing nothing, waiting for somebody to feed it.

Later, working on digital transformation for a private healthcare group expanding across the mainland, I saw the same asymmetry from the buyer’s side. New entrants would arrive fully incorporated, fully compliant, fully capable of invoicing us, and completely unable to explain who inside our organisation had asked for them. Procurement was never the obstacle. Demand creation was. The vendors who won were the ones who arrived with a person we already trusted, and that person was often not an employee of theirs at all.

The three names test

So here is what I run before a client spends a single euro on structure. It takes twenty minutes and it has an unfortunate habit of ruining board decks.

Name three people, actual human beings with actual surnames, who are already in that market, who would take your call, and who have sold something adjacent to your product to the buyer you are targeting. Then tell me, for each of them, what they get paid when you win.

Not what you might offer them. What the deal looks like. Percentage of first-year contract value, or a margin on resale, or a referral fee, or equity in a local venture, or an exclusive territory with a volume commitment attached. Money and terms, written down, in a form they would recognise as serious.

Most founders cannot complete this exercise. That is fine. The point is not to pass on the first attempt, the point is to know that you have not passed. If you cannot name three people, you do not have a market entry strategy, you have a market entry aspiration, and pouring an entity on top of an aspiration does not solidify it.

The second part of the test matters more than the first, and it is where I see the most self-deception. Founders name three people happily. Then I ask what those people earn when a deal closes, and the answer is a vague gesture at “we’d discuss partnership terms.” That means the conversation has never happened. A person who has genuinely agreed to sell for you has, at minimum, argued with you about their cut. If nobody has pushed back on your economics, nobody has agreed to your economics.

For cross-border SaaS distribution this gets sharper still, because the temptation to skip humans entirely is real. Self-serve, product-led, credit card in any currency, no local presence required. It works in some categories and some geographies. It works far less often in markets where the buyer expects a relationship, an invoice in local format, a person to shout at when something breaks, and a reference customer they can telephone. In those markets the payments layer is necessary and nowhere near sufficient. I have seen companies obsess over a merchant of record decision for six weeks while the actual constraint, that no salesperson in the region had a reason to care about their product, went unexamined.

Sequence it the other way round

The practical alternative is unglamorous and I recommend it constantly.

Go and sell something first, badly, without infrastructure. Invoice from your home entity. Accept the wire transfer, accept the currency friction, accept that the tax treatment is suboptimal for the first few deals. Pay your local counterpart as a consultant or a referrer rather than as an employee. Do this for three to five transactions. It will be inefficient and slightly embarrassing and you will learn more in one quarter than a feasibility study will teach you in three.

What you get from those first deals is the specification for your structure. You will discover whether buyers actually demand a local invoicing entity or merely say they prefer one. You will discover whether the data residency question is a genuine blocker or a negotiating point. You will discover whether your partner wants a reseller margin or an agency commission, which changes your entity choice materially. You will discover the true sales cycle, which is almost always longer than the number you put in the model.

Then you incorporate, with a design brief written by reality rather than by a checklist. In my experience the structure you end up building after five real deals is different from the one you would have built beforehand, usually leaner, occasionally in a different city, sometimes in a different country entirely.

There is one honest exception. Some markets and some sectors genuinely gate you at the door: regulated financial services, healthcare, defence, anything where you cannot legally take money without a licence. If that is you, structure first is not a mistake, it is the price of entry. Even then, run the three names test in parallel, because a licence with no distribution behind it is the same expensive filing cabinet, just with a regulator attached.

Everyone else: the entity is not the hard part. It was never the hard part. The hard part is a small number of people in a place you do not live, who have no particular reason to spend their credibility on you, and who will only do so once you have given them a reason that is denominated in money rather than enthusiasm. Find them first. The paperwork will still be there, and it will be cheaper than you fear.

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