Consulting, engineering, training, maintenance and software all share a problem that a manufacturer does not have: nothing can be put in a box and sent for evaluation. A factory sends a sample and the sample argues for itself. A service firm sends a presentation, which argues for nothing, and is then asked a question it is usually unprepared for — not « what do you do » but « who, exactly, will be doing it, and will they still be here next year ».
The proof is the people, not the sample
A Chinese client buying a service is buying the competence of named individuals, and they will ask for them by name. Expect to supply profiles of the people who will actually be on the job, to put them in front of the client early, and to be asked what happens if one of them leaves. Expect, too, the request that follows: a small paid piece of work before the main engagement — a diagnostic, a pilot site, a first training session, a scoping study. This is not an attempt to get the work cheaply. It is the sample, and refusing to sell it is refusing to be evaluated. The one thing that genuinely damages a relationship is winning the work with senior people and then delivering it with someone the client has never met; it is read as a substitution, and in substance it is one.
Billing for time is a hard sell
A day rate asks a buyer to approve a meter whose total they cannot see, for work whose necessity they cannot judge, and to defend that to a finance department that will ask what was bought. The objection is not about the amount. It is that time is an input and a buyer wants to purchase an output. The way through is almost always to repackage rather than to discount: cut the work into phases, give each phase a named deliverable and a fixed price, attach a payment to the acceptance of that deliverable, and let the client stop after any phase. You keep your economics, because you costed the phases from the days, and the buyer gets something they can present internally as a purchase rather than as a commitment.
- A named deliverable per phase. A document, a running installation, a trained group, a report with recommendations. Something that exists and can be pointed at when someone asks what the money bought.
- The boundary of the scope, written as exclusions. What is not in the price, said plainly. Scope stated only as inclusions is read as a minimum, and every service dispute begins with something the client assumed was covered.
- What the client has to supply, and by when. Access, data, a room, an interlocutor with authority, their own people's time. Most service overruns are caused by the client side being late, and that is only recoverable if the obligation was written.
- The price of a change, agreed in advance. Not a threat: a rate and a short procedure for additional work, settled while everyone is friendly. Without it, every extra request is a negotiation you will mostly lose, and declining becomes the only alternative to working for free.
- The acceptance criterion. How both sides will know a deliverable is done, and what happens if nobody says anything. A phase with no stated acceptance is a phase that can stay open until the budget year closes.
A proposal that survives a committee
The person you met will not be the only person who decides, and your document will be read by people you have never spoken to — a technical department, a finance department, someone whose job is to find the risk. That makes the proposal a forwardable object rather than a conversation: it has to stand up without you in the room, in Chinese, and it has to make your advocate's internal job easy. The first page is where that is won or lost. What is being bought, what it costs, what the client has to do, when it happens, and who is accountable — on one page, in Chinese, with no marketing. Everything else is annex.
- Find out who signs, and who can stop it. They are rarely the same person and almost never the one who contacted you. A service sale that has not identified the technical objector loses late, for reasons nobody explains.
- Sell a small paid first step before the engagement. It converts an argument about competence into a demonstration of it, and it tells you what the organisation is actually like to work with before you have committed a team.
- Write the proposal in Chinese, with the first page standing alone. Have it read by someone whose job is to explain it internally, and fix what they stumble over rather than what you find elegant.
- Price the phases, not the days, and attach a payment to the acceptance of each one. Ask for the first payment before the first phase starts, which is normal everywhere and especially normal for a supplier with no local entity.
- Settle how you will be paid and invoiced before the work is agreed, not after. Whether the client can accept a foreign invoice, in which currency, against which document, and what they will withhold is a question their own finance department answers — and the answer sometimes changes the structure of the whole deal.
- Agree who owns what is produced, in writing, before the first deliverable exists. A report, a drawing, a configuration and a piece of software each raise the question differently, and it is a cheap conversation before delivery and an expensive one after.
Protecting a method you cannot patent
A way of working is not an invention and no patent office will register it. That does not leave you without instruments, only without a single one. The name of the method can be registered as a trademark, which makes it yours to licence and makes someone else's use of it actionable. The materials — the manual, the templates, the training deck, the code — are protected as works, which means the contract has to say whether the client may reuse them without you, and in practice this is the clause most often left out. Confidentiality and non-circumvention obligations bind the client and the individuals who sat in the room. And disclosure can be staged: a method is at its most exposed when the whole of it is handed over at the start, and least exposed when each phase reveals what that phase needs.
One structural protection is worth more than all of them, and it is not legal. If the delivery requires judgement — a decision taken on site, a calibration that depends on the case, an arbitration between two bad options — then the method can be read and still not be performed. The firms that lose their method to a client are usually the ones whose method was a checklist, and the honest response to that is to accept it: if the work is genuinely reproducible from a document, sell the document.
Do you need a local entity?
Not for the first engagements, in most cases, and the question is usually asked too early. What actually forces the decision is rarely strategy; it is one of three practical facts. The client cannot process a foreign invoice, or can only do so with a withholding and a procedure that makes the deal unattractive on both sides. Or the work requires people present continuously, which raises an employment question a service contract does not answer. Or the buyer is public, or operates under a licensing regime, and the tender simply requires a local supplier. Until one of those appears, the sensible order is to sell, invoice from home, see whether the work repeats, and let the entity follow the revenue rather than precede it. When it does become necessary, treat the tax and invoicing mechanics as a question for a local accountant on your own facts: the way a service payment leaves China, what is withheld and what document the client needs, depends on the nature of the service and on where the work is performed, and it is not something to infer from someone else's arrangement.
Sell the first paid step rather than the engagement. A diagnostic, a pilot or a first session does four things at once: it gives the buyer the sample they cannot otherwise have, it gets your people in front of the people who will judge them, it shows you what this client is like to work with, and it tells you whether the money can actually reach you. All four are better learned on a small invoice than on a large one.