Alibaba, 1688, agents, inspections. Every guide treats these as a price problem. For a brand owner they are an ownership problem, and the decisive question is not what a unit costs but who ends up holding your mark, your tooling and your tech pack.
Annie Chan··16 min read
Almost every guide to sourcing from China is written for someone who is going to resell a product that already exists. Find the item, compare quotes, negotiate MOQ, order an inspection, ship it. That workflow is fine if your business is arbitrage. It is actively dangerous if your business is a brand.
The difference is not sophistication. It is what you are buying. A reseller is buying units. A brand is buying a manufacturing relationship, and inside that relationship are three assets that will either belong to you or belong to your supplier: your trademark in China, your tooling, and your technical specification. Which of those you end up owning is decided in the first ninety days, usually by default, usually without anyone in your company realising a decision was being made.
The sourcing decision and the intellectual property decision are the same decision. You make it before your first sample, not after your first reorder. Almost nobody does.
Why your supplier is your most likely trademark squatter
China operates a first-to-file trademark system. Rights go to whoever files first at CNIPA, not to whoever used the mark first, and not to whoever owns it in twenty other countries. Everyone in this business has heard that sentence. Very few brands have followed it to its uncomfortable conclusion.
Here is the conclusion. In a first-to-file system, the person best positioned to file against you is the person who learns your brand name earliest and knows it is not yet registered. That is your factory. Not a professional squatter trawling foreign trademark databases, though those exist. Your factory, or your sourcing agent, or the trading company that sat between you and both.
They learn the name when you send the artwork for the first sample. Depending on how fast you move, that can be six to twelve months before your own counsel gets around to a Chinese filing, because the Chinese filing feels like a market-entry task and you are not entering the Chinese market. You are only buying from it.
"You do not need to sell in China to need a Chinese trademark. You need one because your goods are made there, and goods that infringe a Chinese registration can be stopped on the way out."
That last point is the one that catches people. A squatted mark is not merely an obstacle to selling in China. It is a lever over your export flow, because the holder can record the mark with Chinese customs and use it against shipments leaving the country. The pressure is applied at exactly the moment you are least able to resist it, which is when you have a container booked and a retail launch date.
Chinese law has tightened considerably against bad-faith filings, and there are specific provisions aimed at agents and representatives who register a principal's mark without authorisation. Remedies exist. Oppositions and invalidations succeed. But every one of those remedies costs more time and money than filing first would have, and several of them cost more than the entire first production run.
The minimum defensive position
File your mark in China before you send artwork to any supplier. Before, not alongside.
File the Latin-character mark and a Chinese-character version. If you do not choose the Chinese name, the market will choose one for you, and then someone else will register it.
File across the classes that matter, which includes the classes your product sits in and the adjacent ones a squatter would use to box you in.
Record the registration with Chinese customs once granted. This is the tool that works in your favour at the border rather than against you.
Do all of this through counsel who practises in China, not as an extension of your home-market filing programme.
The cost of this package is small relative to a first production run and trivial relative to the cost of getting it wrong. It is skipped constantly because it lands in the gap between legal, who are not running the sourcing project, and operations, who do not own trademarks.
1688 versus Alibaba.com is a liability decision, not a price decision
The standard framing is that Alibaba.com is the export-facing platform with higher prices, 1688 is the domestic wholesale platform with lower prices, and clever buyers use 1688. The price observation is accurate. The conclusion drawn from it is usually wrong.
1688 suppliers are cheaper because they are selling into China. They have not priced in export documentation, English-language account handling, international payment terms, trade-show costs, compliance testing, or the overhead of dealing with a foreign buyer who will ask for a REACH declaration or a CPSIA test report. That is not a markup you are cleverly avoiding. It is a bundle of services you are declining to buy, and if your product needs any of them you will rebuild them yourself at higher total cost.
Buy on 1688 when your constraint is discovery. Buy through an export-facing channel when your constraint is documentation. Getting these backwards is the most expensive routine mistake in brand sourcing.
Discovery is a real constraint and 1688 genuinely solves it. The supplier pool is far deeper, the industrial clusters are visible in a way they are not on export platforms, and you can see what a category actually costs to make rather than what it costs to export. For price discovery alone it is worth using even if you never place an order there.
Documentation is the other kind of constraint. If your product touches skin, gets eaten, plugs into a wall, goes near a child, or crosses a border with a regulatory regime attached, you are buying documentation as much as you are buying goods. A supplier who has never produced a compliance file will not produce a good one for you at the price that made them attractive.
What the platform choice does not solve
Neither platform tells you whether you are talking to a factory or a trading company. This matters more than it sounds. A trading company is not automatically bad, and a good one earns its margin by managing several factories and absorbing coordination risk. But you should know which one you have, because it determines who you are negotiating tooling ownership with and whether your specification survives being passed down a layer.
The tells are mundane. Ask for the business licence and check the registered scope of business. Ask which production lines make your item and how many. Ask a question that only a factory can answer, about a process parameter or a tolerance, and see whether the answer comes back same-day or after a suspicious delay.
The five stages, and the one everyone skips
Stage one: discovery
Build a list of eight to twelve candidates, not three. The purpose of a wide list is not choice, it is calibration. You cannot tell whether a quote is good until you have seen the distribution, and the distribution in Chinese manufacturing is wider than Western buyers expect. Two suppliers making an apparently identical item can quote at a two-to-one ratio for reasons that are entirely legitimate and entirely invisible in the listing.
Stage two: qualification
Narrow to four. At this stage you are testing responsiveness and comprehension rather than price. Send a technical question and see whether the answer engages with it. Ask for references in your destination market. Ask what they will not do, which is the single most informative question available and one that separates factories with real capacity constraints from ones that will say yes to anything and subcontract the difference.
Stage three: sampling
Sample from at least two suppliers, always, even when the first sample is good. A single sample tells you what one supplier can do on a job they know is an audition. Two samples tell you what the category can do, which is the information you actually need.
Pay for samples. Free samples come from the shelf and represent nothing. Paid samples made to your specification are the first real test.
Stage four: the pre-production lock
This is the stage that gets skipped, and it is the stage that determines everything downstream. Before the first production order, three documents need to exist and be agreed.
The technical package. Dimensions with tolerances, materials with grades, finishes, packaging, labelling, and the acceptance criteria for each. If your tech pack is a set of photographs and a paragraph, you have not specified a product, you have described one, and the supplier will resolve every ambiguity in the direction of their own cost.
The golden sample, signed and sealed in duplicate, one copy held by each party. This is the physical reference that every later dispute resolves against. Without it, quality is an argument between two memories.
The agreement covering non-disclosure, non-use and non-circumvention, drafted under Chinese law with a Chinese court or arbitration venue named. A standard Western NDA is close to worthless here, and not because of anything sinister: it typically only prohibits disclosure, which is not your risk. Your risk is use, meaning the supplier making your product for someone else, and circumvention, meaning the supplier selling directly to your customers.
Tooling ownership belongs in the same conversation. If you paid for a mold, the contract should say you own it, say where it is stored, say you can take it, and say what happens on termination. Molds that are legally yours and physically theirs are not yours in any sense that matters when the relationship ends badly.
Stage five: production quality control
Covered in its own section below, because most brands do this at the wrong time.
Inspections: the wrong one at the wrong time
Two things get conflated constantly. A factory audit assesses whether a facility is capable and compliant: equipment, systems, labour conditions, certifications. A product inspection assesses whether a specific batch matches specification. They answer different questions and a good result on one tells you very little about the other.
Most brands book a pre-shipment inspection, which happens when production is finished and the goods are packed. If that inspection fails, your options are to accept the goods, to rework them under time pressure at a factory that just missed your standard, or to reject them and have no product. All three are bad. The inspection has correctly identified a problem at the exact moment you have the least leverage to fix it.
A pre-shipment inspection protects you from shipping bad goods. A during-production inspection protects you from making them. Only one of those is worth what it costs.
The during-production inspection, usually booked when the line is somewhere between ten and thirty percent complete, catches systematic errors while there is still a run to correct. On a first order with a new supplier this is not optional. On the third order with a proven supplier you can reasonably move to pre-shipment only.
Specify your acceptance criteria in the inspection brief rather than letting the inspection firm default to theirs. Sampling and acceptance levels follow published standards, and the levels you choose determine what gets caught. An inspection run to a looser standard than your product needs will pass goods that your customers will return, and you will have paid for the reassurance.
One further discipline that costs nothing: send the tech pack and photographs of the golden sample to the inspector, not just the purchase order. Inspectors check against what they are given. Given a purchase order, they will check quantity and obvious defects and tell you a truthful, useless thing.
What a sourcing agent is actually for, and how they get paid
Agents range from indispensable to actively harmful, and the variable that predicts which one you have is compensation structure.
An agent paid a percentage of order value has an interest in a higher order value. An agent paid a flat fee or a retainer does not. That is not a claim about anyone's integrity, it is a claim about incentives, and it becomes visible in small ways: which suppliers get shortlisted, how hard a price gets pushed, whether a cheaper alternative material gets mentioned.
The other question worth asking directly is whether the agent takes any payment from the supplier side. A commission arrangement in both directions is common and is not always disclosed. Ask, in writing, and prefer the arrangement where you are the only one paying.
The part of the agent business that just disappeared
A large share of the value that agents and forwarders sold to small Western buyers over the last decade was consolidation: gather many small orders, split them into parcels, and use low-value entry exemptions to get goods into the destination market without formal customs entry. That mechanic is finished. The United States has indefinitely suspended its de minimis exemption and Congress has repealed the underlying statutory privilege with effect from July 2027, and the EU is removing its own EUR 150 duty exemption on a similar timeline.
The consequence for you is practical. If a partner is still pitching consolidation as a way to reduce your duty exposure rather than your freight cost, they are either selling you something that no longer works or something that will get you an enforcement problem. The full picture on what changed and what it means for landed cost is here.
The agents worth keeping are the ones who have repositioned around what is now scarce: factory qualification, on-the-ground quality management, compliance file assembly, and consolidated bulk shipment into a properly cleared destination inventory position.
The counterfeit problem is a channel problem
Brands worry about their factory making an extra ten thousand units on the night shift. It happens, and the tooling and NNN provisions above are how you address it contractually. But the more common and more damaging pattern is downstream: your product appearing on marketplaces at a price you cannot match, from sellers you have never heard of, sometimes made in the same industrial cluster and sometimes made in your own factory's neighbourhood by a competitor who reverse-engineered a purchased unit.
You cannot prevent this. You can make it slower and less profitable, and the levers that work are unglamorous.
Registered marks in China plus customs recordal, which gives you a mechanism at the point of export rather than a lawsuit in twenty destination markets.
Design registrations in the markets that matter, filed before public launch, because most jurisdictions have novelty requirements that your own product launch destroys.
A component or process in your product that is genuinely inconvenient to copy. This is a product design decision, not a legal one, and it does more work than any of the above.
Marketplace brand-protection enrolment in every channel you sell in, set up before you need it rather than during the incident.
The strategic version of this point: if the only thing separating your product from a copy is the label, you have not built a moat, you have built a target. Categories where domestic Chinese manufacturers have moved from making the product to owning the brand are covered in the market entry framing, and the pattern usually starts exactly here.
Frequently asked questions
Should I buy on 1688 or Alibaba.com?
1688 is Alibaba's domestic Chinese wholesale platform and prices closer to true factory cost because export services are not bundled in. Alibaba.com is export-facing and the higher prices include documentation, international payment handling and compliance capability. Use 1688 when your problem is finding suppliers and understanding real cost. Use an export-facing channel when your product requires compliance documentation, because rebuilding that capability yourself costs more than the markup you avoided.
Do I need a Chinese trademark if I do not sell in China?
Yes. China is a first-to-file jurisdiction, so rights belong to whoever registers first regardless of use elsewhere. Because your supplier learns your brand name early in the process, they or an associate are the parties best positioned to file against you. A third-party registration can also be recorded with Chinese customs and used to interfere with goods leaving the country, which affects you even if you never sell a unit inside China.
Is an NDA enough to protect my design?
Usually not. A standard Western NDA typically prohibits disclosure only, and disclosure is rarely the risk. The risks are use, meaning the supplier manufacturing your product for a competitor, and circumvention, meaning the supplier selling directly to your customers. An agreement covering non-disclosure, non-use and non-circumvention, drafted under Chinese law with a Chinese enforcement venue, is the standard tool. Have it prepared by counsel who practises in China.
When should I inspect?
On a first order with a new supplier, inspect during production while the run can still be corrected, then again before shipment. Pre-shipment inspection alone identifies problems at the point where your only options are accept, rework under pressure, or have no product. Send the inspector your technical package and photographs of the signed golden sample, not just the purchase order.
What is a golden sample and do I really need one?
It is the approved physical reference unit, signed and sealed, with one copy held by each party. Every subsequent quality dispute resolves against it. Without one, quality arguments become a disagreement between two recollections, and you will lose those arguments because the supplier has the goods and you have a deadline.
How many suppliers should I sample from?
At least two, even when the first sample looks good. One sample tells you what a single supplier can produce for an audition. Two tell you what the category is capable of and what a reasonable price looks like, and the gap between two credible samples is usually more informative than anything in either quotation.
The short version
Sourcing guides optimise for unit cost because unit cost is the number everyone can see. For a brand it is close to the least important variable in the system, because the difference between a good quote and a great quote is a few percentage points, and the difference between owning your mark and not owning it is your business.
"Resellers buy units and can change suppliers on a Tuesday. Brands buy a relationship, and the terms of that relationship are set in the first ninety days by whoever bothered to write them down."
File the trademark before you send the artwork. Write the tech pack before you place the order. Own the tooling in a contract that names where it is stored. Inspect while the line is still running. Everything else in sourcing is negotiable, and those four are the ones that quietly decide whether you have a supply chain or a hostage situation.
Share:
Free · 24 pages
Read Chapter 1 of the Playbook
AI Foundation Models — how DeepSeek and Qwen crossed a billion downloads while Western boards were counting Mag7 capex, and which layers of the AI stack the rotation has not reached. No excerpt tricks: it is the full chapter.
You will also get the weekly newsletter. Unsubscribe any time — the link is in every email.
The Book · $99
China Market Entry Playbook 2026
456 pages across 22 industry chapters — AI, semiconductors, EVs, payments, logistics, beauty, F&B, healthcare, pets, toys. Every chapter runs the same five questions: the anchor number, the rotation underneath it, which layers of the global stack it has reached, and what to do about it. English and Simplified Chinese editions included.
Ex-Transsion Global Digital Marketing Director · Effie Awards Greater China jury member · Guest lecturer, Peking University School of Economics · Writes Annie Chan Talk, your insider lens on China.
