The first time I walked into the Canton Fair complex in Guangzhou with a real product brief in my bag, I had a list of forty-two companies I wanted to meet. I got through nineteen. By the end of day two I had eleven business cards I cared about, a phone full of blurry photos of injection-moulded housings, and the growing suspicion that I had spent two days being sold to rather than doing any selecting.
That was the lesson, and it cost me a year to fully absorb it. Western buyers land in Guangzhou or Shenzhen or Yiwu with a shopping-trip mindset: find the good supplier, shake hands, place the order. Then the golden sample arrives and it is beautiful, and eleven weeks later the first production run arrives and the surface finish is different, the tolerance on one part has drifted, and the person who answered your emails within four minutes in March now takes six days in July.
The fair is not where you find your supplier. The fair is where you eliminate the ones who will hurt you. I go in expecting to disqualify roughly forty companies for every one I keep on the list, and I have never once regretted being ruthless early. It is much cheaper to be wrong in a booth than to be wrong on a container.
The ninety-second test
Half the booths at any large Chinese trade fair are trading companies presenting themselves as factories. This is not fraud, it is a business model, and sometimes a good trading company is exactly what a small European importer needs: they aggregate small orders, they speak better English, they handle export documentation you have never seen before. But you must know which one you are standing in front of, because the failure modes are completely different.
You can usually tell in ninety seconds, and you do not do it by asking “are you a factory?” Everyone is a factory when you ask that question.
I ask three things instead, and I ask them casually, as small talk, while looking at a product on the table. First: how many production lines do you run for this category, and how many people on each line? A factory answers instantly and with a shape, something like “two lines, thirty-eight people, we run one shift and add a second from August.” A trading company gives you a range, or gives you a number that is suspiciously round, or looks sideways at a colleague. Second: what is the largest single order you have run for this exact product in the last twelve months, and what was the cycle time? Third, and this is the one that does the most work: which part of this product do you not make yourselves?
That third question is a gift. It is disarming, because it invites an honest answer rather than a boast, and every real manufacturer has a clean answer: they buy the power supply, they outsource the anodising, they source the PCBA from a partner two hours away. A trading company either says “we make everything ourselves,” which is never true for a consumer product with more than a dozen components, or it produces a vague answer about partners.
Then I stop talking and let the interpreter finish, and I watch. This is the part that people who negotiate only in English miss entirely. Working through an interpreter is not a handicap, it is an advantage, because it buys you four to six seconds of silence in which you can watch two people on the other side of the table decide what to tell you. I learned to hire my own interpreter rather than accept the booth’s helpful English-speaking sales manager. It is a few hundred euros a day. It is the single highest-return expense on the whole trip.
The fastest quote is usually the one to drop
There is a reflex among first-time buyers to reward responsiveness. The booth that emails a full quotation the same evening feels eager, professional, hungry for your business. In my experience that quote is the least reliable number in the pile.
A real quote requires someone to price the bill of materials against current copper, resin and freight, to check tooling capacity, to look at the mould complexity, and to decide what margin the factory needs at your volume. That takes days, not hours, unless the factory already builds something almost identical, in which case they will tell you so and show you.
The same-day quote usually means one of three things. It is a catalogue price with a small adjustment. It is a number designed to win the order, to be renegotiated after you have committed and after tooling has started, which is the moment you have the least leverage in the entire relationship. Or it is a trading company applying a markup to a factory price they already hold. None of those three is a good foundation.
So I now do the opposite of what feels natural. I ask for the quote to come with its assumptions written down: which material grade, which surface treatment, what packaging, what MOQ, what the price does at half the volume and at triple the volume, and what happens if resin moves ten percent. A supplier who can produce that document has a cost model. A supplier who cannot has a hope.
Price is the last conversation. Tooling is the first.
Here is the conversation I have before I discuss unit price at all, and it is the one that separates people who have shipped hardware from people who are about to learn.
Who owns the tooling?
When you build a consumer product with custom parts, you pay for moulds. Depending on complexity, a set can run from a few thousand euros to well into six figures. The invoice says you paid for it. That does not mean it is yours, and it certainly does not mean you can move it. I have seen European companies discover, at the exact moment they wanted to leave a supplier, that the moulds were sitting in a factory in Dongguan and that removing them would be a negotiation, not a collection.
So the first conversation covers: does the tooling agreement say the moulds are my property, is that in the Chinese-language version and not only in the English one, where physically will they be stored, who maintains them, what is the guaranteed shot count, what happens at end of life, and what is the written process if I want them shipped to another factory. Then, separately, whether the supplier will accept a mould transfer clause at all. Some will refuse. That refusal is extremely useful information, and it arrives before you have spent anything.
The reason this comes first is simple. Unit price is a number you can renegotiate every year. Tooling ownership is the thing that determines whether you have any negotiating position at all. When I was building consumer hardware in China, the product line that started as one SKU and grew into eight over several years only worked because we could move production without starting from zero. Every SKU we added was leverage we controlled, not leverage the factory controlled.
The three weeks after you fly home
The fair is theatre. Everyone is on their best behaviour, the booth is rented, the samples are hand-finished, the person you meet may not work for the company you think they do. Nothing you learn in a booth is a decision. It is a hypothesis.
The real qualification happens in the three weeks after you land back in Milan or Munich or Madrid, and it is boring work, which is why so few people do it.
I send the same technical query to every shortlisted supplier on the same day, something specific enough to require an engineer to answer, not a salesperson. Then I measure two things: how long the first reply takes, and whether the answer came from someone who understands the product. A twelve-hour reply from an engineer beats a two-hour reply from sales every time. I ask for photographs of the actual production floor with a specific machine visible, taken this week, not the marketing deck. I ask for the business licence and I check the registered capital, the registered scope of business, and the date of establishment against what I was told at the booth. A company that told me it had been manufacturing since 2009 and shows a licence issued in 2021 is not necessarily lying, but it now owes me an explanation.
Then I visit. Unannounced is unrealistic and rude, but announced with two days of notice rather than three weeks tells you a great deal. On the floor I look at three things that nobody prepares: the state of the incoming goods inspection area, whether the workers are wearing the same uniform as the ones in the photos I was sent, and whether the QC records on the wall are filled in for the last month or only for the last three days.
Somewhere in those three weeks, most of the shortlist eliminates itself. That is the whole point. Forty booths become eleven cards, eleven cards become four serious conversations, four become one supplier and one credible backup you keep warm with small orders, because single-sourcing a product you depend on is a decision to hand someone else control of your business.
Go to the fair. Walk twenty kilometres, drink the tea, collect the cards. Just be clear with yourself about what you are doing there. You are not shopping. You are running a filter, and the value of the trip is measured by how many names you cross out, not by how many you write down.
I write from twenty years of building businesses between Europe and Asia. If your company is facing this, start a conversation.