You find a store. The photos are clean, the price is fifteen percent below the other three you looked at, the seller has been on the platform for twelve years and the page shows a repeat-purchase rate of thirty-something percent.
You feel informed. You are not. What you have is a set of numbers that describe the seller's history with the platform. Not one of them describes what will happen to your order.
1. The badges measure a relationship you are not part of
The basic certification on 1688 — 诚信通 — is essentially an identity check. The platform verified that the company behind the store is a real registered business. That is worth something: it filters out the outright fake listings. It is also the lowest possible bar, because almost every serious seller clears it.
The badges above it mean more, and cost more. The platform sells tiered seller programmes with annual fees, performance requirements, and in some cases an on-site inspection that a third party carries out at the factory. A seller displaying one of those has invested real money and passed a real review.
But read what that review actually covers. It asks whether the seller is a going concern, whether the premises exist, whether they can handle volume. It does not ask whether they will tell you the truth about a lead time, whether they will substitute a cheaper material halfway through your run, or whether they will answer a message after the deposit clears.
And there is a structural problem on top of it: most of the platform's buyer protection applies to orders placed through the platform. If you are a buyer outside China, the conversation usually moves to WeChat, and the money usually moves outside the platform. At that point every badge on the page is decorative. You have taken yourself out of the system that was going to enforce anything.
Twelve years of membership does tell you one real thing: this seller has not blown up badly enough to be removed in twelve years. That is a genuine signal. It is a floor, not a ceiling.
2. The reply is the test, not the storefront
You cannot evaluate a seller from a page they designed. You can evaluate them from how they respond to a question that is inconvenient to answer.
Ask something technical and specific about the exact SKU:
What is the wall thickness of this part, and what material grade is it moulded from? I need it to stay stable at −20 °C in transit — can you confirm the grade holds up?
Three things can come back.
A supplier who knows the product answers with numbers, or tells you the question needs a different product, or says the thickness would have to change and that means tooling. All three are useful. The third is the most useful, because it shows they are thinking about your production rather than your order.
A trader with a good relationship with their factory will say they need a day to check with the plant. Also fine — as long as the answer comes back with a number in it.
What you are filtering for is the reply that says 可以的 — "yes, that's possible" — to every question. It costs nothing to type and it commits to nothing. When someone answers a technical question with pure agreement, they have told you they will also agree to your deadline, your packaging spec and your defect claim, and none of those agreements will mean anything.
A second test that costs one message: ask them to send a photo of how an existing customer's order is packed for export. A seller with real export business has that photo on their phone. A seller who has only ever shipped domestically has to go and ask someone, which you can hear in the delay and the hedging.
3. SKU breadth tells you who you are talking to
Open the store's full product list. Two shapes:
Sixty products, all variations of one thing, with coherent sizing and material options. That is a factory or a specialist.
Four thousand products across kitchenware, garden tools, pet supplies and phone cases. That is a trading company. They buy from factories and resell.
Neither is a problem. This is the part most guides get wrong. For a small overseas buyer, a good trading company is often better than a factory: they will consolidate several items into one shipment, they know export paperwork, they will accept an order below a factory's minimum, and they answer messages faster because answering messages is their job.
The problem is not being a trader. The problem is not knowing which one you are dealing with, because the two fail in different ways. A factory that gets busy stops answering and pushes your lead time. A trader who gets busy cannot push a factory either, so your lead time disappears with no explanation and nobody to escalate to.
So ask directly. "Are you the manufacturer, or a trading company? If trading, which factory do you work with?" A trader who answers honestly in one line has just earned a lot of credibility. A seller who insists they are a factory while listing four thousand unrelated products has told you they will be vague about everything else too.
4. Ask for a sample and watch the process
The sample itself will be good. Samples are always good — that is what samples are for. The information is in how the sample happens.
Four things to record:
Whether they charge, and how much. A sample fee is normal and fair. A sample fee that costs more than the product, non-refundable, is a filter — sometimes a legitimate one, sometimes a way to make money on people who will never order.
Whether the fee is credited against the bulk order. Serious sellers do this without being asked. It signals they expect the bulk order to exist.
What they quote for sample lead time versus what actually happens. This is the single most predictive number in the whole relationship. A supplier who says four days and delivers in four days will usually hit a thirty-day production window. A supplier who says four days and takes eleven has shown you their planning, at a moment when they are still trying to impress you.
Whether they ask you anything. A good seller asks what market the product is going into and what packaging you need, because those change what they build. A seller who ships a sample without a single question is shipping you a catalogue item.
When the sample arrives, take it apart. Cut the seam. Look at the inside of the moulding. Check whether the hardware is the same gauge as the photo. The outside of a sample is a marketing surface; the inside is the product.
5. Cross-check the legal entity, not the storefront
The store page carries company registration information: the full legal name, how long the company has existed, registered capital, registered address. Three cross-checks catch most of the nonsense.
Does the company name relate to the store name? Not always, and there are innocent reasons — a brand name versus a manufacturing entity. But if the store is called one thing, the registered company is called something unrelated, and the registered capital is ten thousand yuan with a founding date eighteen months ago, you are not looking at a twelve-year factory no matter what the badge says.
Does the registered address match where the goods ship from? Same province is normal. Different province, consistently, means the seller is drop-shipping from somewhere else — which for a trader is fine if disclosed and a problem if not.
Is the entity findable at all? China's national enterprise credit system is public and free. A company that has no trace there is not a company.
6. What the review and repeat-purchase numbers actually measure
Repeat purchase rate is the number sellers quote most proudly and understand least. A high rate means domestic small wholesalers buy from this store again. Those buyers order in units of a hundred, they care about price and speed, and they are not asking for the things you are asking for — export packing, English labelling, a defect policy, a certificate.
Reviews are more useful, but not as a score. Read the bad ones and look for repetition of the same complaint. A store with scattered complaints about slow delivery and rude service is a store with ordinary service. A store where four separate buyers describe the same problem — the colour is off, the sizes run small, the material is thinner than listed — has told you exactly how your order will fail.
Ignore volume-based crowns entirely. They measure turnover, and turnover can be bought with promotions.
7. Five questions that separate the two
Send these in one message, in writing, before you discuss price. The answers are worth more than an hour of scrolling.
1. Are you the manufacturer or a trading company? If trading, which factory makes this product?
2. Can you send the specification sheet for this exact SKU — material, dimensions, tolerances, and any test report you already hold?
3. What is your sample cost and sample lead time, and is the sample cost credited against the bulk order?
4. After I pay, who handles my order — you, or someone else? What is their name?
5. If three percent of the units arrive defective, what do you actually do? Please describe the process, not the promise.
Question one and question five carry the most weight, and they work for opposite reasons.
Question one works because honesty is cheap for a competent trader and expensive for a pretender. A trader who says "we are a trading company, the factory is in Dongguan, we have worked with them for six years" has told you their failure mode and made themselves accountable for it.
Question five works because it has no comfortable answer. "We take responsibility for quality" is not an answer — it is a sentence. What you want is a mechanism: send a photo or video of the defect, they replace on the next order, or they deduct, or a third party inspects before shipment at a stated cost. A seller who cannot describe a mechanism has never had a defect claim, which for a factory of any size is not credible.
8. Design the first order so you do not have to trust anyone
Here is the uncomfortable truth that all supplier-vetting advice skips: you cannot establish trustworthiness from the outside. You can only gather evidence that shifts the odds. The decision that actually protects you is structural, not analytical.
Order small, even if it costs more per unit. Paying a 15% unit premium on a pilot order is the cheapest insurance you will ever buy. The price break at ten times the quantity is not a saving if the goods are wrong.
Keep the payment where it can be enforced. If the platform offers escrow or order protection and you move the transaction off it to save a fee, you have traded a small cost for the removal of your only recourse.
Buy an inspection once the value justifies it. Below a few hundred dollars, inspection costs more than the risk. Above a few thousand, a third-party pre-shipment check is one of the highest-return line items in the whole order.
Keep the written record. The chat where they confirmed the material. The quote that stated the carton count. The sample photos with dates. When something goes wrong, what you have is not a grievance — it is a file. The file is what gets a replacement, and it is also what makes the second order cheaper.
You will never know in advance
There is no research path that ends in certainty about a supplier you have not bought from. Anyone who tells you there is — score the badges this way, count the crowns that way — is selling a checklist, not a prediction.
What you can do is make the first order small enough that being wrong is survivable, keep every commitment in writing, and treat the pilot as the actual audition. The supplier who answers your awkward question with a number, hits the sample date they named, and replaces the two units that arrived scuffed is worth more than any store with a wall of certifications.
Everything on the page in front of you is advertising. The order is the evidence.