Remote Market Runs in China: We Walk the Stalls So You Don't Have to Fly In
You run a shop, a stall or an online store in the United States, Australia or Southeast Asia. You buy from a middleman who buys from Guangzhou's Shisanhang, Shenzhen's Huaqiangbei or Yiwu. You would like to cut that layer out — but you cannot close the shop for three weeks and fly over to look at twenty stalls. So we go instead. You send the list; we walk it stall by stall and come back with video, quotes, minimum order quantities and samples.
- This is not a shopping service. Nobody buys anything for you on the first pass. A market run is a survey: we walk the stalls you name, film what is actually on the shelf, ask the price, the minimum order and the lead time, and send it all back in a form you can compare.
- You are paying for a pair of hands and a phone. The deliverable is evidence — continuous video at the counter, a written quote per stall, MOQ per stall, and physical samples where the stall will release them.
- It is priced per stall, per day or per market, and you can see that number next to what a trip would have cost you, because a twenty-stall survey is a fixed, quotable job and a flight is not.
- Two markets have different physics. Shisanhang and Huaqiangbei are dense, vertical and hostile to browsers; Yiwu is enormous, flat and organised by district. The stall-per-hour rate is not the same in both.
- The point is to arrive at the order with the numbers already settled. Whether you then place it through us, through a forwarder, or fly over once and sign in person, you will be negotiating against a list instead of against a smile.
Why the middleman survives, and what it actually costs you
Nobody keeps a middleman because they enjoy paying him. The layer exists because someone has to walk into a building with forty thousand stalls in it, find the four that stock your item, and argue in a dialect about whether the price moves at two hundred pieces. That labour is real. The question is only what it is worth, and whether you are paying it twice.
What the middleman is actually selling you
Strip the mark-up down and a sourcing agent, a buying office or an importer's China desk is doing four things. First, search: finding which of the thousands of stalls actually hold your category, which is not obvious from the street. Second, language: quoting, haggling and confirming specifications in Mandarin or Cantonese, at a speed a translation app cannot match. Third, verification: standing at the counter, looking at the goods, and deciding whether the stall can actually deliver a repeat order. Fourth, consolidation: collecting from eight stalls, packing it together, and moving it to a forwarder.
Of those four, the first three are exactly what a remote market run does. The fourth is a logistics question with well-established answers, covered honestly in its own section rather than passed off as something we invented.
The arithmetic of a stock-up trip
A buyer who flies to Guangzhou to survey stalls is looking at a round trip, a hotel, meals, local transport, a translator or a guide with sourcing experience, and — the part people forget — the cost of their own absence. For a shop owner, three weeks away is three weeks of a closed or thinly-staffed business. Put those on one side of the ledger and twenty stalls on the other, and the comparison is rarely close.
| Line | Flying in yourself | Remote market run |
|---|---|---|
| Flights, return, economy | US$900–1,800 depending on origin and season | Not incurred |
| Hotel, 10–14 nights | US$600–1,600 for a mid-range business hotel | Not incurred |
| Local transport, metro and taxis | US$150–300 across three cities | Included in the day rate |
| Interpreter or sourcing guide | US$150–350 per day, and you still pay for days you spend in transit | That is the service itself |
| Visa, insurance and administration | US$100–250 depending on nationality | Not incurred |
| Your own time off the shop | 10–14 days, priced at whatever your counter earns in a day | A briefing call and a video review, hours not weeks |
| Stalls surveyed | Whatever you can physically reach before your feet stop cooperating — usually 12 to 25 | Whatever is on the list, sequenced by someone who has walked the building before |
Read the last row twice. A trip has a natural ceiling that has nothing to do with how many stalls you needed to see: it is how many you can walk past before the day ends. A remote run has no such ceiling except the one you write on the list, because the person walking is paid to walk and has no flight to catch.
The honest counter-argument
If you are placing a large first order with a manufacturer — tooling, moulds, a private label, a production line that has to be audited — a remote run is not a substitute for being there. That is factory work, not market work. What a remote run replaces is the aisle trip: the reconnaissance sweep that tells you which four of twenty stalls deserve your attention at all.
Where this fits in the sequence. Most buyers do it in this order: remote run first to find the four stalls worth knowing, samples second, one focused trip third — and that trip is now three days instead of three weeks because the shortlist already exists. Skipping straight to the flight is how people spend a fortnight and come home with a stack of business cards and no comparable prices.
The three markets, and why they cannot be walked the same way
Shisanhang, Huaqiangbei and Yiwu get named in the same breath by buyers, as if they were three branches of one shop. They are not. They differ in verticality, in how stalls are numbered, in whether a stall will sell you twenty pieces or two hundred, and in how much of the price is real. A survey plan that treats them as interchangeable produces a wasted day in each.
Guangzhou Shisanhang — the clothing and fabric maze
Shisanhang is not one building. It is a district of a dozen or so wholesale malls along a stretch of road, each with its own speciality and its own floor logic: one mall for knitwear, one for denim, one for accessories, one for fabric by the metre. Floors inside a single mall often split by category, not price, which is why a buyer who walks in blind spends the first hour in the wrong building entirely.
The characteristic behaviour here is that price is quoted per piece and falls steeply with quantity, and that many stalls will not sell below a small minimum at all — they are wholesale-only and will simply wave you on if you ask for three. Fabric sells by the metre with a minimum roll, and the roll is where the real cost sits. This is a market where the same garment appears in four stalls at four prices and the difference is mostly fabric weight, which nobody tells you unless you ask.
Shenzhen Huaqiangbei — vertical, electronic, fast
Huaqiangbei is the electronics counterpart and it is laid out differently again: several towers, each with a fixed identity — one for components and integrated circuits, one for consumer gadgets, one for LEDs, one for repair and rework. The floors are dense and the stock turns over fast enough that a component list written last month can be obsolete this month.
Two things make Huaqiangbei different in kind. First, specification is the negotiation: the price is almost secondary to whether the part is genuine, refurbed, a drop-in clone or a bin-sorted pull. Second, the unit of sale is small. This is one of the few wholesale environments in China where a stall is genuinely happy to sell you ten pieces, which makes it far friendlier to a small buyer than the clothing malls are.
Yiwu — enormous, flat, organised
Yiwu International Trade City is the largest small-commodity market in the world and it is organised in a way the other two are not: five districts, each with a theme, each laid out in numbered streets and aisles. You can, with a map and discipline, navigate it. That is unusual; most Chinese wholesale malls defeat navigation on purpose.
The behaviour to expect: many stalls are display-only showrooms for a factory elsewhere, so the thing on the shelf is a sample of a sample and the real MOQ is set by a workshop you will never see. Yiwu is also the most foreign-buyer-aware of the three — more English, more card terminals, more stalls accustomed to shipping single cartons abroad. That convenience has a price attached, usually in the quoted unit cost.
| What matters | Shisanhang (Guangzhou) | Huaqiangbei (Shenzhen) | Yiwu |
|---|---|---|---|
| Layout | A dozen malls on one road; floors split by category | Several towers, each a fixed speciality | Five themed districts on a published grid |
| Typical stall MOQ | Often wholesale-only; per-piece minimums bite | Frequently low — tens of pieces is normal | Mixed; showroom stalls route you to a factory MOQ |
| Negotiation axis | Fabric weight and quantity band | Genuineness and grade of the part | Unit price against carton size |
| Stalls a competent run covers in a day | 12–18 if pre-mapped; far fewer if not | 10–15, because each counter takes longer | 15–22, because the grid is walkable |
| Foreign-buyer friendliness | Low — expect little English | Low to moderate, technical English is better | Highest of the three |
| The trap | Same garment, four fabrics, one price tag | A "new" part that is a refurbished pull | A showroom price that a factory will not honour |
If you are going to come anyway. Half the value of a remote run is that it tells you whether the trip is worth taking. If the survey comes back with four stalls worth visiting, you book a short trip and walk in with a route; if it comes back thin, you have saved the flight. Should you decide to come in person, the guided in-market days we run for Yiwu and for Huaqiangbei are the natural next step — built for a buyer who is standing there, not watching a screen.
What a market run actually is, minute by minute
A service you cannot picture is a service you cannot buy. So here is the whole thing, stripped of adjectives: what happens from the moment your list arrives to the moment the quote sheet lands in your inbox, and what is deliberately not part of it.
The four things you get, and nothing else
Every run produces the same four artefacts. If a provider cannot name their four, you are buying a vague promise with a day rate attached.
- A clip per stall. Continuous footage shot at the counter, not a highlight reel — long enough that you can see the shelf behind the counter, the quantity on it, and the person answering.
- A quote line per stall, in writing. Unit price at your stated quantity, the price at the next quantity band up, the minimum order the stall will actually accept, the lead time, and whether packing is included.
- A verdict per stall. One of three, and the page uses the same three words throughout: can stock for a stall that will sell at your size and can repeat, needs a call for one where something is unresolved, and walk away for one that cannot do what it claims.
- Samples where the stall will release them. Paid, receipted and shipped to you — or refused, and the refusal recorded, because a stall that will not release a sample at your quantities is telling you something useful.
The sequence, from list to quote sheet
- You send the brief. The list of stalls, products or categories; the market; the quantity you are realistically buying at; and any hard constraint such as a certification or a size run. A stall number is ideal; a category plus a photo of the product is enough to start.
- We map it. The list gets sequenced by building, floor and aisle so the day is a walk rather than a search. Anything unreachable in the time available gets flagged before we go — at this stage, not afterwards.
- We walk, and we film. At each stall the same four questions go out in the same order. Uniform questions are what make the resulting table comparable; a stall visited with a different set of questions is a row you cannot use.
- We push back once. The first price at a wholesale counter is an opening position. A run includes one round of negotiation per stall, and the sheet shows both numbers so you can see what moved and what did not.
- We photograph the paperwork. Business licence or stall card where it is offered, the WeChat of the person who quoted, and the sample receipt. This is what lets you follow up directly.
- You get the bundle. Clips, the quote table, verdicts and sample tracking, delivered together rather than as a stream of messages, so you can open one document and compare.
- You choose the next move. Order, sample further, ask us to go back with a narrower brief, or stop. Nothing is ordered on your behalf in this phase, and no money changes hands with a stall without your explicit instruction.
What is deliberately not included
Four exclusions, stated plainly because a service that hides them is a service that will surprise you later.
- No purchasing on the first pass. The first run is evidence. Buying comes after you have read the sheet and decided, and it is a separate instruction with separate costs.
- No quality inspection of a production run. Looking at the sample on the shelf is not the same as inspecting a batch before it ships, and batch inspection is a factory-level job with its own protocols.
- No legal or customs advice. We tell you what the stall says and what the paperwork shows. Duty, tariff classification and import licensing depend on your country, and that is your broker's territory.
- No guarantee of a price. A quote from a stall is a quote from a stall. It can move with material cost, which is exactly why the sheet carries the date on every line.
One row per stall, twelve columns: stall identifier, market and floor, product as photographed, price at your quantity, price at the next band, minimum order, lead time, packing included or not, sample released or not, the verdict, the date, and the link to that stall's clip. Twelve columns is not decoration — it is the minimum set that lets two rows be compared without opening anything else.
The verdict column is the one most providers leave out, and the one that saves the most time, because sorting twenty rows into three buckets is the entire decision you were trying to make.
How to write the stall list
The quality of a market run is set before anyone leaves the office, by the list. A good list is specific about the product and generous about the stall; a bad list is the reverse. Twenty stalls named by a real product beat fifty named by a vague category, and the reason is arithmetic: every stall on the list costs time, and time spent on a stall that could never have supplied your item is time not spent on one that could.
What a usable list entry contains
Four fields. If you have all four for a stall, the run can proceed with no further questions. If you have only the first, that is still workable — it just costs more hours.
A photo of the actual item you want, not a category name. "Winter jackets" sends us looking at forty stalls in three buildings. A photo of the jacket you are already selling — ideally with the label visible — sends us to the four stalls in one building that stock that construction.
If you are currently buying from a middleman, photograph what he sent you. He is buying it from one of the stalls on your list, and the photo is the fastest route to finding which.
The number you would actually order, not the number you would like to be able to order. A stall that will sell at a hundred pieces is a different stall from one that starts at a thousand, and if we ask with the wrong number we get a price that is real but irrelevant to you.
Give a band: "I want 200, I could stretch to 500 if the price drops meaningfully." That single sentence changes what we ask and what we report back.
Whatever would make you reject a stall regardless of price. A certification your market requires, a size range, a voltage, a material you cannot sell, a packing format your shipping method demands. These are written down and asked at every stall, so the sheet comes back pre-filtered instead of pre-sorted by you.
A stall number, a building name, a business card from a previous trip, a photo of a shopfront, or the name of the person who quoted you last time. Any of these collapses the search step and buys back hours. If the entry has a stall number and the other three fields are missing, we will come back and ask — a stall number with no product is a stall we cannot usefully visit.
How many stalls is a run?
There is no fixed bundle, because the honest unit of a run is the hour, and hours vary by market. What can be said precisely is how many stalls a competent walker covers in a working day in each market, which is the number you should budget against.
| Market | Stalls per full day | Minutes per stall | Note |
|---|---|---|---|
| Yiwu | 15–22 | 20–30 | Fastest of the three: pre-mapped districts, and the twenty-minute end of the range applies to short stops |
| Shisanhang | 12–18 | 25–35 | Slower if the list crosses several malls on the same road |
| Huaqiangbei | 10–15 | 30–45 | Each counter takes longer; technical questions and authenticity checks |
| Two markets in one day | Not advised | — | Shisanhang and Huaqiangbei are in different cities; even within Guangzhou, cross-mall days lose hours to transit |
These are planning figures, not a promise, and the spread matters more than the midpoint. A run whose stalls are all in one mall on one floor with numbers already known will beat the top of the range. A run whose list is "find me suppliers of X" with no names will finish the day below the bottom of it, and should be quoted as a search day rather than a survey day.
Send the list before you send the money. The single most useful thing you can do is send the draft list and let us tell you what it will actually take. A list that reads as twenty stalls on paper sometimes resolves to eleven once the buildings are laid out — and sometimes to thirty, because three of the categories are in a mall you did not know existed. That conversation costs nothing and prevents the usual disappointment, which is a survey that ran out of day at stall fourteen.
The twenty-stall run, walked
Here is the same job you described, worked through as it actually happens: a small buyer in Australia wants women's knitwear, currently taking a 35% mark-up from an importer in Melbourne, and has a list of twenty stalls across two Guangzhou malls. The run below is what the day looks like from the first door to the last clip.
The brief that came in
Knitwear, women's, autumn and winter. Target order 300 pieces per style, three styles. Must be a machine-washable wool blend, no pure acrylic, because the buyer's customers have complained about pilling. Ships to Melbourne. Twenty stalls named, all in two malls on Shisanhang road, eight with stall numbers from a previous trip and twelve known only by mall and floor.
The day, hour by hour
- 08:40, on site before the stalls open. The twenty entries get sequenced by mall, floor and aisle. Eight have numbers, so they anchor the route; the twelve without get pinned to the departments their category lives in. The plan: fourteen stalls before lunch, six after.
- 09:10, first counter. The same four questions go out in the same order. Price at 300. Minimum. Lead time. Will you release one sample at that size. The answer arrives in ninety seconds and the camera is rolling before the third question.
- 10:20, the first rejection. Stall four stocks the right construction but will not break a carton below 600 pieces. That is not a "no" — it is a needs a call, because at 500 across two styles the number may work. It is recorded, not discarded.
- 11:05, the fabric question. Three stalls quote the same garment at visibly different prices. Asked directly, the difference is wool content: 30%, 45% and 60%. This is the single fact a video alone would have missed, and why a sample gets requested at each of the three rather than one.
- 12:00, lunch, and the list is re-ordered. Two of the twelve unnamed entries resolved to the same stall. That is a real finding: when a category has fewer physical sources than you assumed, the negotiating position changes. The two freed slots go to a fabric stall that appeared on a walk-through.
- 13:30, the negotiation round. Back to the six best counters for one more conversation each. Four prices move, two do not. Both outcomes are recorded — a price that will not move at 300 pieces is information about that stall's scale.
- 15:10, the sample requests. Five stalls release a sample at the buyer's quantity; three refuse. The refusals go in the sheet with the reason given, because a stall that will not release a sample at 300 pieces is unlikely to serve a 300-piece order well.
- 16:00, paperwork. Business cards and WeChat handles for the five releasing stalls, business licences where offered, sample receipts for the four that took payment.
- 17:20, second mall, six remaining stalls. Walked in forty minutes because the anchors were already known. Two more can stock verdicts, one walk away — the stall claimed a wool blend and the label said 100% polyester — and three needs a call entries.
- 18:00, clips uploaded. Twenty stalls, twenty clips, the last one shot minutes earlier. Nothing is edited into a highlight reel; the buyer gets the counter as it sounded.
What came back
Twenty stalls surveyed, six shortlisted, five samples in transit, and a sheet the buyer could read in fifteen minutes. The middleman's 35% was not eliminated that day — that takes a first order and a shipping decision — but the number it was hiding was. Two stalls could match the Melbourne importer's landed cost at less than half his mark-up, and one of them was the stall the importer was almost certainly using.
45% wool blend, the construction in the photo. Quoted per piece at 300, moved once on the call. Will release a sample for a small fee, receipt provided. Lead time quoted in working days, packing included.
Why it made the shortlist: the fabric matched the brief on the first touch, the counter could show the same garment on the shelf behind them, and the stall quoted a price at the buyer's actual quantity without changing the subject.
Correct fabric, correct construction, but the counter will not confirm a lead time inside the buyer's window without a supervisor. Left with a contact and a note to call back.
Why it is not a rejection: everything except the timing was right. A stall whose only open question is a date is worth one more conversation.
Quoted the lowest price of the day. Label on the goods read 100% polyester against a spoken claim of a wool blend, and the counter would not open a sealed carton to show more.
Why it is the most useful row in the sheet: the cheapest quote and the least honest one landed on the same stall. A survey that only recorded price would have sent the buyer straight at it.
What you would pick
Sorting the twenty rows into the three verdicts is the decision you were trying to make, and it is done in one pass. The chips below are the six shortlisted stalls — turn them on and off to see how the shortlist is the only part of the sheet that ever needs a second look.
Three of the six are switched on, matching the three stalls that cleared every question in the brief on the first pass.
What it costs, and what it saves
A day rate on its own tells you nothing. The number only means something next to the cost of the alternative, which is a flight, a hotel and two weeks off the shop. So the page quotes both, side by side, and lets you do the subtraction yourself.
How the service is priced
Three units, and you can buy any of them. In practice most buyers choose by the shape of the job rather than by price: a stall list is priced per stall, an open search is priced per day, and a broad multi-market sweep is priced per market.
- Per stall. A named list up to a stated count, each stall walked, filmed, quoted and given a verdict. The cleanest unit, because the deliverable is countable.
- Per day. An eight-to-nine-hour working day in one market, with the list built on the day. Right when you know the category but not the counter.
- Per market. A multi-day sweep across two or three cities, sequenced so travel time is not billed as survey time.
- Add-ons, quoted separately. Sample purchase and onward shipping at cost plus a handling fee; a return visit to a shortlisted stall.
- A return flight from the US, Australia or Southeast Asia, at whatever your season is charging — and a shop owner pays the retail fare, with no corporate rate behind it.
- Ten to fourteen nights of hotel in a business district, in a city where a trade-show week can double the rate.
- A guide or interpreter per day, including the transit days on which nothing is surveyed at all.
- Your own absence. For a one-person shop, the counter is closed or the online orders go unanswered.
| Question | Do it yourself | Remote market run |
|---|---|---|
| How long until you have the numbers? | Two to four weeks — booking, flying, acclimatising, walking | Days, once the list is agreed |
| How many stalls can you realistically see? | 12–25 across a whole trip, limited by stamina and geography | 15–22 per day, on a list you set |
| What happens when a stall is wrong? | You find out in person, after paying to get there | You find out on video, and it costs one row of the sheet |
| Do you get comparable prices? | Only if you ask every stall the same questions in the same order | Yes, by construction — the questions are fixed before the day starts |
| What does your business do meanwhile? | Runs without you, or not at all | Unchanged |
| When is flying in still right? | When you are signing a factory, approving tooling or placing a large first order | — a survey is not a substitute for that, and this page does not claim it is |
The part that is genuinely hard to price
Two costs never make it onto a comparison table. The first is the price of not knowing: a buyer who has never seen the stall behind their supplier negotiates against a question mark, and the mark-up survives partly because the alternative looks expensive and uncertain. The second is the cost of a bad first order: a container of the wrong fabric, discovered after it lands, costs more than a year of surveys.
You send the list. We count the stalls, check that they are walkable in the time available, and quote a number and a duration. You see both before anything is booked.
If the list is larger than a day can hold, the quote says so and offers the honest options: two days, a narrower list, or a search day to find the stalls and a survey day to price them. What it will never do is take the money for a day and quietly walk fourteen of your twenty stalls.
The video and the quote sheet
Everything the run produces exists to answer one question at a distance: is this stall the one? Video answers what the goods look like and what the counter sounds like. The quote sheet answers what it costs and whether it will deal with you at your size. Neither works alone.
Why the footage is unedited and continuous
A highlight reel is a sales document, and you are not buying from us — you are buying from the stall. So the clips are shot the way a witness would shoot them: one continuous take at the counter, no cuts, no music, questions audible. That format is harder to watch and far more useful, because the tells that matter are in the parts an editor would remove — the pause before a price, the glance at a colleague, the hesitation when a lead time is quoted.
Three details in the footage are worth more than the rest. The shelf behind the counter, because it shows what the stall actually holds rather than what a sample wall displays. The quantity on that shelf, because a stall with four units is a sample room, not a stockholder. And who is answering — an owner, a family member or a hired hand, often the difference between a price that holds and one that evaporates.
What the quote sheet looks like
Twelve columns, one row per stall, and no column is optional. Two of them do most of the work: the next-band price, which tells you how much cheaper the goods get if you double your order, and the minimum order the stall will actually accept, which is the number that decides whether the stall is available to you at all.
| Column | What it holds | Why it is on the sheet |
|---|---|---|
| Stall ID | Market, mall, floor and stall number | Without it, a row cannot be re-found on a later trip |
| Product as seen | The item, described as photographed | Stops a later misunderstanding about which variant was quoted |
| Price at your quantity | Per piece, at the number you gave | The only price that is actually relevant to you |
| Price at next band | Per piece, at the next break | Shows whether your order size is leaving money on the table |
| Minimum order | What the stall will actually accept | Decides availability, independent of price |
| Lead time | Working days, as quoted | Sets whether the goods arrive for your season |
| Packing | Included or excluded, and in what unit | Packing is where quoted prices quietly diverge |
| Sample | Released or refused, and at what charge | A refusal at your size is a signal about the stall's scale |
| Verdict | Can stock, needs a call, walk away | The decision, pre-sorted into three buckets |
| Date | The day the quote was captured | Quotes age; a date makes the ageing visible |
| Clip link | The continuous take for that stall | Turns an abstract row back into a counter and a voice |
| Contact | WeChat, phone or card, if given | Lets you follow up directly, on your own timeline |
Reading the sheet in fifteen minutes
Three passes, and then stop. First pass: filter by minimum order. Anything whose minimum is above what you can buy is not a row, it is a note — it tells you the stall exists and is out of reach for now. Second pass: sort by verdict. The can stock rows are the shortlist; the needs a call rows are a phone call each; the walk away rows are why you did not fly over. Third pass: read the clips of the shortlist only. This is the whole point of a survey done properly — you spend your attention on four stalls instead of forty.
If the shortlist happens to be in Yiwu, the in-person option is a different service. Yiwu's published grid makes it the one market where a buyer who does decide to come can be genuinely effective in a couple of days. When that is the case, the guided market days we run for buyers who are physically there are the right next step rather than another remote run — and there is a page that sets out how those in-market Yiwu days work if the survey points you toward booking the flight after all.
MOQ, samples and the first order
The survey ends with a sheet. The order begins with three decisions that the sheet was built to inform: which stall, what size, and what has to be verified before any money moves. This section is about the second half of the job, because a survey that leaves you holding a spreadsheet has done only half of what you paid for.
Minimum order: the number that decides everything
Every stall has a minimum, and it is usually not the one printed on a card. It is the smallest quantity at which the person behind the counter considers the transaction worth their time, and it moves with the season, the stock on the shelf, and whether you are asking on a slow Tuesday. Three things are worth knowing.
- It is negotiable, once. A stated minimum of 600 often becomes 400 if you are buying two styles and paying promptly. What does not happen is 600 becoming 60 — the gap between a wholesale counter and a retail one is real, and a stall quoting wholesale prices at retail quantities is a stall to be suspicious of.
- It differs from the factory minimum. A showroom in Yiwu may quote you a stall minimum of 200 and then route the order to a workshop that will not run below 2,000. The stall minimum is the one you can transact at; the factory minimum is the wall behind it.
- It is per style, not per order. 300 pieces sounds like a modest order until you learn the stall means 300 per colour per style, which is 2,700 across three colours and three styles.
Samples: what to do with them when they land
A sample is not a souvenir; it is a test with a protocol. The buyers who get value from samples run the same three checks every time, and it takes twenty minutes.
- Weigh it against the sheet. If the stall quoted a 45% wool blend, the sample goes to a fibre test if the order is large, or at minimum gets compared by hand against the sample that came from the other stall quoting 45%. Two stalls claiming the same content at different prices is exactly the case a sample was requested for.
- Wash it, if it is textile. Pilling, shrinkage and colour bleed are the three complaints the buyer in the worked example had, and all three show up in one wash cycle.
- Count and measure. Carton quantity, unit dimensions, packing method, and whether the packing matches what the sheet says the quote includes. Discrepancies here are cheap to find now and expensive to find in a container.
Shortlist from the verdict column. Three to six stalls, no more. A shortlist of fifteen means the brief was not specific enough.
Order samples from all of them. Two of the three stalls with the best prices will often drop out here, and the sample fee is the cheapest filter in the whole process.
Place a small first order with one. Not the largest order you will ever place — the smallest that is still above the stall's minimum. A first order is a test of the stall's reliability, not of your budget.
Then, and only then, talk about volume. A supplier who has delivered once at 300 pieces is in a different conversation about 3,000 than a stranger with a good price.
When the survey says come over after all
Occasionally a remote run ends with the opposite of its premise: the shortlist is strong, the category is fabric-led, and the honest recommendation is that the buyer's next move is a flight. That is a good outcome, not a failed one. In that case the useful next step is a guided in-market day rather than another remote pass — the same buyers who come for Huaqiangbei's electronics towers often find that a day with someone who knows which building holds which part is worth more than a week of looking, and the in-market Huaqiangbei arrangement is set out here for exactly that case.
One honest limit. Nothing in this section replaces inspection of a production batch. Seeing a sample is not seeing the goods that ship, and if the order is large enough to matter, the sample-based checks above are the floor, not the ceiling. Batch inspection before shipment is a separate service with a separate protocol, and a survey page that blurred the two would be selling you confidence it does not have.
Seven things that go wrong, and the first hour of the fix
Market runs fail in patterns, not in random ways. The same seven problems show up across buyers, markets and years, and each has a response that is cheap on the day and expensive later. This section is the run's own risk register.
The goods are beautiful, the price is fine, and the shelf behind the counter holds exactly one unit. The stall is displaying for a factory somewhere else, and the minimum it quotes is the factory's, not its own.
Fix in the first hour: ask where the goods physically are, and whether the stall can ship from its own stock today. A showroom that cannot point to stock belongs in the needs-a-call column with a note about the factory behind it.
The quote was good because the stall answered a question you did not quite ask: a thinner fabric, a lower grade, a smaller size run. This is the most common way a survey produces a misleading sheet.
Fix: the four questions are asked in the same order at every stall precisely so this is caught at the counter. When it is not, the clip is the evidence — re-watch it before blaming the stall, because the ambiguity is usually visible in the question as it was asked.
You assumed a category had twenty sources and it has twelve, because several of the "different" stalls are one counter counted twice.
Fix: record it as a finding rather than a failure. Fewer real sources changes your negotiating position, and knowing which you are in is worth a day of walking.
The lowest price of the day came from the stall that would not open a sealed carton, would not confirm the content, and quoted a number no one else would approach.
Fix: treat a price far below its neighbours as a question to answer, not a bargain to take. Above, the cheapest row and the least honest row were the same stall for exactly this reason.
300 pieces sounded affordable until the sheet revealed it meant 300 per colour per style, and the order is nine times larger than the buyer planned.
Fix: the unit of the minimum is now a column, asked explicitly at every stall, because a figure without its unit is not a figure.
The person who quoted was helpful, fluent and gone by next season, and nobody at the stall remembers the price or the buyer.
Fix: the sheet names who quoted and their role where it can be established. A quote from an owner survives a year; a quote from a seasonal assistant often does not.
A quote captured in a busy season is stale within weeks, because the material cost moved, the shelf turned over, or the stall's minimum changed with its stock position.
Fix: every row carries the date it was captured, and the sheet is presented as a snapshot with an explicit shelf life rather than a standing price list. When the shortlist is narrow enough, one return visit re-prices it in a morning.
The pattern behind all seven
Six of the seven are failures of resolution rather than of honesty: the stall was not lying, the sheet was not detailed enough. That is why the questions are fixed and asked in a fixed order, why every row carries a unit, a date and a verdict, and why the clips are unedited. The instrument is built to fail loudly and early, when failing is cheap.
Paying, shipping and staying legal
A market run stops at the sheet. What happens next — how money crosses the border, how goods leave the country, and what a small buyer is responsible for — is a separate chain with its own mechanics. This section is candid about it, including the parts that are not ours to solve.
Paying a stall you have never visited
Three routes, and the trade-offs are real rather than a matter of preference.
- Direct bank transfer to the stall or its company. Cheapest, slowest, and it puts the entire risk on your judgement of the counterparty. Workable for a stall that has a business licence and a track record, and worth avoiding for one that will only take a personal account.
- An agent or buying office in the middle. Costs a percentage, absorbs some of the payment risk and the language work, and reintroduces a layer — which is not automatically bad, because the layer you are trying to remove is the one that marks goods up 35%, not the one that charges a fee for a service you can name.
- A platform escrow or trade-service arrangement. Middle cost, middle protection, and its usefulness depends entirely on whether the stall will work inside it, which many wholesale counters will not.
Whichever route is chosen, the structure of the payment matters more than the method: a deposit on order and the balance against documents or on delivery, never the whole sum up front to a stall with no verifiable licence.
Getting the goods out
The consolidation step is the one genuinely useful thing the middleman did, and it is worth understanding rather than avoiding. Goods from several stalls have to be collected, packed, and moved to a forwarder, and each stall's packing unit is different. Two practical consequences.
- Eight stall receipts become one shipment with one set of export documents instead of eight.
- A single carton can be packed to survive a container instead of arriving in eight small parcels.
- One forwarder and one tracking number, rather than chasing eight vendors about eight parcels.
- Duty, tax and classification in your own country — that is your broker's job, and it varies by tariff line and origin rules.
- Product compliance in your market: safety marks, labelling law, restricted materials. The stall's certificate is not your compliance.
- The exchange-rate exposure between quoting and paying, which is real for a small order and invisible until it is not.
The compliance line a small buyer crosses without noticing
Two pieces of paperwork matter out of proportion to their size. The first is the business licence: a stall with one is a company you can transfer to and, if necessary, pursue; a counter that quotes prices but cannot produce a licence is an informal operation, and the price advantage it offers is often the compensation for exactly that. The second is whether the goods are what the certificate says, because a supplier's test report for a different production run protects nobody.
Where the factory question belongs. All of the above is market-level: stalls, counters and cartons. The moment a buyer moves from buying stock to commissioning production — tooling, a private label, a run to your own specification — the instrument changes completely. That is a factory audit with its own checklist and its own visit, set out in this site's guide to factory sourcing and floor inspection rather than repeated here, because the two jobs share a destination and almost nothing else.
- Before you transfer
- A business licence you have actually seen, a named contact with a role, and a payment structure that holds back the balance until goods are ready.
- Before you ship
- A packing list that matches the invoice, one forwarder rather than eight, and a decision about insurance you made on purpose.
- Before you sell
- Your own market's labelling and safety requirements checked against the goods, not against the stall's certificate.
- Before you scale
- One completed order at small size. A supplier with a delivery behind them is a different conversation from a supplier with a good price and nothing else.
Questions people ask about remote market runs
The six that come up in almost every first conversation, answered the same way each time.
Not on the first run. A market run is a survey: we walk, film, quote and verdict, and you receive the sheet. Purchasing is a separate instruction with separate costs, and it happens only after you have read the sheet and told us which stall to deal with. If the stall accepts payment from overseas directly, the cleaner route is usually for you to pay them yourself, because that keeps the supplier relationship in your name rather than ours.
Twenty is a reasonable first list and it is what the worked example uses, but the number matters less than the specificity. Twenty stalls named with a photo of the product and a realistic quantity is a strong brief. Fifty stalls named by category is a search, not a survey, and it will finish the day short. If you are unsure, send the draft and we will tell you what it resolves to once the buildings are laid out.
Yes, and it is priced differently because it takes a different kind of day. A search day resolves a category to a set of stalls rather than pricing a list you already have; the output is a list with locations, not a sheet with quotes. Many buyers run a search day first and a survey day second, which is usually the most efficient way to spend two days in an unfamiliar market.
That happens, and it is a real result rather than a failure. Stall prices are per piece before freight, duty and inland delivery, so a stall quote should be compared against your middleman's landed cost, not his selling price. When a stall is genuinely above your current cost after that calculation, the useful finding is that your middleman is earning his margin.
Not as part of a market run. Looking at a sample on a shelf is not inspecting a production batch, and treating the two as the same thing is how buyers get surprised by a container. Batch inspection before shipment is a separate service with its own protocol, and for a factory-level order the relevant work is a floor audit rather than a market visit.
The constraint is usually your list, not our calendar. Once the brief is agreed and the list is mapped, a single-market run is a day of walking; footage and the compiled sheet follow shortly after the day ends. Multi-market runs take longer because travel between cities is not survey time and we do not bill it as such. The one genuine delay is a category that needs the stalls found first, which means a search day before a survey day.
How this page was built, and what it is not
A page that quotes stall counts, day rates and margins owes you an account of where those numbers came from.
What the planning figures are based on
The stalls-per-day ranges in Table 3 are working estimates from market runs in Shisanhang, Huaqiangbei and Yiwu, not a published statistic and not a guarantee. They move with how pre-mapped the list is, the time of year, and whether the buyer is on a call during the walk. Where a figure is a range, this page shows the range rather than a midpoint, because the spread is the part that helps you plan.
Three things it deliberately does not do
- It does not promise a price, because a stall's quote is the stall's to give and to change, and a snapshot with a date on it is the only honest form a price can take at this distance.
- It does not give legal, customs or tax advice. It describes how the systems work and what the paperwork shows, which is not the same thing.
- It does not blur a market survey into a factory audit. The two jobs share a destination and almost nothing else.
What it cannot tell you
How a specific stall will behave on a repeat order, because that takes a second order. Whether a price will hold into your next season, because material costs move. And whether the goods clear your market's rules, because that depends on your tariff line, your labelling law and your broker — three things no survey of a Chinese market can settle from Shenzhen.
- Market runs and guided buying days conducted in Guangzhou Shisanhang, Shenzhen Huaqiangbei and Yiwu International Trade City
- Yiwu International Trade City's published district and street layout, and the floor-category conventions of the Shisanhang malls
- Stall-level quoting practice as observed at wholesale counters, including minimum-order units and packing conventions
- Business licence and unified social credit code practice under the National Enterprise Credit Information Publicity System
- Consolidation, forwarding and export documentation as handled by domestic forwarders for small export orders
Write the list. We will walk it.
You already know which market your goods come from, because your middleman's invoices say so. What you do not have is the twenty stalls behind the invoice, a price at your own quantity, or a video of the counter that quoted it. That is the whole gap, and it closes in a day of walking.
Send the market, the category and a photo of the item you are buying now. If you have stall numbers, send those too — they shorten the day. What comes back is a sheet you can read in fifteen minutes and a shortlist you can act on: can stock, needs a call, or walk away.
What we need from you to quote: the market or markets, the number of stalls you want walked, and your real order quantity. What you get back: the number of days it takes and the price, before anyone picks up a camera.
