Sales Lead, China. 中国区销售负责人。

Take a Chinese seller doing real numbers on Amazon. The factory is theirs, or it is two streets away. The product is genuinely good. On paper the unit economics work.

Then ask two questions.

Who are your customers? Nobody can answer. Amazon has the names, the emails, the reviews and the repeat purchase history. The seller has an order count.

What happens if you switch the ads off for thirty days? Everybody knows the answer to that one. The orders stop.

That is not a detail to fix later. That is the difference between a business you own and a business you rent by the day.

We wrote this letter because the fix is not mysterious. It is well understood, it is repeatable, and almost nobody in Shenzhen is hearing it in plain language from a team that has actually watched a hundred thousand stores try it.

Amazon is a wonderful place to start and a hard place to stay. Nothing in this letter says leave it. What it says is that Amazon should not be the only thing you have.

What you are actually renting:

• The listing is not yours. It can be hijacked, copied or suspended. One account action and revenue is zero the same afternoon, with no way to reach a single past buyer. The figures below are what that looked like in 2021.
• The customer is not yours. No email, no phone number, no pixel, no way to sell a second time except by paying for the same click again.
• The ranking is bought. With price, with ads, with reviews. Everybody in your category buys it the same way, so nobody gets ahead for long.
• The only lever left is price. A factory in the next district can always take another five percent off. The category grinds margin down until the only party reliably making money is the platform.
• You never learn why. A conversion rate on a listing you cannot restructure teaches you nothing you can act on. You cannot change the argument, only the price and the photo.

There is a word in China that does not survive translation: 内卷. Involution. Working harder and harder inside a system that returns less and less for the effort.

Anybody selling out of Shenzhen already knows the feeling, and it has very little to do with foreign competitors.

• A product works, and six weeks later there are forty listings of it. Same street, same supplier, same mould, often the same photographs. Differentiation lasts about as long as the tooling lead time.
• Ad costs rise while the ticket price falls. Everyone bids on the same keyword, and the platform keeps the difference between what the category used to earn and what it earns now.
• Everyone works harder for less. Longer hours, faster launches, thinner margin. The whole category can end a year busier and poorer than it started.
• Price is not a moat, it is a countdown. A team surviving on five percent is one currency move or one freight spike from not surviving.

The uncomfortable part, and the reason this is in a recruiting letter at all: going overseas does not escape it by itself. A brand that opens a US marketplace account and competes on price has exported 内卷 to a market with higher advertising costs. The arena changed. The rules did not.

You get out by changing what you compete on, not where. A brand people ask for by name. A page that argues instead of listing. An offer nobody else can assemble. A customer who comes back without being bought a second time. Those are the four things a platform cannot flatten, and every one of them lives on a site you own.

Everything before this chapter is an argument. This one is just maths. Two costs moved against Chinese sellers in the same year, from opposite directions, and neither is coming back.

Historically the 铺货 model, listing volume at low prices and shipping small parcels direct, accounted for roughly 70 to 80 percent of industry sales. That model was built on three things: no inventory, low price, and a tariff exemption. The third one is now gone, and for a small parcel the duty can exceed the cost of the goods.

So the mindset is not wrong, it is simply out of date, and the person you are talking to may not have noticed yet. A seller who has spent eight years getting good at listing volume and squeezing cost is being asked to become good at something with no obvious relationship to it: knowing one buyer well enough to write a page they cannot stop reading.

The Chinese industry press calls the phase we are in 品牌精耕, brand refinement, and says the single-channel traffic bonus is over. CNBC put it more bluntly in June 2025: Chinese exporters increasingly need to build brands to survive. Academic work on the same question finds Chinese firms rarely place brand at the centre of an international strategy at all.

None of that is a criticism of the person across the table. It is the reason the job exists. You are not selling software to somebody who lacks a tool. You are selling a second skill to somebody who is world class at the first one.

Before any more opinions, some counting. Everything in this chapter comes from StoreLeads, the store intelligence database, for Shopify stores registered in China as of Q3 2026.

Look at that list again and notice what is not in it. Not one of the three most installed apps changes the argument being made on the page. They patch trust, they answer questions, they say where the parcel is. None of them decides why somebody should buy.

Put the whole chapter together and you get a picture that is uncomfortably consistent: the same province, the same categories, the same three apps, the same theme, often the same photographs from the same supplier, and almost no social presence to build anything of your own on top of.

And the top of the table says the same thing from the other end. DJI is the highest ranked Chinese Shopify store in the world, at global rank 233. The next two are a mechanical keyboard shop and an anime merchandise printer. Both are good businesses. Neither is a household name.

So: thousands of world-class factories, and roughly one world-class independent storefront brand. That gap has nothing to do with product quality, and everything to do with the four chapters that follow.

This is not a talent problem and it is not an effort problem. We work with sellers who are sharper and faster than most of the Western brands we see. The strengths are real, and they are very hard to copy.

Put those two columns next to each other and the outcome is predictable. Excellent product, weak page, so the only argument left is price, so the margin goes, so there is no budget for content, so the page stays weak. The strength never gets to show up.

The good news in that diagnosis is that everything in the second column is buyable. It is a skill set, not a birthright. It is also the exact skill set that a company like ours has spent ten years assembling.

Abstract diagnosis is easy to nod along to. So here are two real Chinese brands, both successful, both on Shopify, looked at the way you will look at a prospect's site before a meeting.

This loop is the single biggest reason investment in the independent site stays small.

Because Amazon covers the salaries, the site is never urgent. It stays a side project for years, and every year the conclusion "the independent site does not work for us" gets a little more confident, on evidence that was never valid.

That is also why we sell it the way we do. Not a website project. One funnel, for one product, live in days, tested from the first week, with a number you can look at before you decide to spend more.

Everybody uses the word brand and almost nobody defines it. Most definitions cannot be checked, which is why they survive: a brand is a promise, a brand is what people say when you leave the room. You cannot run those.

So here are three that can be run, ordered by how quickly you get an answer.

The third one is the most brutal and the most conclusive, which is why we lead with it in conversation. It is also the slowest and the most expensive, and it lies in one specific case: in a low frequency category like mattresses or luggage, thirty days of silence proves nothing, because nobody was going to buy twice anyway. There, test two is the honest one.

Both of those can make money. Only one of them is worth something on the day you want to sell the company, raise money, or take a month off without watching the dashboard.

This matters commercially, not philosophically. A business where every order is bought fresh has a valuation multiple close to its inventory. A business with repeat customers, an email list, a name people search for and a page that converts without a discount is valued on its earnings. Same product, same factory, completely different number.

And it is not all or nothing. Most healthy brands sit somewhere in between. The question is only which direction you are moving in this year.

Most sellers file the independent site under "we should probably have one". That framing is the problem. It puts a website in the same drawer as a business licence and a trade show banner.

Try the sentence the other way round. Your brand does not live in your logo, your trademark registration or your factory. It lives in the only place a stranger meets it and decides what it is worth. For a cross border seller, that place is a page.

And the buyers are already there. This is not a bet on a channel that might arrive. In 2025 Shopify processed $378.4B in GMV, up 29% year on year, holding over 14% of US ecommerce, second only to Amazon. There are 4.8 million active storefronts, and the markets a Chinese brand wants are the ones most used to buying this way.

Australia holds around 138,000 stores and Japan around 32,000. These are places where a shopper buying from a brand they have never heard of, on a site that is not a marketplace, is completely ordinary behaviour. The buyer is not the obstacle. The page is.

So what is the site actually doing? Not looking professional. Not so the factory has a page to show at a trade fair. A website does four jobs that a marketplace will never do for you, and if it is not doing these four it is just an expensive brochure.

• It owns the customer. Email, phone, WeChat, the pixel, the repeat order. Everything about who bought and why. This is the whole asset.
• It owns the price. There is no buy box. You decide what the product is worth, and the page has to argue for it. That is a harder job and a much better business.
• It owns the story. On a marketplace you are a row in a list, compared on price by a shopper who is comparing forty rows. On your own page you get a few uninterrupted minutes to explain why yours costs twice as much.
• It owns the data. Which angle they clicked, which section they read, where they left, which offer they took. That is the only real feedback loop in this business, and on a marketplace you do not have it.

The website is also where every other channel finally pays off. TikTok, influencers, retail, a viral video: they all send people somewhere, and if that somewhere is a marketplace listing, the platform keeps the customer you paid to create.

Everybody in this business starts with a watermelon. The seller cuts it up and sells it by the slice, and from that morning on the only decision left is the price of a slice. The stall next door can always cut thinner or charge less, and eventually the whole street is selling the same slice for less than it costs to buy the fruit. That is 内卷, drawn.

The entrepreneur looks at the same watermelon and asks a different question: what do people actually want, and what would they pay for it. Cold pressed and bottled, with a label, a name and a reason. Sold at a price that has nothing to do with what the fruit cost. And crucially, it does not have to stay a watermelon at all.

The reason this picture belongs in a letter about websites is that the page is where the transformation actually happens. A marketplace listing can only ever sell the slice, because it only lets you fill in weight, photo and price. Your own page is the bottle, the label, the name and the reason, all at once.

This is the chapter most sellers skip, and it is the one that explains everything after it. If you have only ever sold in China, buying from a brand's own site is not a habit you have, or one you have watched anybody else have.

The usual explanation is that Western consumers got the internet earlier, so they are simply used to it. That is a small part of it. The real reason is market structure, and it is much more useful to understand because it is not going to change.

Think about what Taobao does for a seller before the seller has said a word. It vouches for you. It holds the money until the buyer is satisfied. It shows a rating built over years, a return policy the buyer already trusts, a checkout they have used a thousand times, and a comparison against every competitor on the same screen. By the time a Chinese buyer reaches your listing, the platform has already done perhaps eighty percent of the persuading. Your job is to be the best option inside a frame somebody else built.

Now put that same seller on their own site in Ohio. No escrow. No rating built up over years. No familiar checkout. No frame. A stranger arrives from an advertisement. Everything the platform used to provide, the page now has to provide: who you are, why this works, what happens if it does not, why now, and why you rather than the tab they still have open.

And it is also why the Western consumer is willing to do this at all. They were never trained to expect one place that has everything, because they never had one. Ecommerce in the United States is roughly 17% of retail and Amazon is only about 6% of retail overall. The number two, three and four are not marketplaces, they are shops with their own websites. Buying from a brand you have not heard of is not a leap of faith there. It is Tuesday.

In Europe it goes further. Third-party marketplaces are only 39% of ecommerce, and more than 65% of online shoppers say that when they can buy directly from the brand, they prefer to. Read that twice, because it is the exact inverse of the instinct a Chinese seller has been trained into. The majority of your European customers would rather not buy from you on Amazon.

These two words get used loosely by everyone selling services, so let us be exact. If you already know this, skim it. If you do not, this is the single most valuable part of the letter, and almost nobody explains it properly.

A landing page is not your homepage

A landing page is the one page an advertisement points at. Not the homepage. Not the catalogue. Not the general product page that everybody sees.

It is built for one reader, who clicked for one specific reason, and it makes one offer. If your ad said "stop your dog pulling on the lead", the page opens with a dog pulling on a lead. It does not open with your logo and a hero image of your whole product range.

The reason this matters is simple maths. You paid for that click. Every second the visitor spends looking for the thing the ad promised is a second of paid attention you are burning. Most sellers send every ad to the same two pages and then blame the traffic when the conversion rate is one percent.

A funnel is the path around that page

A funnel is the sequence a stranger walks through, from the first second of attention to the second order. Four steps. Most sellers only build two of them.

The standard mistake is pointing paid traffic straight at a product page. A product page answers "which one and how much". It does not answer "why should I care", and a cold visitor has not asked the first question yet.

There is a second mistake underneath the first one. The page they land on is not only the wrong kind of page. It is the same page as everybody else's.

A few hundred thousand Shopify stores are built from the same handful of themes. Same section order, same hero, same three trust badges under the button, same review widget in the same place. A buyer who has clicked ten ads this month recognises the shape before reading a word, and the shape tells them: this is one of those. That is a large part of why conversion sits under one percent. The page is not ugly. It is generic, and generic is a conversion problem, not a design problem.

Which is why the traffic increasingly does not go to the store at all. It goes to a page whose only job is to warm the click up, and the store is where the visitor arrives once they already want the thing.

The four shapes of a pre-sell page

Four formats do this job, and all four are just different ways to earn the right to make an offer.

The last chapter said what a landing page and a funnel are. This one opens one up. Everything below was counted rather than argued, and a rep who can describe it in a first meeting is describing something the seller has never had explained to them by anybody.

Most advice about conversion starts at the page. The page is the second lever, not the first. When a test does nothing, it is almost always because the thing being tested was never the constraint.

What a converting landing page actually contains

Our team pulled apart more than two hundred high performing landing pages across beauty, health, wellness and consumer electronics. The same blocks keep showing up, in roughly the same order. This is the checklist that came out of it.

Eight pages from that research set, in eight different categories. Scroll past the branding and the same blocks are in the same order almost every time.

Eight pages from that research set are shown above, in eight different categories.

Where merchants actually point their tests

Every A/B test run through GemPages over ten months, about three hundred a month, 2,506 in total. Nearly half of them are aimed at landing and pre-sell pages, not at the product page and not at the theme.

The part everyone skips

Upselling is not one thing, it is four positions in the journey, and most stores only ever install the second one.

• Before purchase. Bundle and save on the product page, frequently bought together, build your own bundle.
• In the cart. Cart drawer offer, free shipping threshold nudge, progressive gift ladder.
• At checkout. Order bump, checkout extensions, warranty or protection add on.
• After payment. One click volume discount with no card re-entry, a mystery box at a price that is obviously a deal, a downsell when they decline the first offer, then the thank you page offer and the email sequence.

The last position is the cheapest revenue in the business. The hard part of a sale is the decision to trust you, and that is already done. A second offer thirty seconds later is a different question with a much lower bar. The reason it gets skipped is that it sits after the conversion event, so it does not show up in the metric most teams are judged on. It shows up in average order value, and average order value is what pays for the ads.

One rule: never break the original order. If the upsell asks for the card again, the offer is dead before it is read.

Where this is going

Today the angle lives entirely in the ad. Eight different ads, eight different reasons someone stopped scrolling, and every one of them lands on the same argument. The version worth building routes the click: the fear-led ad opens the fear-led page, the comparison ad opens the comparison, the deal-led ad opens the deal.

Building four full funnels used to cost four times as much, which is why nobody did it. With AI writing and building the versions, each extra angle is nearly free. The constraint moved from "can we build it" to "do we know which four to build". That is why a company like ours can exist now and could not have five years ago.

A seller will tell you the traffic was bad. It usually was not. Conversion fails in seven recognisable places, and the useful skill in this job is naming which one, in the first meeting, out loud.

The Latin America row is the sharpest example, and it is the one we bring up most often. A brand builds one beautiful page, ships it to eight countries, and cannot understand why Brazil and Mexico look broken next to the United States. The answer usually has nothing to do with the product. The page is too heavy for the phone it is being opened on, and it asks for a card from a buyer who expected to pay the courier in cash.

A separate, deliberately light, mobile-only landing page with a short cash-on-delivery form is not a downgrade for that market. It is the correct page. Building it used to be a second project nobody had budget for. That is exactly the kind of thing that is now cheap enough to just do.

Everything in the last few chapters comes down to one line. It is worth writing down, because it decides where a seller should put the next bit of money.

Most sellers work the equation from the left. Traffic is the visible lever, the one an agency can be hired for tomorrow, and the one that shows movement the same afternoon. It is also the only one you pay for by the unit. Double it and the bill doubles.

The order that actually works

This is the part of the letter you will end up saying out loud most often, so it is worth having it in a form you can deliver in ninety seconds without notes.

Conversion optimisation has been through four eras in ten years, and we have built tooling for all four. Knowing where the line is heading is most of what makes a sales conversation with a serious operator interesting rather than generic.

• Scarcity, countdowns and theatre do nothing. An agent does not feel urgency. It reads price, rating and specification, and it does not care that a timer is running.
• Structured data becomes the sales copy. If the spec, the price and the review score are not machine readable, the agent cannot recommend you even when you are the better product.
• The comparison happens off your site. You are being evaluated next to competitors in a context you do not control and cannot design.
• Human persuasion still matters. A person still approves the purchase. You now have to win two very different readers with the same page.

A candidate who has sold anything before will get to this page and think: everything above is true, and none of it proves anybody will pay. Some of these brands have never run a serious ad campaign. Some do not know what a landing page is. So we should answer it here rather than let you find out in month three.

Not every Chinese brand is a customer, and we are not going to pretend otherwise. Three groups, and only one of them is yours this year.

The company is GemCommerce, built in Hanoi, Vietnam. For ten years we have made the software Shopify merchants use to build the pages that sell. That seat is unusual: we do not have one brand's opinion about what converts, we have a decade of pages, tests and results across every category and market a merchant can sell into.

Three products, one job. A page builder gets you a page. It does not tell you whether the page is right, and it does not run the next version for you. The gap between those two is where almost all the lost revenue sits, so that gap is the product.

We also run ShopX, the community arm: small senior rooms for cross border operators, in thirteen cities across Asia Pacific, Shenzhen and Shanghai among them. That is not marketing decoration. It is where the person in this job will already have warm rooms to walk into.

Ten years is easy to write and hard to picture, so here it is in four photographs.

Some of the brands on it, from the top of the ranking: Eberlestock, Hume Health, Lucy & Yak, Angelus Direct, Heritage Foods. Hunting packs, health hardware, British fashion, sneaker paint, food. No category in common, one thing in common: they all live or die on the page.

It is a fair question and a Chinese seller should ask it in the first meeting. The honest answer, which the person in this job will give a hundred times:

• We understand how you actually decide. Factory owned or one street away, family money, fast calls, real price pressure, no patience for a six week discovery phase and a brand workshop. A San Francisco agency genuinely does not understand that, and it shows in the first meeting. Vietnam works the same way we do.
• Vietnam is a hub for Shopify software. Apps built in this country, GemPages, PageFly, Avada and Firegroup among them, power more than a million Shopify stores between them. A lot of the engineering that Western merchants quietly run their stores on is already written in this part of the world.
• And it is not only builders. Vietnam runs one of the strongest print on demand communities anywhere, with billions of dollars flowing through it every year, alongside one of the region's most active dropshipping scenes. This is a country that has been selling to American and European consumers, at volume, for a decade.
• We sit between East and West by default. Our team is in Asia. Our customers are in the United States, Europe and Australia. Every working day is a translation between how a product gets made here and how it gets sold there. That is precisely the gap a Chinese brand is trying to cross.
• The cost is sane. A Vietnamese team costs a fraction of a US or European agency for the same output, and we are one hour from your time zone, not twelve. Same day answers, not next day.
• We are already in the next market. Southeast Asia is where a lot of Chinese brands go second. Vietnam, Thailand, Malaysia, Indonesia, the Philippines. We are not studying that region, we live in it.

Let me say the awkward part first, because a Chinese seller reading this has already thought it. Why would a Vietnamese company want to help Chinese brands go overseas? What is in it for them, and should I be careful?

The honest answer, and it is not a warm one: it is built into how we make money.

We are not your competitor. You make products. We make the software that sells them. There is no version of this where we take your customer or your category, because we do not sell anything a consumer buys. In ten years we have never launched a product that competes with a merchant, and we never will, because the day we do, a hundred thousand merchants stop trusting us and the company ends.

Our business only works if yours keeps working. The commission structure in this letter pays our own salesperson every month an account stays alive. The subscription works the same way. We do not get paid for signing you, we get paid for you still being here in year three. A company built on retention cannot afford to sell you something that does not work. That is a better guarantee than any promise I could write.

And there is a trade here that runs both ways. Vietnam is roughly where China was fifteen years ago in manufacturing, and we know it. China is roughly where Vietnam is in learning to sell to Western consumers, and you know it. Neither of us is doing the other a favour. We each have something the other needs, and we are close enough, culturally and geographically, to actually work together on it.

The mission, said plainly: China makes an enormous share of what the world buys, and captures a small share of what the world pays. That gap is not made of tariffs or logistics. It is made of pages, brands and the relationship with the person who bought. Those are exactly the three things we have spent a decade building tooling for, and we would rather spend the next decade closing that gap with Chinese brands than watch somebody else charge four times as much to do it worse.

Chris Koh, founder, Hanoi

We can build funnels. We have ten years of data about what converts. What we do not have is a person in the room in Shenzhen, in Chinese, with the sellers who need this.

To be clear about who we are looking for: roughly three years of sales, roughly 26 to 30, with more ambition than CV. We are not hiring somebody who has already run a department, because a person who has done it already will not get much out of building this from nothing. Somebody three years in will.

That is the role. One person to start. You are based in Shenzhen, in the densest concentration of cross border sellers on earth. You talk to brands all day. Our team in Hanoi builds what you sell: strategists, designers, engineers, CRO. You sell it, they deliver it, you keep the relationship.

You are not selling software into a vacuum. There are ShopX rooms in Shenzhen and Shanghai, a decade of case studies, and a product you can show working on a real store inside one meeting.

The size of the thing you would be selling into

Worth knowing before you decide whether this is a job or a decade. Cross border B2C ecommerce is on track for $7.9 trillion by 2030, close to doubling this decade. Southeast Asia alone did $159B in ecommerce GMV in 2024, inside a regional digital economy heading for $1 trillion by 2030.

How we work, so you know what you are joining

• Ship, then learn. We put real work in front of real merchants fast, measure it, and iterate. Momentum over meetings.
• Own the outcome. No hand-offs to hide behind. You own a problem from the idea to the number it moves.
• Taste matters. We hold a high bar for craft, including in how we sell. Nobody here sends the mass blast.
• Small team, high trust. The AI does the heavy lifting so a handful of people ship like a much larger company. That only works when everyone owns their end of it.

If it works, this becomes a team, and you are the person who builds it. That is not a line to make the job sound bigger. It is the plan, and it is why we are hiring for ambition and not only for a track record.

The opportunity in this letter is wide enough that a new person will go and talk to everybody. Do not. Most of the loss in a first sales year is time spent on brands that were never going to buy.

You do not present. An agency arrives with a deck about itself; you arrive having already read their page. That difference is what you are selling, and you can do it from your first week.

A commission plan tells you what a company actually wants. Ours is below, in the open, before you apply rather than after.

What that adds up to

Written from your side of the table. If a question you actually have is not here, that is our omission, so ask it in the email and we will answer it straight.

Send an email to chris@gemcommerce.com with "Sales Lead, China" in the subject line. In the body, tell us three things, in whichever language you write better.

1. One thing you sold, and specifically how you got the first meeting.
2. One Chinese brand you think is selling badly overseas, and what you would say to them.
3. Whether you want to be a salesperson, or build a sales team.

A CV is welcome. It is not the part we read first.

If you would rather not use email at all, put your WeChat in the form and Chris will add you there. That is how most of this conversation will happen anyway.

Or send an email instead All open roles