On 12 February 2023, in the first quarter of the Super Bowl at State Farm Stadium in Glendale, Arizona, a thirty-second film appeared twice. The Kansas City Chiefs were on their way to beating the Philadelphia Eagles 38 to 35. Between the downs, a young woman walked through a street of bargains while a jingle told a hundred million viewers she was shopping like a billionaire. A dress in the spot could be had for less than ten dollars. Wireless earphones were shown at about eight and a half. The company behind the film was five months old. It was called Temu. It had been set up that September in Boston, and its owner was PDD Holdings, the Nasdaq-listed parent of Pinduoduo, a Chinese group-buying app that most of the people watching the game could not have named.
By the next morning the jingle was already a fact of domestic life. Phones that had never held a Chinese shopping app now held one. The parcel that followed, a grey poly mailer with a tracking number and no shop assistant attached, would arrive ten or twenty days later on a porch in Ohio or a landing in Manchester, and the person who opened it would perform a small private audit. The stitching might be straight. The charger might be warm to the touch after twenty minutes. The toy might shed a wheel. That audit, repeated hundreds of millions of times a year, is the real subject of the Chinese shops that have come west. Temu is only the loudest of them. AliExpress had been teaching Western buyers the same lesson since 2010. Shein had been dressing them since a rented villa on the edge of Nanjing. Banggood had been feeding hobby benches with motors and flight controllers. DHgate had been selling factory lots, small enough for a corner shop and large enough to skip the old trading house, since Diane Wang launched it in Beijing in 2004. Joybuy, JD.com’s European shop, arrived later and with a different promise: stock already on this side of the customs line, and a courier who could be shouted at in the same city.
Taken together they did something the container ship and the big-box store had not quite done. They put the workshop, or a very good imitation of it, on the kitchen table. A Western customer could buy a single socket set, a single dress, a single drone frame, at a price that used to require a wholesaler, a letter of credit, and a minimum order that would have filled a garage. The revolution was not that China could manufacture. Everyone who had bought a microwave since the 1990s knew that. The revolution was that the minimum order fell to one, the shop window became a feed, and the state’s old bargain with the foreign parcel, the rule that a cheap packet could come in almost as if it were a letter, held long enough for the habit to form. When that bargain was withdrawn, in America in 2025 and in the European Union in the summer of 2026, and the duties and safety rules followed, the shops did not vanish. They changed shape. The parcel at the door remained. What changed was who held the stock, who paid the duty, and how much trust a buyer was willing to extend to a listing with no shop floor behind it.
A name taken from Dunhuang
The oldest of these shops did not begin with a Super Bowl. In August 2004, Diane Wang, who had already built Joyo, an online bookstore she co-founded with Lei Jun and later sold to Amazon, started a company in Beijing and named it for Dunhuang. Dunhuang was the oasis where the caravan routes pinched together, a place of caves, manifests, and middlemen. Wang’s line, often repeated since, was that she wanted a Silk Road on the internet. The site, DHgate, opened in 2005. It was not a boutique. It was a small-lot wholesale floor: electronics, clothing, household goods, the long tail of what a Guangdong factory would sell if a foreigner could order forty units instead of four thousand. Payment sat in the middle, so the buyer was not wiring a stranger. The languages multiplied. By the end of the 2010s the company was describing itself as a bridge used by buyers in more than two hundred countries. It never became a household jingle. It did not need to. Its customer was often a person who would resell: a market stall, an eBay shop, a small Amazon seller who had discovered that the old importer’s margin could be captured by anyone with a browser and a tolerance for uneven quality.
That tolerance is the whole moral of DHgate, and it is why the site still matters when louder names take the advertisements. On a retail app, the platform often stands between you and the factory, sets the price, and calls itself the seller. On DHgate you are closer to the bench. You can ask for a custom run. You can also receive a carton of the wrong specification, because the person on the other side of the chat is a trading company of six people in Shantou, not a brand with a laboratory. Reliability, here, is not a property of the website. It is a property of the supplier you chose, the photographs you demanded, and the sample you paid for before the bulk. Western buyers who learned that discipline did well. Buyers who treated the site as a department store sometimes did not. The platform’s own reputation has swung with fashion. In the spring of 2025, American social media briefly treated DHgate as a place to find lookalike luxury, and the old wholesale site had a strange season as a consumer destination. The underlying machine had not changed. It was still Dunhuang: a pass, not a palace.
Alibaba had been in that pass since 1999, when Jack Ma and seventeen others started an English-language directory in a Hangzhou flat so that foreign buyers could find Chinese factories. That was Alibaba.com, a B2B bulletin board, not a shop. The consumer shops came later, and at home first: Taobao in 2003, built to answer eBay, and Tmall in 2010 for brands that wanted a cleaner floor. AliExpress, opened in April 2010, turned the directory inside out. Instead of a purchasing manager flying to the Canton Fair, a person in Madrid or Moscow could order one phone case and wait. For a decade the site’s great market was not America. It was Russia, Spain, Brazil, France, the places where a long delivery was acceptable if the price was a fraction of the local shop. Cainiao, the logistics network Alibaba formed in 2013 with several Chinese express companies, existed to make that wait less of a rumour. A tracking page became part of the product. So did the argument in the comments, in broken English and perfect photographs, about whether the received object matched the listing.
Before AliExpress, a smaller pioneer had already proved the demand. LightInTheBox, started in 2007, sold wedding dresses and home goods straight from Chinese suppliers to foreign brides and householders, grew fast enough to list in New York in 2013, and then watched AliExpress take the general trade. The lesson was not that the pioneer failed. It was that a single category, however romantic, could not hold a buyer who had just discovered she could order the dress, the lights, the table runner, and the hairpins from one search box. Wish, an American company rather than a Chinese one, spent the same years as a carnival of Chinese sellers on a Western app, two-dollar gadgets and a feed that felt like a joke you might win. It taught millions of people the gesture, then lost them when the joke wore out and the parcels disappointed often enough. The gesture remained. Chinese companies would later perform it with more capital and less irony.
A villa on the edge of Nanjing
Shein does not like to be grouped with the bazaars, and the distinction is real enough to honour. Xu Yangtian, known in English as Chris Xu or Sky Xu, was born in 1984 in Zibo, Shandong, to factory parents, and he does not give interviews. Former colleagues have said his own staff have failed to recognise him in a lift. In 2008 he and partners started a cross-border trading business out of Nanjing. Chinese reporting describes an early office in a rented villa on the city’s edge: one floor for work, one for sleeping, one for packing, electric fans against the summer, the founders answering customer mail themselves. The consumer brand that became Shein began as SheInside, selling wedding clothes and then women’s fashion to buyers abroad. Around 2015 the name shortened, and the centre of gravity moved from Nanjing to Guangzhou, closer to the workshops. In 2022 the legal home moved again, to Singapore, and the company’s private valuation was reported at about $100 billion. Xu, by then, was one of the richer men most of his customers could not picture.
What he built was not a market so much as a machine for guessing. Shein put small runs into production, watched what sold within days, and cut more of the winners. A traditional European chain still thought in seasons. Zara had already collapsed the season into a few weeks, using Chinese and other workshops but keeping the design hand in Spain. Shein collapsed the week into something closer to a social-media cycle. A dress seen on a phone on Thursday could be a listing by the weekend and a parcel the week after, not because a designer had a vision, but because a feed had a slope. The Guangzhou suppliers, often small and specialised, one shop making collars and another making the body, were paid to be fast rather than to be famous. Later the company added a marketplace on top of its own designs, so that third parties could list beside the house label, and the line between retailer and bazaar blurred in the way these lines always blur once the traffic is large enough to rent.
The numbers, when they finally appeared in a prospectus rather than in leaks, were the numbers of a major clothing company, not a startup. Forbes, reading the Hong Kong filing in July 2026, reported fiscal 2025 revenue of $41.8 billion, up 8 per cent, and net profit of $2.06 billion, down 38.7 per cent. A table circulating from the same filing put active customers at 273 million and orders at about 1.08 billion. Growth that had once been frantic had become ordinary, and ordinary at that scale is still a shock to anyone who remembers when a fast-fashion chain boasted of a few billion. The first quarter of 2026, the same reporting said, swung to a loss. Duties had arrived. Marketing had become expensive. The machine still ran. It no longer ran for free.
Quality, in this system, is a distribution, not a grade. Some Shein garments are perfectly adequate clothes: a cotton T-shirt, a summer skirt, a thing you wear six times and do not mourn. Some are thin, twisted after a wash, or cut for a photograph rather than a body. The company’s critics, and they are not a fringe, point to labour conditions in the supplier belt, to the environmental cost of clothes designed to be cheap enough to abandon, and to designs that sit close enough to other people’s work to start lawsuits. None of that cancelled the commercial fact. A generation in Europe and America learned to dress from a Chinese feed, and the high street, from Primark to Zara, had to live with a competitor whose design room was a data system and whose shop was a mailbox. When Shein later took space inside Paris, at BHV, with the affable American executive Donald Tang as the public face, it was the mailbox trying to become a room. The mailbox remained the business.
The hobbyist’s workbench
Not every Chinese shop that reached the West was trying to sell a person her whole life. Some were trying to sell him the part that the local electronics counter had stopped stocking. Banggood’s trail begins around 2006, when a website was registered in Guangzhou for a trade that barely existed: selling Chinese electronics straight to foreign consumers, in English, one board at a time. The company behind it, Guangzhou Banggood Technology, was formally established on 23 February 2009. It was never a Super Bowl advertiser. Its customers were, and are, people who know a part number. Drone frames, flight controllers, soldering stations, LEDs, RC cars, the small machinery of a weekend. The catalogue widened into tools and household goods, and the firm has described a reach of more than two hundred countries, with the United States, Britain, Germany, France, Spain, Australia, and Russia among the important ones. It also sold through other people’s platforms, eBay and Amazon and AliExpress among them, which is a clue to its real nature. Banggood was a retailer with a warehouse habit, not only a stall inside someone else’s app.
On a hobby forum the reputation is specific, and it is worth listening to because it is the opposite of a brand campaign. Buyers will tell you that a motor arrived as described, that a clone of a well-known flight board worked, that a parcel took three weeks and then a second parcel had to be sent because the first vanished in a sorting centre. They will also tell you that they would not buy a child’s toy there, or a mains charger, without knowing the exact cell and the exact certification. The site occupies a middle rung that the celebrity apps have made harder to see. It is more serious than a lucky dip. It is less accountable than a shop on a European high street with a manager and a returns desk and a trading-standards officer who can walk in. For a person building a quadcopter, that middle rung was a revolution of its own. The parts that used to require a specialist importer, or a friend who was flying to Shenzhen, became a listing. The same workshops that turned border villages into the world’s electronics bench acquired a retail window.
Geekbuying, another Shenzhen-centred gadget shop, sits on the same rung, a little smaller in the public mind and just as familiar to anyone who has waited for a single-board computer or a robot vacuum part. These sites did not revolutionise the dress shop. They revolutionised the workbench, and through the workbench they changed what a Western amateur could attempt. A teenager in Leeds could fly something that, fifteen years earlier, would have been a laboratory project. The reliability was the reliability of a known component from a known class of factory, not the reliability of an infinite catalogue. When the component was a no-name power bank, the old risk returned. When it was a motor with a published specification and a hundred forum threads, the risk was ordinary engineering risk. That distinction, between a category where the buyer can check and a category where she cannot, is the useful one. It will matter again when the same factories list the same motor on Temu, without the forum, at a lower price, and with a photograph that may or may not be of the motor in the envelope.
Boston, and the price of a game
Pinduoduo was founded in Shanghai in 2015 by Colin Huang, a former Google engineer. Its trick at home was social. You gathered friends, or strangers assembled by the app, and the price fell as the group grew. The goods were not glamorous. Fruit, tissues, phone cases, the bottom of the Chinese consumer market, served with games and coupons and a feeling that shopping was a small collective win. The company listed on Nasdaq in 2018. Huang later stepped back from the executive titles. By the end of 2025 the firm was run by co-chairmen and co-chief executives Lei Chen and Jiazhen Zhao, with its executive offices in Dublin and its incorporation in the Cayman Islands, the ordinary offshore architecture of a Chinese internet company that also wants American shareholders. In its 2025 annual report, PDD Holdings recorded revenue of RMB 431.8 billion, about $61.8 billion. It did not publish Temu as a separate line. The outside world has had to estimate.
Temu, founded in Boston in September 2022, took the group-buying instinct and stripped it for export. The American and European user did not need to assemble a group of neighbours. The platform would assemble the demand itself, tell the Chinese merchants what to make, often set the price, and subsidise the early parcels so fiercely that the first purchase felt like a mistake in the customer’s favour. Oceania followed in March 2023, Europe in April. By the end of 2025 the company’s own report said Temu was serving shoppers in the United States, Japan, Germany, Britain, France, Canada, and Italy, among other places. Trackers disagreed on the exact gross merchandise value, which is the usual fate of a number the company will not confirm. One widely cited set of estimates, drawn from YipitData, Bernstein, and ECDB, put the merchandise value at a few hundred million dollars in 2022, about $15 billion in 2023, about $54 billion in 2024, and about $92 billion in 2025. Treat the later figures as estimates. Treat the direction as fact. Nothing else in retail grew like that from a standing start, and the Super Bowl spots of 2023 and 2024, the second year reported at about $21 million for three commercials, were the advertising bill for a land grab.
The model is worth describing plainly, because it is the source of both the delight and the hazard. A classic marketplace such as early AliExpress is a landlord. Sellers list, ship, and wear the complaint. Temu, in its first incarnation, behaved more like a very aggressive retailer that did not own the factories. Merchants in China supplied goods into a system that controlled the price, the promotion, and often the logistics. The consumer saw one brand, Temu, and a price that local shops could not touch, because the goods had not yet paid a shop rent, a Western wage, or, for a time, a customs duty. The games and the coupons were not decoration. They were the Pinduoduo method, exported: keep the hand in the app, make the next click cheaper than the last, turn a purchase into a small win. People who enjoy that will find the criticism of it pompous. People who have opened a carton of goods that smelled of the wrong solvent will find the fun less charming. Both are describing the same machine.
The European Commission described it in the language of law. Temu was designated a very large online platform. On 31 October 2024 the Commission opened formal proceedings. On 28 May 2026 it fined the company €200 million under the Digital Services Act, the largest fine under that law up to that point, for failing to assess the systemic risk of illegal products on the service. The Commission’s account was specific. Temu’s 2024 risk assessment had leant on generalities about e-commerce rather than on evidence about Temu. A mystery-shopping exercise found that a very high percentage of selected chargers failed basic safety tests, and that a high percentage of tested baby toys posed medium or high risks, whether from chemicals over the legal limit or from parts that could come off and choke. Recommender systems and influencer promotions, the Commission said, could amplify the spread of those goods. Temu was given until 28 August 2026 to produce an action plan. The company has said the fine is disproportionate. The finding stands as the most formal verdict yet on what a price-first catalogue does when nobody with a laboratory is standing between the listing and the child.
The courier who can be found
JD.com was built on the opposite bet. Richard Liu started with a small electronics counter and, after SARS pushed him online, decided that China’s early couriers were not reliable enough to carry his reputation. JD bought warehouses and hired its own delivery staff. In China that choice made the company the place where a buyer who wanted a genuine phone, and wanted it tomorrow, was willing to pay a little more than on a pure marketplace. Authenticity and speed were the product. The marketplace sellers came later, under a stricter hand than Taobao’s free-for-all. When JD looked west, it did not look like Temu, and it did not look like Shein.
It also did not succeed at first. An earlier Joybuy, launched in the middle of the 2010s as a cross-border shop in the AliExpress manner, never became a habit. At the end of 2021 JD shut the consumer sites and turned the name toward business trade. A fresh attempt, Ochama, opened in the Netherlands in 2022 and spread as a niche, noticed mainly by shoppers already inclined to Asian groceries and goods, and not by the general public. The company has been unusually willing to say that the lesson was localisation and delivery, not another advertisement. In August 2025 Ochama began migrating to the Joybuy name. Germany saw a launch on 23 August 2025, with same-day delivery promised in parts of North Rhine-Westphalia. On 16 March 2026 JD made the wider move Reuters reported: Joybuy opened in Britain, Germany, France, the Netherlands, Belgium, and Luxembourg, aimed at Amazon, with delivery presented as the point. Free delivery over 29 euros or 29 pounds. A subscription, JoyPlus, introduced at 3.99 euros or pounds a month. A network described as some sixty warehouses and depots, and a last-mile service of JD’s own. The same company had agreed to buy Ceconomy, the German group that owns MediaMarkt and Saturn, for 2.2 billion euros, a bricks-and-mortar customer list that no app launch can counterfeit. Regulatory approval was still the gate. The intention was clear. JD wanted rooms, not only mailers.
It is too early, as of that launch spring, to mark Joybuy’s reliability the way one can mark Temu’s fine or Shein’s prospectus. What can be marked is the design. Goods held in Europe can be returned to Europe. A same-day promise can be broken in public, in a city, rather than explained away by a tracking page that stalls in a hub outside Zhengzhou. JD’s Chinese reputation was earned by firing couriers who were late and by keeping fakes off the self-run shelves with a harshness that sellers resented and buyers liked. Whether that harshness survives contact with European labour law, European rents, and a grocery aisle that still lacks, by the account of analysts who watched Ochama, the dull essentials that make a household switch its weekly shop, is the open question. The strategic contrast is already the most interesting thing in this generation of Chinese shops. Temu and Shein taught the West to accept a wait and a gamble in exchange for a price. Joybuy is an attempt to sell the thing JD sold at home: the absence of the gamble. If it works, it will not be because the goods are Chinese. It will be because the stock is local, and someone can be found when the blender does not blend.
Shops that are also shows
The list does not stop at the five names a Western newspaper reaches for. TikTok Shop, owned by ByteDance, took a Chinese habit, buying inside a video, and carried it into Britain, the United States, and parts of the European Union after it had already become ordinary in Southeast Asia. The seller may be a person filming in a Guangzhou flat or a brand that has hired one. The goods run from cosmetics to kitchen tools to clothes, and the reliability is the reliability of whoever is speaking, plus whatever checks the platform bothers to run. A live host can sell a thousand units of a pan before lunch. She can also sell a pan whose handle was never going to survive a dishwasher. The addiction of the format is not a side issue. It is the shop. You did not search for the pan. The pan arrived in the middle of a song, and the buy button was closer than the doubt.
Miravia, Alibaba’s Spanish effort, tried to be a cleaner local shop rather than a window onto Yiwu, and it has remained a Spanish story more than a continental one. AliExpress itself spent the early 2020s trying to escape its own reputation, with a Choice programme that promised faster delivery and a more managed assortment, Cainiao still underneath. The European Commission opened proceedings against AliExpress in March 2024 and, in June 2025, set out preliminary findings on risk assessment, illegal goods, and the traceability of traders. That is not the same as the Temu fine. It is a reminder that the oldest of the big consumer windows was still being asked, a decade and a half after launch, to prove it knew who was selling. Shein, for its part, was placed under formal DSA proceedings on 17 February 2026, on addictive design, the opacity of its recommendations, and illegal products, including items the Commission described as child sexual abuse material, such as child-like sex dolls. Opening a case is not a conviction. It is a measure of what a fashion feed can be made to carry when the catalogue is effectively infinite and the moderation is not.
Western incumbents did not sit still, which is part of the revolution rather than a footnote to it. Amazon built a bargain aisle of its own once Temu had proved the demand. Dollar chains found their prices were no longer the floor. Zara and H&M had already been accused of borrowing silhouettes; they now faced a rival that could borrow them before the original had reached the provinces. The important point is not that Chinese apps won a war against Amazon. Amazon remains the default for a person who wants a known brand tomorrow. The important point is that the floor of the market moved. A price that had been available only to an importer became available to his customer, and every shop that lived on the gap between those two prices had to justify itself with something else: speed, trust, a fitting room, a person at a counter, a guarantee that meant a local refund.
A dress, a charger, a drone
Quality is a word that collapses distinctions the parcels themselves refuse to collapse. It helps to keep three objects on the table, because each of them indicts a different failure, and each of them explains a different success.
The dress is Shein’s object, and Temu’s, and AliExpress’s when a seller is in apparel. Its failure mode is disappointment rather than danger: a photograph that was kinder than the cloth, a seam that parts, a colour that was never the colour on the screen. Its success mode is also real. A person on a small budget wears something new to a wedding, and the garment performs. At the prices these sites charge, the rational expectation is not a lifetime. It is a few outings. Critics who compare the dress to a coat from a tailor are answering a question the buyer did not ask. Critics who ignore the waste, the copied print, and the workshop at the other end are answering a question the public ought to ask. Both can be true in the same parcel. The revolution in clothing was temporal. Fashion stopped being a season and became a notification. Western high streets had spent twenty years speeding up. Shein changed the unit of time again, and the unit of shame went with it, because a dress that costs less than a lunch is easy to hide in a drawer and easier to throw toward a recycling bin that will not recycle it.
The charger is Temu’s object in the Commission’s file, and it is the object that should end any romantic talk about the bazaar as a pure good. A charger is not a dress. It sits on a mains socket. It can burn. The mystery shopping that informed the 2026 fine found basic safety failures at a rate the Commission called very high, in a sample, not in every charger on earth, but in a sample chosen to see what a European shopper would actually meet. Baby toys failed in a different way, chemicals and small parts. This is what a price system does when the penalty for a bad unit is a refund, perhaps, and the penalty for a good-enough photograph is a sale. AliExpress has hosted the same class of goods for years, with the difference that a wary buyer could sometimes read a seller’s history. Temu’s managed model took that history away and offered a single logo instead. A single logo is a promise. The fine was the bill for a promise the risk assessment had not earned. None of this means a person cannot buy a wooden spoon or a roll of tape from the app and be perfectly satisfied. It means the catalogue contains categories where satisfaction is the wrong test. Safety is the test, and a platform that will not pay for it has chosen a side.
The drone, or the motor, or the development board, is Banggood’s object and Geekbuying’s, and it is the object that saves the whole trade from being only a story about rubbish. A buyer who can read a datasheet is not in the same position as a buyer choosing a toy from a thumbnail. Chinese factories in Shenzhen and Dongguan really do make the world’s hobby electronics, just as they make a great share of the world’s serious electronics. The chokepoints in chips are a different essay, but the workbench is the civilian edge of the same map. A flight controller that matches its listing is not a miracle of globalisation. It is a routine product of a supply chain that learned, over forty years, to build to a specification at a price the West no longer attempts for low-volume parts. The sites that serve that buyer have varied reliability in shipping and in clones, and a clone can be a perfectly honest product if it is sold as a clone. They have a higher ceiling of trust than the lucky dip, because the community checks them in public. When those same parts slide onto a general app, without the part number and without the thread, they fall back into the charger’s category. The platform erases the knowledge that made the purchase rational.
There is a fourth object, and it does not belong to any of these marketplaces, which is why it belongs in the argument. Anker’s batteries, DJI’s cameras, Xiaomi’s phones: these are Chinese brands that won Western trust by being brands, with warranties, firmware, and a name that can be embarrassed. Xiaomi is not a bazaar stall, whatever the easy comparison. A person who buys a Xiaomi phone from a proper channel is not performing the same act as a person who buys an unbranded phone from a mailer. The shopping sites are often the channel through which the unbranded twin of a good product travels. They are rarely the reason the good product is good. Conflating the two, as Western commentary often does, produces a fog in which every Chinese good is either a miracle or a scam. The last ten years have been a sorting. Brands climbed. Infinite catalogues raced to the bottom and, at the bottom, found both grateful customers and regulators with laboratories.
The letter that was actually a shop
None of this volume was only a story about apps. It was a story about a customs rule. For years the United States allowed a de minimis entry, under Section 321 of the Tariff Act, for goods valued at $800 or less. A parcel under that line could enter without the duty and the paperwork that a container faced. The European Union had its own version: after July 2021, value-added tax was due even on small parcels, the old €22 VAT exemption having been abolished and an Import One-Stop Shop created so platforms could collect it. Customs duty was another matter. Consignments worth up to €150 still slipped through without duty. The shops designed themselves around the gap. Air cargo, not ocean freight, became rational when the packet was light and the duty was zero. A dress and a charger crossed the world as if they were postcards.
The United States closed the gap first. Duty-free treatment for products of China and Hong Kong ended on 2 May 2025. On 29 August 2025 the suspension extended to goods from everywhere else. Congress, in the same season’s tax law, repealed the commercial de minimis rule outright with effect from July 2027, and customs authorities spent 2026 teaching the postal system new entry types for the small packet that had never needed one. Temu’s direct-from-China American business, the thing the Super Bowl had advertised, was the obvious casualty. The company shifted toward sellers holding stock inside the United States. Shein talked of making more in Turkey, Mexico, and Brazil, which is to say it tried to put the factory on the favourable side of a line, or at least on a side that was not the one just singled out. The parcel did not stop. It became a domestic parcel, or a slower and dearer international one. Trade press citing Customs and Border Protection described a sharp fall in the old low-value flood after the change: on the order of a third fewer such parcels, and a drop in their recorded value from about $65 billion toward $48 billion. The precise series can be argued. The behaviour cannot. A business model that was a customs classification died as a classification, and was reborn as warehousing.
Europe followed on a delay that let another record year of packets through. Commission figures put low-value items entering the EU from outside the bloc in 2025 at nearly 5.9 billion, roughly twelve million a day, against about 1.4 billion in 2022. On 1 July 2026 the €150 duty exemption ended. In its place, until 1 July 2028, sits a flat €3 customs duty per item type on consignments up to that value. After 2028 the Union intends to charge ordinary tariff rates, once a single customs data hub replaces twenty-seven national systems. Three euros will not make a €4 dress into a €40 dress. It will make a €4 dress less obviously worth the air freight, and it will make a parcel of many cheap items, each triggering the charge, look suddenly foolish. National treasuries and high-street lobbies called this fairness. Shoppers who had organised their children’s birthday around a feed called it a price rise. Both descriptions fit. The exemption had been a subsidy to a particular kind of trade, paid for by everyone who still used a shop that paid duty, rent, and a local wage. Ending it does not restore 1995. It ends the pretence that a commercial shipment is a letter.
Britain, still outside the Union’s duty line and inside the same consumer culture, watched both experiments and opened its own consultation, with a change widely expected later in the decade. The direction is the same everywhere that the flood was heaviest. The political fact caught up with the physical fact. Once twelve million parcels a day are a trade policy, they will be taxed like trade.
What changed in the Western shop
It is easy to narrate this as decline: the high street hollowed, the charity shop full of last month’s Shein, the independent electrician unable to sell a socket set against a screenshot. The decline is real in particular streets and particular product lines. It is a bad description of the whole. Western retail was already being hollowed by its own national champions, by out-of-town sheds and then by Amazon, long before a Boston app bought a Super Bowl spot. What the Chinese shops changed was the remaining excuse. A local store could no longer claim that the alternative was a slow special order from a catalogue. The alternative was in a pocket, and it was cheaper, and it would arrive before the customer had forgotten why she wanted it. The stores that survived that comparison were the ones selling something the mailer cannot: food that spoils, a fit that has to be seen, a repair, a community, a brand whose next purchase depends on this one not being a disappointment. That is a harder business than being the only place in town with stock. It is also a clearer one.
The gain on the other side of the ledger is not trivial, and it should not be said in a mumble because the charger story is ugly. A household with little money could clothe a child, replace a broken kettle, try a hobby, without asking a bank for room. The same factories that supply the battery firm that overtook Tesla sit in an industrial system capable of serious work, and the trivial work comes off the same benches, often the same week. Access to that system, at the scale of one unit, is a change in the life of a poor consumer in a rich country comparable to the arrival of the supermarket in the previous century. Supermarkets were hated too, for what they did to butchers. They were not a hallucination. Nor is a €6 dress. The hallucination is the idea that the €6 dress is either a human right or a crime, rather than a product with a cost that was hidden in a customs code, a carbon account, and a workshop the buyer will never see.
Trust moved, and this is the part Western firms still underestimate. For twenty years the trusted name in online shopping was the platform: Amazon, because it would take the return. Chinese retail split that trust into two experiments. Shein and Temu asked customers to trust a price and a refund policy, and to stop expecting the object to be an heirloom. JD, through Joybuy, and the serious brands, asked customers to trust a standard, and spent money on warehouses to make the trust physical. AliExpress and DHgate and Banggood left trust where it had always been in a bazaar, on the individual seller or the individual part, and accepted that many buyers would never learn the skill. The West’s own platforms are now copying all three experiments at once: a bargain aisle, a premium aisle, and a marketplace where the third-party seller wears the complaint. The copy is the compliment. It is also the proof that the innovation was not a cultural curiosity. It was a set of answers to a question every retailer now has. How little certainty will a customer sell you, and at what price?
There is a geographical answer as well. The Canton Fair, the buying agent, the Hong Kong trading house, the letter of credit: these were the old pass at Dunhuang. They have not closed. A retailer ordering ten thousand units still flies, or still sends a staff member who flies. What closed was their monopoly on the small order. Macau spent centuries as a licensed door between empires. The poly mailer was a door that did not bother to ask for a licence, until the licence, in the form of a duty and a safety file, was imposed from the other side. The companies that thrive after the imposition will be the ones that can live with a licence: local stock, named sellers, test reports, a person who can be fined in a jurisdiction the buyer recognises. That is Joybuy’s theory, and it is the theory behind every Western chain that has spent the last three years hurriedly learning to sell a cheaper own-brand without poisoning its customers. The companies that cannot live with a licence will shrink back into the categories where the buyer is a willing gambler. There will always be willing gamblers. There will not always be twelve million parcels a day.
The porch, again
Go back to the morning after Glendale. The jingle has faded. The parcel has not yet come. When it comes, the person on the porch is performing, without knowing it, the entire reform of a trading system. She is the final inspector, because the state declined the job for any packet under eight hundred dollars, and because the platform found it cheaper to refund than to test. She is also the beneficiary, because a factory she will never visit has agreed to deal with her as if she were a wholesaler. The dress, the charger, the drone, and the toy are not one product. They are four different contracts, sold through windows that happen to look alike on a phone.
Temu asked her to feel rich. The Commission’s laboratories later asked whether the charger on that feeling was safe, and answered, for too many of the units they bought, that it was not. Shein asked her to feel current, and the prospectus showed a business the size of a European champion, growing more slowly now that the postcard rate of postage is over. AliExpress asked her, years earlier, to wait, and is still being asked to say who the seller was. Banggood asked a narrower person, the one with a soldering iron, to trust a part number, and often deserved it. DHgate asked a small merchant to skip the trading house, and he did, with his eyes open or shut. Joybuy, newest in this particular European costume, asks her not to wait and not to gamble, and has spent the warehouse money that makes the ask plausible. Whether she believes it will be settled by blenders and birthday presents, not by a launch in London.
The revolution in Western trade is already in the cupboard. It is the object you do not remember ordering, which nevertheless arrived, and which either worked or did not. Chinese shops did not invent desire, or bargains, or deception. They industrialised the sample. They made the sample the ordinary unit of world trade, and they did it so quickly that the law, the high street, and the customs shed were all left addressing a previous decade. The correction has begun. Duties are back on the small packet. Fines have started. Stock is moving onshore, which is to say the factory’s advantage in labour and cluster has to be joined, again, to someone else’s advantage in proximity. That is not the end of the trade. It is the trade growing up, in public, on a porch, where anyone can open the bag and see what the price was hiding.
Diane Wang named her company for an oasis on the old road, a place caravans reached because the desert on either side was worse. The desert now is not sand. It is the difference between a listing and a thing. The shops that came west have all had to cross it, some with a jingle, some with a datasheet, some with a warehouse and a subscription at 3.99. The customer on the porch is the oasis and the judge. She has already changed what the West is willing to pay. She is in the middle of changing what it is willing to forgive.
