The Factory Videos Did Not Sell Much. That Is Exactly Why They Hurt.
The buy-direct-from-the-factory wave on TikTok converted badly, and the app-store data proves it. What it left behind was a price anchor in your customer's head, and a threat model your supplier NDAs were never designed to cover.
Annie Chan··13 min read
In April 2025, American social feeds filled with Chinese factory owners standing in front of production lines, holding up a handbag or a pair of leggings, and explaining that this is where your brand's product is made, that the branded version costs eight times more, and that you can buy it directly from them instead.
One video, since removed, drew more than 6.1 million views on a claim that over 80 percent of luxury handbags are actually made in China, according to CNN's April 16, 2025 reporting. Another widely shared clip put the manufacturing cost of a pair of Lululemon leggings at 4 dollars, a figure asserted in the video rather than verified by anyone, and reported as such by MIT Technology Review on April 28, 2025.
The immediate brand response was a mix of legal review, quiet takedown requests, and public silence. Fifteen months later that response looks like a category error, because it treated the wave as a distribution threat when it was never one.
The factory videos barely sold anything. The commercial evidence is unambiguous. What they did instead was install a number in the consumer's head, and a number is much harder to remove than a competitor.
The commercial ledger, with dates
Take the trend on its own terms first. If Chinese factories were about to disintermediate Western brands, the effect would show up in the marketplace apps consumers were being directed toward. It did, briefly, and then it did not.
DHgate was the main beneficiary. On April 13, 2025 the app recorded about 117,500 iOS downloads in a single day, roughly 732 percent above its 30-day average, with 65,100 of those from US users, up about 940 percent. It went from around number 352 in the US shopping charts into the top three, reaching number two on the US App Store behind ChatGPT. By April 18, 2025 it was the top shopping app in 98 countries. Within DHgate itself, sales of home appliances rose 962 percent and security technology 601 percent between April 12 and 19, 2025, per figures cited by MIT Technology Review.
Then the policy landed. The de minimis exemption, which had allowed shipments valued at or under 800 dollars from China and Hong Kong to enter the United States duty free, ended at 12:01am EDT on May 2, 2025 under Executive Order 14256. Covered goods faced a 120 percent ad valorem duty, or for postal items a per-item charge of 100 dollars, scheduled to rise to 200 dollars on June 1. A subsequent order on May 12 lowered the de minimis mail rate to 54 percent effective May 14, 2025. Rates in this area have continued to move, so treat specific percentages as point-in-time.
The effect on the consumer proposition was immediate. Caixin, citing Sensor Tower, reported that by May 8, 2025 DHgate had fallen to 29th among free shopping apps on the US App Store, from the top of the chart three weeks earlier.
"Three weeks at the top of the chart, then back to the fourth page. As a channel shift, the factory-direct wave was noise. Any brand that measured it by lost sales concluded, correctly, that nothing had happened."
The product claims did not hold up either. Reporting from CNN, Forbes and NPR in mid-April 2025 converged on the same assessment: manufacturers genuinely producing for major luxury houses are typically bound by confidentiality terms that prohibit disclosure, most top luxury handbags are still officially manufactured in Europe, and much of what appeared in the videos was likely to be high-quality dupes or counterfeits rather than authentic overruns. Some of the accounts were straightforwardly selling fakes under a supply-chain-transparency framing.
So why does it matter
Because the videos were not competing for the sale. They were competing for the frame, and they won that.
Before April 2025, a consumer looking at a 90 dollar item asked a question about the item: is this good, do I like it, is it worth 90 dollars to me. After April 2025, a meaningful share of consumers ask a question about the ratio: what does this actually cost to make, and what multiple am I paying.
That second question is unanswerable in your favour. It does not matter whether the 4 dollar leggings figure was accurate, and it almost certainly was not, because it ignores design, sampling, materials specification, quality control, compliance, freight, duty, returns, retail rent and the roughly half of units that sell at a discount. Accuracy is irrelevant to the mechanism. Once a specific number exists in the consumer's mind, every price you set is evaluated as a multiple of that number rather than against your competitors' prices.
A price anchor does not need to be true to be expensive. It needs to be specific, memorable, and unopposed. The factory videos supplied all three, and most brands responded with silence, which is the condition under which an anchor sets hardest.
The threat model your NDAs were built for is the wrong one
Here is the part that legal teams keep getting backwards, and it is the most useful thing in this piece.
Your supplier agreements prohibit your contract manufacturers from disclosing that they make your product. Those agreements largely work. The factories that actually produce for major brands have long-term relationships, revenue at stake, and lawyers who have read the contract. They are, on the whole, not the ones posting.
The accounts driving the wave were mostly factories that do not have your contract. They produce in the same industrial clusters, often for smaller clients, sometimes for nobody in particular, and they are contractually free to say anything they like. They cannot legally claim to make your product, so they do something more effective: they show a product that looks like yours, describe the category, and let the viewer make the connection.
"Your confidentiality regime binds exactly the set of factories that would never have posted, and none of the set that did. It is a perfectly enforced defence against the wrong opponent."
This is why enforcement has been such a poor investment. Takedown requests against an account that never named you are legally weak and reputationally awful, because the visible action is a large brand silencing a small manufacturer over a video about how things are made. That is a story with a predictable second act.
It also explains why the wave will recur. Nothing about the underlying capability has been removed. Chinese factories began experimenting with video content around 2020 and treated the channel as commercially serious from roughly 2022, per MIT Technology Review's account. The annual cost of platform visibility they described, in the range of 30,000 to 40,000 renminbi, roughly 4,000 to 5,400 dollars, is not a barrier to anyone. April 2025 supplied a trigger, not a capability.
Who the videos were actually for
A detail that reframes the whole episode: for most of these factories, the consumer sale was never the point.
Consider the economics from their side. Chinese apparel exports to the US grew less than 1 percent in 2024 while average unit price fell 7.6 percent, per figures cited in the MIT Technology Review piece, in a manufacturing sector long characterised by thin margins. A factory in that position has two things it can do with an English-language audience. It can attempt direct consumer sales, which means fulfilment, returns, customer service and customs exposure in a foreign market, at retail volumes it is not set up to handle. Or it can use the audience to reach sourcing managers, private-label buyers, small brand founders and resellers, which is its actual business, at order sizes that make sense.
One creator profiled in that reporting had built 1.2 million Instagram followers. That is a lead generation asset for business-to-business sales, and a modest content revenue stream, achieved at a cost of a few thousand dollars a year.
Which means the damage to your brand premium was a byproduct, not an objective. That has two consequences that most brand responses miss.
You cannot out-compete them for the consumer sale, because they are not trying to win it. Improving your direct-to-consumer conversion, matching their price, or proving that ordering from them is inconvenient does not touch their incentive.
The behaviour is durable and rational at almost any tariff level, because tariffs affect consumer shipments and this is business development. Policy changes that killed the consumer arbitrage in May 2025 did not weaken the reason to keep posting.
A test for how exposed your category actually is
Exposure to this pressure is not evenly distributed, and the variable that predicts it is not price, prestige or country of manufacture. It is whether your product has a liquid resale market.
A resale market is a continuous, public, third-party verification of value. If a used version of your product reliably sells for a meaningful fraction of retail, then the market has priced your brand independently of anything you or a factory says about manufacturing cost. A consumer weighing a dupe against the real thing can see that the dupe resells for nothing and the original resells for a number, and that comparison settles the argument without you participating in it.
Categories with deep resale liquidity, luxury leather goods, watches, certain sneakers, some furniture and instruments, absorbed the factory-video wave with very little damage. Categories with no resale market, basic apparel, cosmetics, small electronics accessories, home textiles, are fully exposed, because there is no third party continuously confirming that the branded version is worth more.
The exposure question is not how much do we charge. It is whether anyone other than us is willing to pay for our product secondhand. If nobody is, then the only evidence of your premium is your own price tag, and a factory video is a direct attack on the only evidence you have.
This turns an abstract brand-equity discussion into a concrete, cheap piece of research. Pull thirty days of resale listings for your top SKUs on the relevant secondary platforms. Calculate realised resale price as a percentage of retail. That ratio is a better read on the durability of your premium than any brand tracker, and it costs an analyst a week.
Four responses that do not work
Denial and correction
Publishing a rebuttal explaining that those are counterfeits and your products are made in Italy is factually correct and strategically self-defeating. It moves the conversation onto the terrain of manufacturing cost and markup, where you cannot win, and it signals that the claim was worth answering. The correction gets a fraction of the reach of the original and confirms the topic.
Increased supply chain secrecy
Tightening disclosure looks like something to hide, and it does not work mechanically, because the factories posting are not the ones you have agreements with. It also runs directly against the disclosure direction that regulation and institutional buyers have been pushing for a decade.
Discounting
A promotional response validates the multiple. If the accusation is that your price is inflated relative to cost and your answer is 30 percent off, you have conceded the point in public and reset your reference price in a market where every historical price is permanently searchable.
Craftsmanship storytelling
Atelier footage, artisan interviews and heritage films are the standard luxury reflex, and they are answering an emotional claim with an emotional claim while the consumer is asking an arithmetic question. They work when the consumer already believes in the premium and wants reinforcement. They do not work on the consumer who has just watched a factory video and is doing division.
What actually works
Own the cost story with specific numbers before someone else supplies them. Not a vague statement about quality but an actual account of where the price goes: materials, labour, compliance and testing, freight and duty, returns, retail cost, discount rate, margin. Brands that have published something in this direction have generally found it defensible, because the honest breakdown is far less damning than the imagined one. The imagined one is what you are currently competing against.
Move the purchase toward what a factory cannot ship. Warranty economics, repair, service, returns handling, authentication, resale support, community. These are the components of value that survive the arithmetic, because the dupe genuinely does not have them and the consumer can verify that in a week.
Audit your supplier base for social presence. Know which of your contract manufacturers and their subcontractors run consumer-facing accounts, in which languages, and what they post. This is a two-week exercise that almost no brand has run, and it converts a surprise into a monitored risk.
Build or endorse a resale channel if your category can support one. An official resale or authentication programme creates the third-party price signal that defends you automatically. Brands still resist this because resale appears to cannibalise new sales, which is a defensible worry in a world without dupes and a serious strategic error in one with them.
Make authentication cheap and public. Where the dupe is visually indistinguishable, the differentiator is verifiability. Serialisation, digital product passports and free authentication services turn an argument about appearance into a binary check.
None of these are marketing campaigns, which is why the factory-video problem tends to be assigned to the function least able to solve it. The response lives in pricing, product, aftersales and legal, and it needs somebody senior enough to convene all four.
Frequently asked questions
Were the TikTok factory claims true?
Mostly not, as far as reporting from CNN, Forbes and NPR in April 2025 could establish. Manufacturers genuinely producing for major luxury brands are typically bound by confidentiality terms, most top luxury handbags remain officially manufactured in Europe, and a substantial share of the goods shown were likely high-quality dupes or counterfeits. Some specific claims, such as manufacturing costs quoted on camera, were assertions with no independent verification.
Did the buy-direct trend actually change where people shop?
Only briefly. DHgate went from around 352nd in the US shopping charts to number two on the US App Store within days in April 2025, then fell to 29th among free shopping apps by May 8, 2025 according to Caixin citing Sensor Tower, after the de minimis exemption for China and Hong Kong ended on May 2, 2025. The channel shift did not persist. The change in consumer beliefs did.
Should we tell customers where our products are made?
If you are not saying it, someone else will characterise it for you, and their version will be worse. The useful disclosure is not the country name but the cost account: what the price consists of and why each component exists. Brands generally find their real breakdown more defensible than the version consumers imagine, which is the whole reason silence is expensive.
Which categories are most at risk?
Those with no liquid resale market and visually replicable products: basic apparel, cosmetics, small electronics accessories, home textiles, and most mid-market fashion. Categories with deep secondary markets are structurally protected, because resale prices are a continuous third-party confirmation that the branded item is worth more, and no video can argue with a realised transaction price.
Will this happen again?
Very likely. The capability was built well before April 2025, the annual cost to a factory of maintaining social visibility is a few thousand dollars, and the underlying incentive is business-to-business lead generation rather than consumer sales, which means tariff changes do not remove it. April 2025 supplied a political trigger. The next trigger will find a more experienced supplier-creator ecosystem.
The number to write down
Take your highest-volume SKU. Write down its retail price, then write down the number a hostile, reasonably informed person would assert as its manufacturing cost. Not your actual landed cost. The number a factory owner would say on camera to make a point.
The gap between those two figures is what your brand is now being asked to justify, in public, by consumers who did not previously think in these terms. If your organisation cannot explain that gap in three sentences that a sceptical customer would accept, you do not have a communications problem. You have an unpriced liability, and the factory videos simply moved it onto the balance sheet.
"The wave sold almost nothing and changed the question everyone asks at the point of purchase. Those are not in tension. The cheapest attacks on a premium are the ones that never needed to make a sale to work."
For the same dynamic playing out inside China, where domestic competitors dismantled foreign price premiums by making comparison free rather than by undercutting, see the guo chao analysis. For how Chinese brands are converting that manufacturing base into premium positions in Western markets rather than cheap ones, see why they are out-premiuming you.
Share:
Free · 24 pages
Read Chapter 1 of the Playbook
AI Foundation Models — how DeepSeek and Qwen crossed a billion downloads while Western boards were counting Mag7 capex, and which layers of the AI stack the rotation has not reached. No excerpt tricks: it is the full chapter.
You will also get the weekly newsletter. Unsubscribe any time — the link is in every email.
The Book · $99
China Market Entry Playbook 2026
456 pages across 22 industry chapters — AI, semiconductors, EVs, payments, logistics, beauty, F&B, healthcare, pets, toys. Every chapter runs the same five questions: the anchor number, the rotation underneath it, which layers of the global stack it has reached, and what to do about it. English and Simplified Chinese editions included.
Ex-Transsion Global Digital Marketing Director · Effie Awards Greater China jury member · Guest lecturer, Peking University School of Economics · Writes Annie Chan Talk, your insider lens on China.
