Why Delisting Counterfeits Is Not the Same as Stopping Them

TL;DR

• Temu reports proactive removals outpacing complaint-based takedowns 331 to 1, with a 99.97% cleanliness rate independently assessed by REACT.
• Blocking a listing removes the storefront, not the factory. The same stock relists, rebrands, or moves to a channel with weaker screening.
• The report's real deterrence signal is offline: test purchases that suspend sellers in four days, plus engagement with Europol, OLAF, and the WCO.

Temu published its latest Intellectual Property Protection Report in June 2026, and the headline number is hard to ignore. Across the year to May 2026, the marketplace says its systems removed suspected infringing listings proactively at a rate of 331 for every one pulled after a rights holder complained, more than 60 percent up year on year. An independent assessment by REACT put the platform's cleanliness rate, across listings involving alleged counterfeiting and unfair trademark use, at 99.97 percent. On paper, the storefront is close to spotless.

The machine behind the 99.97 percent

The report is, above all, a description of scale. Temu's proactive monitoring database now holds more than 47 million images and over 9.5 million keywords, a roughly ninefold jump in image coverage from the previous period. It screens listings with text and image recognition, logo detection, and optical character recognition before a product can go live, then keeps scanning after publication. More than 40 percent of seller applications were rejected during onboarding. Over 16,000 stores were terminated for repeat violations. The average complaint now resolves in under 24 hours. Direct collaboration reached more than 3,000 brands, around 500 of them small and medium-sized businesses that rarely have the legal muscle to police a global marketplace on their own.

Read as an operations report, it is genuinely impressive. The whack-a-mole that brand-protection teams have complained about for two decades, spot the fake, file the notice, wait, repeat, is being replaced by interception before the listing publishes. Blocking a listing before a customer ever sees it is strictly better than removing it after the sale.

A blocked listing is not a stopped counterfeiter

Here is the limit of the model. Every figure in the report counts listings, sellers, and searches. None of them count counterfeiters, and those are not the same thing.

A delisted listing is a removed storefront, not a removed factory. The producer who printed 10,000 fake jerseys does not stop printing because one URL went dark. The stock still exists. It relists under a fresh seller name, migrates to another platform, moves through social channels, or ships to a market with weaker screening. The 99.97 percent figure describes the shelf, cleaned continuously. It says almost nothing about the warehouse behind the shelf, and a marketplace has no line of sight into that warehouse at all.

This is the structural blind spot of platform-side enforcement. It acts on representations of goods, the title, the image, the keyword, rather than the goods themselves. You can suppress the representation perfectly and leave the physical supply completely intact. For the class of counterfeiting that runs on organized supply chains, and often braids into other serious crime, a delisting is a cost of doing business, not a reason to stop.

The economics make the gap worse. Opening a new seller account is close to free, while producing the counterfeits is where the money and the risk already sit. When enforcement only ever reaches the account, it taxes the cheapest and most replaceable part of the operation and never touches the expensive part. Temu's own repeat-offender figure hints at this. More than 16,000 stores were terminated over the year, a number that is large precisely because banned sellers keep returning under new names. Termination at that volume is not a sign the problem is solved; it is a sign the same actors are being removed again and again.

Where the report quietly touches the physical world

The most interesting parts of the report are the smallest. Between the dashboards are the few places where Temu stops behaving like a search engine and starts acting on physical goods and real accountability.

One is the Test Purchase Initiative, launched in 2025. In the case Temu cites, it enabled identification of counterfeits, blocking of the sales, and suspension of the sellers within four business days. That is a different act from delisting. Someone bought the product, confirmed in the physical world that it was fake, and a seller lost their account as a result. It ties an online listing to an offline fact.

The other is the company Temu now keeps. The report lists engagement with the UK Intellectual Property Office, WIPO, Europol, the European Anti-Fraud Office known as OLAF, and the World Customs Organization. In December 2025 the platform banned the sale of counterfeit airbags in the United States, a move the Automotive Anti-Counterfeiting Council welcomed on safety grounds. Customs bodies, fraud offices, and police forces can do the thing a marketplace cannot: hold the shipment, open the container, and put a name in front of a prosecutor.

Deterrence is an offline cost

Deterrence is not the same as removal. Removal is what happens to a listing. Deterrence is what happens to the person deciding whether to make the next batch. It requires that person to face a cost a new seller account cannot absorb: seized inventory, a customs hold, a referral, a court date.

None of that shows up in a cleanliness rate, which is exactly why the cleanliness rate is the wrong number to judge the fight by. A platform can be 99.97 percent clean and sit directly downstream of a supply chain that is entirely intact. The 331-to-1 ratio proves the detection works. It does not prove that anyone stopped.

The real value in Temu's report is that it is unusually specific about the online layer and, read closely, honest about where that layer ends. The proactive systems are real and they help. But the report's own most durable moments are the offline ones: the test purchase that ended in a suspension, the airbag ban, the seat at the table with Europol and the World Customs Organization. That is where a listing stops being a listing and becomes evidence. The rest, however well done, is housekeeping.

Hero image: "Shopping online with bank card" by Bogdan Hoyaux / European Commission, licensed under CC BY 4.0, via Wikimedia Commons.

Source

• Temu: Temu Intellectual Property Protection Report (June 2025 to May 2026)