Your Trademark Is Registered. Nobody Is Watching It.

CNIPA publishes every application before it registers, and that publication is the only moment you can stop a squatter cheaply. The window is sixty days under the 2026 revision, down from ninety. It starts on the publication date, not the day you find out. Monitoring is the one part of trademark ownership with no invoice attached, which is exactly why it is the part that gets skipped.

Registration feels like the end of the project. The certificate arrives, the file gets closed, the renewal date goes into a calendar, and the brand owner moves on to the next thing.

For the squatter, that moment is the beginning.

China operates a first-to-file system. Whoever files first owns the mark, with narrow exceptions. That rule is not a loophole or a malfunction -- it is the deliberate architecture of the register, and it is explained without apology in the first-to-file explainer. What follows from it is that the register is a live competitive space. Every week, thousands of applications enter it. Some of them are yours, filed by someone else.

Nobody tells you. CNIPA does not notify prior holders when a similar application is published. There is no alert, no email, no equivalent of the watch service that trademark offices in some jurisdictions offer as a matter of course. The register is public and the obligation to read it belongs entirely to the party with something to lose.

Quick answer: Every Chinese trademark application is published before registration and can be opposed within two months of publication under Article 36 of the 2026 revision, down from three months. The clock runs from publication, not from discovery. Miss it and the mark registers -- but that is not the end: an invalidation request under Article 51 is available for five years from registration on relative grounds, and under Article 50 with no time limit on absolute grounds. Separately, a registration that has not been used for three consecutive years can be cancelled by anyone under Article 57. Official fees: CNY 450 for opposition, about USD 67; CNY 675 for review, about USD 101; CNY 450 for cancellation, about USD 67, at the October 2026 rate.

The Window Nobody Watches

The Chinese trademark process has one gate that a third party can walk through, and it is narrow.

An application is filed, examined, and if it clears examination it is published in the Trademark Gazette for opposition. That publication is the moment the mark becomes visible to the world and the moment the clock starts. Under Article 36, anyone may oppose within two months of the publication date. After that, the mark registers and the gate closes.

Two months sounds generous until you subtract what actually has to happen inside it. A brand owner has to notice the publication, decide the filing is worth fighting, assemble evidence of prior use or prior rights, draft the opposition, translate the supporting documents into Chinese, pay the fee, and file. In a well-run process that is four to six weeks of work. In a process where the filing is discovered on day fifty, it is ten days of panic and a filing that is either late or thin.

That is the real argument for monitoring. Not that it produces information -- the information is public and free -- but that it buys the calendar time needed to use the information properly. A squatter caught on day three of publication is a manageable problem. The same squatter caught on day fifty-five is an expensive one, and caught on day seventy it is a different problem entirely.

What Changed on January 1, 2027

The 2026 revision shortened the opposition period from three months to two. Article 36 now reads that within two months of the publication date, a prior right holder or interested party may oppose on relative grounds, and any person may oppose on absolute grounds.

Thirty days is not a large number in absolute terms. It is a large number relative to the work the window has to contain. Under the old three-month period, a brand owner monitoring monthly had a reasonable chance of catching a filing and still having six or seven weeks to act. Under a two-month period, monthly monitoring can consume half the window before anyone knows there is a problem.

That is the concrete reason the monitoring interval has to change with the law. A quarterly or monthly watch cycle was defensible under a ninety-day window. It is not defensible under a sixty-day one. The comparison between the old and new periods, and what it does to filing practice, is set out in the sixty-day opposition deadline analysis and in the two-month opposition playbook.

Why Watching Your Own Mark Is Not Enough

The most common monitoring failure is not that nobody watches. It is that the wrong thing is watched.

Brand owners monitor for their exact mark. That is close to useless, for a structural reason: an application for an identical mark in an identical class will usually be refused by the examiner, who cites the earlier registration automatically. The system already catches exact copies. It does not need your help.

What the system does not catch reliably is the near miss. A squatter who understands the examination guidelines does not file your mark. They file something that survives citation:

• An extra character. A suffix or prefix that changes the visual impression enough to argue the marks are distinguishable.
• A different transliteration. Your brand has a Chinese name; the squatter files a homophone using different characters with the same pinyin. On paper they are different marks. In a supermarket aisle they are the same word.
• A swapped element. A composite mark where the word element stays and the device element changes, or the reverse.
• An adjacent class. Your registration covers class 25 apparel; the squatter files class 18 bags and class 35 retail services. No citation, because the classes do not conflict under the examiner's grouping -- but the shelf does.
• A sub-class shift. Within the same class, a move into a different similar group. The Chinese class system is divided into subclasses and similar groups, and two marks in the same class can avoid citation entirely if they sit in different groups. The structure is explained in the subclasses and similar groups guide.

Effective monitoring therefore covers four dimensions at once: the Latin word mark, the Chinese character version, the pinyin transliteration, and the device or logo element -- each across the core classes plus the two or three adjacent classes where the brand realistically appears on a shelf. That is a broader watch than most brand owners set up, and it is the difference between a watch that produces alerts and a watch that produces catches.

Four Kinds of Squatter

"Squatter" is used as a single word for four different actors, and they require different responses. Treating them as one category is why enforcement plans misfire.

The supplier or factory

The most common, and the most damaging, because the relationship makes it personal. A factory that has produced goods bearing the mark for two or three years files the mark in its own name, sometimes out of genuine misunderstanding, sometimes to secure leverage in a price negotiation. The response here is rarely a pure legal one: it is a commercial negotiation backed by the registration and, where available, a customs recordation. The pattern and its remedies are covered in the supplier hijack recovery guide.

The former distributor

An ex-partner who registered the mark locally during the relationship, on the theory that they were building the market and deserved the asset. These cases turn on the contract and on evidence of the relationship, and the invalidation route under Article 22 -- unauthorized registration by an agent or representative -- is the specific tool. They are also the cases most likely to end in a buy-back, and the economics of that negotiation are set out in the buy-back negotiation guide.

The professional hoarder

A shell company holding hundreds of marks across unrelated classes and industries, filing not to use but to sell. This is the actor Article 19 of the 2026 revision was written for, and it is discussed below. These are also the cases where cancellation for non-use is most effective, because a professional hoarder almost never uses anything.

The competitor

A genuine trade rival filing a blocking mark in a class where it does not operate, to prevent the brand from expanding. Rare, expensive, and usually visible early -- a competitor's filing pattern is easier to predict than a hoarder's.

The general framework for assessing whether a filing was made in bad faith, and what evidence supports that characterization, is set out in the bad faith registration defense guide. What matters at the monitoring stage is that the category determines the route, and the route determines the deadline.

Three Routes, Three Clocks

Once a hostile filing is identified, there are three procedural routes and they are not interchangeable. Each has its own clock, its own evidentiary burden, and its own cost.

Two things about this table matter more than the rest of it.

The first is that invalidation under Article 51 has a five-year window. That is not a rounding error on the opposition deadline -- it is a completely different timescale. A brand owner who misses the sixty-day window has not lost the mark. They have changed which proceeding they will use, and increased the cost. That distinction is the single most valuable piece of information in this article, because the opposite belief causes brand owners to give up on marks that were entirely recoverable.

The second is that cancellation for non-use requires proving nothing about the brand owner at all. It is discussed separately below because it is the most underused route in the entire system.

The Cancellation Route Nobody Uses

Article 57 provides that where a registered trademark has not been used for three consecutive years without justified reason, any entity or individual may apply to cancel it. The fee is CNY 450 online, about USD 67.

Read the evidentiary requirement again. To oppose, a brand owner must prove prior rights, prove the marks are confusingly similar, and usually prove reputation or bad faith. To invalidate, they must prove a ground. To cancel for non-use, they must prove one thing: that the registrant has not used the mark in China for three years.

The burden is reversed. Instead of the brand owner proving who they are, the registrant has to prove they used the mark. In practice that means producing invoices, contracts, packaging, advertising, and dated evidence tying the mark to goods or services in the Chinese market. A squatter holding a mark as an asset has none of it. A professional hoarder with three hundred marks has none of it for any of them.

There is a strategic asymmetry here that is worth stating plainly. Opposition is a fight about identity: who has the better claim to the mark. Cancellation for non-use is not a fight about identity at all. It is a compliance check, and the squatter usually fails it. For a mark that has already registered and where the opposition window has closed, cancellation is frequently the cheapest route to the same outcome -- and it does not require the brand owner to establish that they are anybody in particular.

The mechanics, the evidence a registrant typically produces, and the defenses that actually work are covered in the non-use cancellation guide.

The Five-Year Mistake

The belief that a missed opposition deadline means a lost mark is widespread and wrong. It is worth taking apart carefully because it changes the entire posture of a monitoring programme.

Under Article 51, where a registered trademark violates Articles 20 to 22, Article 23(1) or Article 24 -- the relative grounds, covering prior identical or similar marks, well-known marks, and unauthorized agent filings -- a prior right holder or interested party may request invalidation within five years of the registration date. The decision period is twelve months, extendable by six months.

There is an exception written into the same article: for a well-known mark registered in bad faith, the five-year limit does not apply. A famous brand facing a bad-faith registration is not on a clock at all.

Under Article 50, where a registration violates Articles 15, 16(1), 17 to 19, or 25 -- the absolute grounds -- the authorities may invalidate on their own initiative, and any entity or individual may request it. No time limit. The decision period is nine months, extendable by three.

So the practical picture is this. Opposition is fast, cheap and narrow. Invalidation is slower, more expensive, and far more forgiving on timing. A brand owner who discovers a squatter's registration two years after it registered is not out of options -- they are in the most common scenario, and Article 51 was designed for it.

What changes is the evidence. In an opposition, the question is whether the marks conflict. In an invalidation filed years later, the question is usually broader: whether the registrant acted in bad faith, whether the brand was well known at the relevant date, whether there was a relationship. That is more evidence to assemble and more time to assemble it. Which is why the honest framing is that missing the window costs money, not rights.

Article 19: The New Attack Surface

Article 19 of the 2026 revision provides that an application for trademark registration that is not made for the purpose of use and clearly exceeds the normal business needs of the applicant shall not be registered. It also prohibits filing by deception or other improper means.

The second sentence is a restatement of existing bad-faith doctrine. The first sentence is new, and it is operationally useful in a way that bad-faith arguments never quite were.

Bad faith is a state of mind. It is proved indirectly, through patterns, relationships, prior knowledge and commercial context, and it is always contestable. "Clearly exceeds normal business needs" is closer to an observable fact. A company with two employees, registered capital of CNY 1 million, and four hundred trademarks across forty unrelated classes is not a difficult case to argue. Neither is a registrant whose entire portfolio consists of foreign brands it has no relationship with.

For monitoring purposes this changes what is worth recording. Under the old framework, a suspicious filing was worth noting but hard to act on until you had built a bad-faith case. Under Article 19, the registrant's overall portfolio becomes evidence in itself -- the size, the spread, the absence of any operational footprint. That is a monitoring output, not just an enforcement input, and it is why a watch programme should record the applicant's filing history alongside the mark itself.

Article 19 sits inside a broader set of changes to the 2026 revision, and its interaction with the bad-faith provisions is set out in the eight changes analysis and in the 2027 compliance checklist. One related development is worth flagging separately: the authorities now have ex officio cancellation power, meaning a registration can be removed without anyone filing against it. That mechanism is covered in the ex officio cancellation analysis.

How to Actually Monitor

Monitoring is not a product you buy once. It is an operating rhythm, and the rhythm is set by the law's clock rather than by convenience.

The interval

Every two weeks during normal operation. Weekly once a suspicious filing appears, until it is resolved.

The reasoning is arithmetic. An application typically reaches publication four to six months after filing. A two-week check means a squatter's publication is caught within fourteen days, leaving roughly forty-six of the sixty days for evidence, translation and filing. A monthly check means up to thirty-one days consumed, leaving twenty-nine -- workable but tight. A quarterly check is a coin flip.

The scope

Four dimensions, all of them, every cycle:

1. The Latin word mark, including close variants -- one letter changed, a plural, a hyphen, a spacing difference.
2. The Chinese character mark, including visually similar characters and characters that differ only in a radical.
3. The pinyin transliteration, including homophones with different tones and different characters.
4. The device element, where the brand has a distinctive logo, checked as a visual similarity search rather than a text search.

Across the core classes plus adjacent classes. A brand in class 25 apparel should be watching class 18, class 24 and class 35. A brand in class 3 cosmetics should be watching class 5, class 21 and class 35. The adjacency is commercial, not legal -- it is where the brand appears on a shelf next to itself.

The record

Every hit gets logged with the application number, the applicant name, the applicant's address, the filing date, the publication date, the classes, and the applicant's total portfolio size. That last field is what turns a list of alerts into an Article 19 argument, and it takes ten seconds to record at the moment of the hit and considerably longer to reconstruct later.

The decision

Not every hit is worth fighting. A filing in an unrelated class by an unrelated company with a genuinely different mark is noise, and treating noise as signal produces fatigue, which produces a monitoring programme that gets abandoned in month four. The filter is commercial: does this filing threaten a market the brand actually sells in or intends to enter within two years? If yes, act. If no, log it and move on.

What It Costs

Official fees for the three routes, at the October 29, 2026 mid-rate of 6.6805 CNY per US dollar:

Paper filings cost CNY 500 and CNY 750 respectively -- about USD 75 and USD 112. The difference is small enough that the choice should be driven by whether the filing needs to be accompanied by physical evidence, not by the fee.

These numbers are not the real cost, and pretending otherwise would be misleading. Agent fees for an opposition or invalidation in China commonly run several times the official fee, and where translation, notarization and evidence collection are involved, substantially more. The honest comparison is not "USD 67 versus doing nothing." It is "the cost of a contested proceeding now versus the cost of buying the mark back later, or losing the market segment entirely." The buy-back alternative has its own price range, and it is rarely lower. That comparison is worked through in the buy-back negotiation guide and in the trademark squatters overview.

Five Moves

1. Put the watch on a two-week cycle and give it an owner. Not a quarterly review, not an annual portfolio audit. Two weeks, named person, calendar entry. The law gives sixty days; the watch has to be faster than the law.
2. Widen the watch to near misses. Latin word mark, Chinese characters, pinyin homophones and the device element, across core and adjacent classes. Watching only the exact mark catches only the filings the examiner would have refused anyway.
3. Record the applicant, not just the mark. Application number, applicant, address, filing date, publication date, classes, and total portfolio size. The portfolio size is an Article 19 argument waiting to be made.
4. Diarize all three clocks, not one. Sixty days to oppose. Five years to invalidate on relative grounds, with no limit for a well-known mark registered in bad faith. Three years before non-use cancellation becomes available. A missed opposition is a change of route, not a loss.
5. Check the register before any expansion, not after. New class, new product line, new Chinese brand name -- run the search first. The cost of discovering a blocking registration before launch is a search fee. The cost of discovering it after launch is a rebrand.

The Part With No Invoice

Trademark ownership in China is usually described as a filing problem. It is not. Filing is a one-time transaction with a clear price, a clear process and a clear deliverable. Ownership is an ongoing position in a live competitive register where thousands of applications enter every week and some of them are aimed at the brand you spent two years building.

The register is public. The gazette is free. The information has never been the scarce resource. What is scarce is the discipline to read it on a fixed interval, the judgment to distinguish a near miss from noise, and the willingness to spend sixty-seven dollars on an opposition instead of discovering the problem in year three and spending five figures on a recovery.

Nobody sends a reminder about any of this. The squatter is counting on that.

Sources:

• Trademark Law of the People's Republic of China (2026 Revision) -- CNIPA official text, adopted June 26, 2026, effective January 1, 2027: Article 19 (use requirement and bad faith), Article 36 (two-month opposition period), Article 50 (absolute grounds invalidation, nine months), Article 51 (relative grounds invalidation, five years, twelve months), Article 57 (three-year non-use cancellation)
• National People's Congress -- full text of the 2026 Trademark Law revision, adopted at the 22nd session of the 14th Standing Committee
• CNIPA Trademark Fee Guide -- official fee standards: opposition CNY 450 electronic, review CNY 675 electronic, cancellation CNY 450 electronic, updated October 9, 2026
• China Trademark Office -- Trademark Gazette, published by issue for public opposition and registration notices
• CNIPA -- official announcements and the 2026 Trademark Law revision notice, June 26, 2026
• Bank of China -- published foreign exchange rates, the reference used for the USD conversions in this article (October 29, 2026 mid-rate: 6.6805 CNY)