Protecting Waterproof Bag IP in China: NNN Guide

Buyers send a technical package to three factories, ask each to sign a non-disclosure agreement, and feel protected. They are not, and the reason is specific rather than cynical: an NDA obliges the recipient not to talk about your design, and almost nothing about whether they may make it. A factory can honour a confidentiality clause perfectly while running your pattern on its own line, selling the result to someone else, and never disclosing a word. Closing that gap is the whole job.

This guide covers the difference between non-disclosure and non-use in practical terms, what a properly constructed NNN agreement adds and why the third N is the one that gets enforced, how to make a document genuinely enforceable in China including jurisdiction and language and which legal entity must sign, liquidated damages, what enforcement actually looks like when it happens, why registration has to come before disclosure, who owns moulds and cutting dies and artwork, how to disclose in stages, what to leave out entirely, and the uncomfortable conclusion that supply chain structure protects you more than any contract does. QUANZHOU JUNYUAN BAGS — custom waterproof bag production since 2014, 4,950 m² SGS-verified facility — works to MOQ 500 pieces per style, with sampling in 6–10 working days and bulk in 35–50 days, FOB Xiamen.

Custom waterproof tote with proprietary branding details — A non-disclosure clause stops talking. Only a non-use clause stops manufacturing.

Waterproof backpack with custom hardware and pattern — The tooling is your design in physical form, and whoever holds it holds the product.

Waterproof dry bag produced to a custom specification — Split the supply chain and no single party holds enough to copy you.

Why an NDA on its own is close to useless here

The standard NNN agreement China discussion starts from a misdiagnosis. Buyers assume the risk is that the factory will tell someone about their design. It almost never is. The risk that actually costs money is that the factory will make it — for you, for your competitor, or for its own brand — and a confidentiality clause is silent on that. Non-disclosure governs information. Manufacturing is not disclosure. This is the gap that a well-built programme of waterproof bag IP protection has to close first, before anything else is worth doing.

There is a second, quieter problem: a generic NDA is usually a downloaded template governed by a foreign law, signed by a trading company rather than the producer, in English only, with no stated remedy. If it were ever tested, three of those four defects would be fatal before anyone reached the merits. A document that cannot be enforced is worse than no document, because it produces confidence without protection — and confident buyers disclose more.

And there is a category-specific reason this matters more for bags than for many products. A waterproof bag is not protected by a patent in most cases; it is protected by a combination of registered design, trade mark, and accumulated construction know-how — panel geometry, weld sequencing, the way a roll-top is tensioned, which tape width survives which fabric. Almost none of that is secret once a factory has made it once. The know-how transfers to the line whether or not anyone intended to disclose it.

So the objective is not secrecy in the abstract. It is control over three specific things: whether the factory may make your design for anyone else, whether it may use your tooling to do so, and whether it may sell the resulting product into your markets. Those are use questions, and they need use obligations. The international framework for intellectual property is administered by WIPO, and the trade-related minimum standards most relevant to cross-border manufacturing sit under the WTO TRIPS agreement.

The three obligations an NNN adds, and why non-use is the one that matters

An NNN agreement is not a mystical instrument; it is a confidentiality agreement with two additional covenants. The name comes from the three obligations, and understanding which one does the work tells you how much effort each deserves.

Obligation	What it prohibits	What it does not do	Where it fails in practice
Non-disclosure	Revealing your information to third parties	Does not stop the recipient manufacturing the product	Almost never the actual breach that costs you money
Non-use	Using your information for any purpose other than your orders	Does not by itself stop sales if a copy already exists	The clause that actually matters, and the one most often missing or vague
Non-circumvention	Going around you to your customers, agents or contacts	Does not create rights in the design itself	Useful when you introduce a factory to your own retail accounts
All three together	Talking about it, making it, and selling around you	Still requires registration and enforceability to bite	The realistic minimum for a custom programme

Non-use is the load-bearing covenant and it needs drafting care rather than a heading. It should define the purpose narrowly — manufacture and supply of the identified styles to you alone — prohibit use of the information, the tooling, the patterns and any derivative for any other purpose, and extend expressly to the factory’s own brand and to any affiliated entity. Without that last extension, the obligation binds a company that has no assets while its sister company does the manufacturing.

Non-circumvention is the one buyers over-value. It matters in a specific situation: you introduce a producer to your retail accounts, or you use an agent who controls the relationship. In a direct buyer-to-factory relationship where the factory never meets your customers, it adds little. Include it, but do not mistake it for protection of the design.

Also add the two clauses that make the covenants operational. An obligation to disclose nothing to subcontractors without the same obligations flowing down in writing, and an obligation on termination to return or destroy all files, patterns and tooling, with written confirmation. Both are mundane; both are the difference between having a covenant and being able to prove a breach.

Enforceability: jurisdiction, language version and the correct legal entity

A beautifully drafted NNN that cannot be enforced is a decoration. Four elements determine whether it is real, and they are all unglamorous.

The correct signing entity. Get the full registered name in Chinese, the unified social credit code, and the company chop. A contract signed by a trading company does not bind the factory that actually holds your tooling.
The correct jurisdiction. Choose the courts where the counterparty’s assets and operations are, or arbitration with a seat whose awards are recognisable there. A foreign judgment that has to be recognised before it bites is a slow remedy.
A controlling language version. If you sign Chinese and English, state which governs. A bilingual document with "both equally authentic" invites an argument about translation at exactly the wrong moment.
Liquidated damages stated as a number. An injunction you cannot obtain quickly is worth less than a sum a court will award without you proving loss.
Signature by someone with authority, with the company chop applied. In practice the chop matters more than the signature, and a document without one is frequently rejected as unauthenticated.

The entity point deserves emphasis because it is the most common fatal defect. Many buyers contract with a Hong Kong or mainland trading company that presents itself as the manufacturer, or with a sales office whose name resembles the factory’s but is not it. The NNN then binds an entity with no production assets, no tooling and no employees who have seen your design. Verify the name against the business licence, and check the factory address against the address on the licence.

The chop is the second practical detail. Company seals are the operative signature in mainland practice; a contract with a signature but no chop is weak, and a contract with a chop but no signature is usually fine. Ask for a photograph of the chop on the page and keep the original by courier rather than relying on a scan. Our supplier audit checklist includes the licence and entity verification step that catches this early.

Liquidated damages: turning a promise into a number

The reason most IP clauses never get enforced is remedial rather than substantive. If your only remedy is an injunction, you must persuade a court urgently that irreparable harm is occurring, which is slow, expensive and in practice rarely granted fast enough to matter. If your remedy is a stated sum, you must prove breach and the sum follows.

So state a number, and make it large enough to be a deterrent and small enough to be awards. For a custom bag programme, a figure in the range of one to three times the annual order value is a common construction, sometimes expressed per breach and sometimes per unit of infringing product. Per-unit constructions are often better because they scale with the harm and are easier to evidence from a seizure or an invoice.

Add two supporting provisions. A fee-shifting clause so the prevailing party recovers legal costs, because without it the cost of enforcement can exceed the recovery. And an evidence provision: an obligation on the counterparty to disclose production and sales records on request, or a contractual acknowledgement that records it produces are admissible. Evidentiary difficulty is the main reason IP claims fail in practice, and a contractual admission removes a lot of it.

One caution: liquidated damages that are wildly disproportionate can be reduced by a court as a penalty. Aim for a defensible multiple tied to a real commercial measure — the value of the orders, or a per-unit royalty you could plausibly have charged — rather than an arbitrary large number that looks punitive.

Enforcement reality: what actually happens when a factory copies you

It is worth being honest about the sequence, because expectations here are usually wrong in both directions. Enforcement is not a phone call that stops a line, and it is also not the hopeless exercise it is sometimes portrayed as. It is a procedure with a cost and a timeline, and knowing the shape of it changes how you design the protection in the first place.

Realistically, the sequence runs: you discover the copy, typically through your own market monitoring rather than through any formal channel; you gather evidence, which means buying samples, photographing them, and establishing that the same tooling or the same construction was used; you send a demand referencing the agreement and the registered rights; and if that fails you escalate to an administrative complaint or a court or arbitration proceeding. The first three steps take weeks. The fourth takes months.

What that means practically is that the deterrent matters more than the remedy. A supplier who believes you will detect a copy promptly, have a registered right, and have a document with a number in it, behaves differently from one who believes you will notice eventually and do nothing. Detection capability is therefore part of the protection system, not an afterthought — market monitoring, and being visibly present in the channels where a copy would appear.

It also means the commercial sanction is often more effective than the legal one. For a factory whose revenue depends on a handful of export customers, the credible threat of losing your programme plus a demand circulated to two other customers you know they supply is frequently enough. Legal remedies are the floor; commercial consequences are usually what actually stop the line.

First-to-file: register where it counts, before you disclose

This is the step that most buyers skip and it is the one that determines whether anything else works. Trade marks and designs in China operate on a first-to-file basis. Whoever files first owns the right, regardless of who invented it or who used it first elsewhere. If your brand name or your distinctive design is registrable and you have not filed, someone else can — and in a dispute the registrant is in a far stronger position than the originator.

Right	What it protects for a bag programme	When to file	Consequence of not filing
Trade mark (word and logo)	Your brand name and device on goods and packaging	Before first disclosure to any factory	A factory or agent can register your brand and block your own exports
Design patent / registered design	The ornamental appearance: shape, panel geometry, distinctive closure	Before the design is shown publicly or to a factory	No registered right means no clean basis for an administrative complaint
Copyright in artwork	Prints, patterns, photographic and graphic artwork	On creation; consider recordal	Harder to evidence ownership without a record
Utility model or invention patent	A genuine technical mechanism — a novel valve or closure	Before disclosure, if the mechanism is truly novel	Rarely applicable to bags, and expensive to pursue
Trade secret (contractual)	Process know-how, material specification, weld parameters	Always, via NNN rather than registration	Registration is impossible; contract is the only route

The sequencing rule is simple and unforgiving: file before you disclose. A design shown to a factory before filing may already be public in the relevant sense, and in a first-to-file system that is the end of the registered right. Filing is not slow relative to a sampling cycle, and it is inexpensive compared with a lost programme.

Also register the trade mark in the classes you actually need, including the class covering the goods and, if you sell direct, the class covering retail services. Buyers frequently register in one class, discover that a factory has registered in another, and find their own goods blocked at export. And register the mark in Chinese characters as well as in Latin script if you trade in the region, because a transliterated mark filed by someone else is a common and entirely legal trap.

Moulds, cutting dies and artwork: who owns the physical embodiment

For a manufactured product, the most valuable intellectual property is often not a document — it is a tool. Your cutting die is your panel geometry. Your welding electrode is your seam profile. Your injection tool is your hardware shape. Whoever holds those objects can reproduce your product without holding any information at all, and no confidentiality clause has any effect on physical possession.

Three provisions are needed and they are usually all missing. Ownership: the tooling is yours, acknowledged in writing, whether or not you paid for it directly. Possession and marking: each tool is marked with your reference, photographed, and entered in the supplier’s tool register as belonging to you. And release: on request or on termination, the tools are released to you or your nominated carrier within a stated number of days, with no right of set-off for disputed sums.

Add an exclusivity-of-use provision, which is the one that protects day to day. Absent a clause, a factory holding your electrode will run another customer’s bag on it at night, and you will have no way to detect it. An express prohibition on using your tooling for any product other than yours, with liquidated damages attached, is worth more than most of the confidentiality language combined.

Artwork needs the same treatment in a different form. State who owns the artwork files, whether the supplier may reuse the print screens or embroidery tapes, and what happens to them on termination — returned, or destroyed with written confirmation. Screens and embroidery tapes are cheap to keep and expensive to have in someone else’s drawer. The commercial and cost mechanics of tooling, including what "free tooling" usually means, are set out in our tooling and mould costs guide.

Staged disclosure: never send the complete technical package first

The single highest-value operational habit in this entire subject is staged disclosure, and it costs almost nothing. The instinct is to send a complete tech pack to every factory you are evaluating so you can compare quotations. What that actually does is hand your complete design to three parties, two of whom you will never order from, before any obligation of any kind is in place and before you have any evidence of how they behave.

Stage	What you send	What you hold back	Gate before the next stage
Enquiry	Category, target volume, size range, material family, a mood image	Panel geometry, pattern, artwork, tooling drawings	Signed NNN with the correct entity, plus a price indication you can live with
Quotation	Outline dimensions, material specification, feature list, target price	Full pattern, weld sequence, hardware drawings, artwork files	Commercial terms agreed in principle
Sampling	Full specification for the style being sampled	Artwork finals, tooling for the full range, hardware for future styles	Sample approval and a signed production agreement
Production	Everything needed for the ordered styles	Anything relating to styles not yet ordered	Tooling ownership documented and marked
Range extension	One style at a time	The rest of the roadmap	Performance on existing styles, verified by lot data

Notice what this does to your negotiating position as a side effect. A factory that has only seen an outline specification cannot quote precisely, which is fine, because at enquiry stage you want a range rather than a number. And a factory that has invested in sampling one style with you has something to lose, which is a better position from which to be difficult than one where they hold your entire range and you hold nothing.

It also shortens the argument later. When a copy appears, the first question is how they got the information. A staged disclosure record — what was sent, to whom, on what date, under which agreement — turns that from a suspicion into a document. Our design file submission guide covers what a tech pack needs to contain at each stage.

What to leave out of the tech pack entirely

Some information does not need to be in the technical package at all, and including it is pure downside. This is not paranoia; it is the observation that a factory needs to know how to make your bag and does not need to know how you sell it.

Your retail prices, channel margins or target retail price. Give a target manufacturing cost if you must, never a retail price.
Your customer list, your distributors, or any named retail account. Name the market, not the account.
Your full product roadmap. One style at a time, with future styles disclosed only when ordered.
Your other suppliers, or anything that lets them map your supply chain.
Process parameters you developed yourself, where the factory can achieve the result by its own method. Specify the outcome, not your recipe.
Your material supplier’s identity where only the specification is needed. Specify performance, not the mill.

The fifth item is the one that surprises people. Buyers often send weld temperatures and dwell times copied from a previous supplier, believing they are specifying quality. What they are actually doing is transferring process know-how for free, and usually specifying something the new factory cannot calibrate to anyway. Specify the outcome — bond strength in newtons, leak test method and acceptance — and let the factory determine how to achieve it. That is both better protection and better engineering.

The same logic applies to materials. Naming a mill hands over a sourcing relationship; specifying film thickness, coating weight, hydrolysis resistance and colour fastness gives the factory what it needs to buy correctly and leaves the relationship where it is. Where a specific material is genuinely required, name it in the specification but not in the enquiry.

Supply chain structure protects you better than any contract

This is the core judgement and it is worth stating as clearly as possible: the most effective protection for a bag design is not a document, it is an arrangement in which no single party holds enough of your product to reproduce it. Contracts are what you rely on when structure has failed. Structure is what stops the situation arising.

The reason is straightforward. Every contract-based protection is retrospective: it operates after someone has decided to copy you, and it requires detection, evidence, a forum and time. Structural protection is prospective and needs no enforcement at all, because the thing you are protecting is never assembled in one place. A factory that has your cutting dies but not your welding electrodes, or your panels but not your hardware, or your assembly but not your branded components, cannot produce a competing finished product even if it wants to.

Practical forms of this, in rough order of accessibility. Split tooling ownership so that no single supplier holds the full set, with panels cut at one facility and welded at another. Source distinctive hardware yourself from an independent supplier and consign it, so the bag maker never holds the tooling for the most recognisable part. Use branded components — zipper pulls, buckles, labels, tape prints — that you control the supply of, so a copy requires counterfeiting a component rather than merely making a bag. And consider keeping final assembly or final branding in a second location.

Each of these has a cost: more logistics, more coordination, sometimes a slightly higher unit price. The right comparison is not against zero but against the cost of a copy appearing in your market, which for most brands is a multiple of the margin on the entire programme. Where the design is genuinely distinctive and the volume justifies it, structural protection is usually the better spend. This is also the argument for an OEM rather than ODM relationship: the design originates with you, which is the cleanest structural position of all.

Factory tours, photographs and the information you give away free

Factories learn a great deal from what you do in front of them, and none of it is covered by any agreement because none of it is disclosed information. It is observation, and it is entirely lawful.

The obvious examples: photographing another customer’s product on their line is both rude and instructive to them about what you consider competition; discussing your sales volumes, your growth plans or your retail accounts in a meeting tells them exactly how valuable your programme is and how much slack they have; asking for a quote on six styles so you can compare tells them you are developing six styles. None of these require a breach of contract to be useful to the other side.

Less obvious but more consequential: the questions you ask reveal your priorities. If you spend an hour on weld parameters and none on cost, you have told them weld performance is your differentiator. If you ask whether they can make a component you currently buy elsewhere, you have told them your supply chain and your intent. Be deliberate about what a question reveals.

None of this means being secretive to the point of dysfunction — a factory cannot help you if it does not understand the product. It means deciding in advance what is shareable. Volumes, market category, functional requirements and quality standards are shareable. Margins, customer names, roadmap and process recipes are not. Working with a supplier as a design partner is genuinely valuable, and our collaborative design guide covers how to do it without handing over the parts that matter.

When the leak is on your own side

It is uncomfortable but necessary to say that a meaningful share of design leakage does not involve the factory breaching anything. It involves someone in your own chain — a freelance designer, a sourcing agent, a former employee, a trading partner — who has your full technical package and a relationship with three factories.

Sourcing agents are the most common vector, because their business model is connecting buyers to factories and their incentive is volume rather than exclusivity. An agent who has placed your design with one factory this year may place a near-identical one with another next year for a different client, and if your agreement is with the factory and not the agent, your covenants may not reach them at all. Put obligations directly on any intermediary, name them in the NNN, and make sure the non-use covenant binds them personally as well as corporately.

Two structural controls help. Keep ownership of the design files and issue them rather than sharing a folder, so you can see what was sent and when. And separate who designs from who sources where you can, so no single external party holds the full package plus the factory relationships.

Also consider the mundane controls: watermarked and dated tech packs, files shared as view-only links rather than attachments, and a simple register of who has what. These cost almost nothing and they convert a suspicion into a record — which is exactly what you need if you ever have to act.

A disclosure protocol you can actually run

Everything above collapses into a short operational sequence. It is not complicated, and the discipline of running it consistently is worth more than any single clause.

File trade marks and designs before the first enquiry, not after the first sample.
Verify the legal entity, licence number and company chop before sending anything beyond an outline.
Sign an NNN with non-use as the operative covenant, liquidated damages as a number, and the correct jurisdiction and governing language.
Disclose in stages, holding artwork finals and tooling for unlaunched styles until they are ordered.
Specify outcomes rather than recipes: bond strength, leak test method, acceptance criteria.
Document tooling ownership, marking and release at the time the tooling is paid for, not later.
Register what was sent, to whom, on what date, and under which agreement.
Monitor your own market, because detection speed is what makes everything else credible.

The last item is the one people omit, and it is the one that makes the rest real. A protection system that cannot detect a breach is a system that will never be tested, and an untested deterrent is not a deterrent. Set up a simple recurring search on your own product name and on a couple of distinctive visual features, and actually look at the results.

If you are starting a custom programme and want a counterparty that will sign a proper agreement, this is a reasonable place to begin: MOQ 500 pieces per style, sampling in 6–10 working days with test evidence attached, bulk production 35–50 days, FOB Xiamen, and tooling ownership documented at the point of payment. The sequence from first enquiry through sampling into bulk, including what is disclosed at each stage, is described in the production process guide on our main site.

Frequently Asked Questions

Q1. Why is an NDA not enough to protect a bag design?

Because it prohibits revealing information, not making products. A factory can honour confidentiality perfectly while running your pattern on its own line and selling the result, without disclosing anything.

Q2. What does the second N in NNN actually do?

Non-use prohibits using your information for any purpose other than your orders, and extends to the factory’s own brand and affiliates. It is the covenant that actually stops manufacturing.

Q3. When is non-circumvention worth including?

Mainly when you introduce a producer to your own retail accounts or work through an agent who controls the relationship. In a direct buyer-to-factory relationship it adds relatively little.

Q4. Which legal entity should sign the agreement?

The one that actually holds your tooling and employs the people who see your design. Verify the registered name in Chinese, the unified social credit code and the company chop against the business licence.

Q5. Why does the company chop matter more than a signature?

Because company seals are the operative signature in mainland practice. A contract with a chop but no signature is usually accepted; one with a signature but no chop is frequently rejected as unauthenticated.

Q6. Should the agreement be in English or Chinese?

Usually both, with one stated as controlling. A bilingual document with both versions equally authentic invites a translation argument at exactly the wrong moment.

Q7. Why are liquidated damages so important?

Because proving loss and obtaining an urgent injunction is slow and expensive, while a stated sum follows from proof of breach. A number is a far more practical remedy than an injunction.

Q8. How large should liquidated damages be?

Large enough to deter and small enough to be awards — commonly one to three times annual order value, or a per-unit sum. Wildly disproportionate figures risk being reduced as penalties.

Q9. Why must I register before disclosing?

Because trade marks and designs operate on a first-to-file basis. Whoever files first owns the right regardless of who created it, and a design shown before filing may already be public.

Q10. Should I register my brand in Chinese characters?

Yes, if you trade in the region. A transliterated mark registered by someone else is a common and entirely legal way to block your own exports.

Q11. Who should own the cutting dies and welding electrodes?

You, in writing, with marking, a register entry and a release clause. The tooling is your design in physical form, and possession matters more than ownership on paper.

Q12. What is staged disclosure?

Sending an outline at enquiry, a full specification at sampling and artwork or tooling only when a style is ordered. It means two of three evaluated factories never hold your complete package.

Q13. Should I specify weld parameters in my tech pack?

Usually not. Specify the outcome — bond strength in newtons, leak test method and acceptance criteria — and let the factory determine the recipe. You protect know-how and get better engineering.

Q14. Is supply chain structure really better protection than a contract?

Yes, because it is prospective. If no single party holds the full set of tooling, components and artwork, a competing finished product cannot be assembled regardless of what any agreement says.

Q15. Can a sourcing agent be the leak?

Often. Their incentive is volume rather than exclusivity, and if your agreement is with the factory only, the covenants may not reach them. Put obligations on intermediaries directly and in their personal capacity.

Q16. What should I never put in a technical package?

Retail prices and margins, customer and distributor names, your full product roadmap, your other suppliers, and process recipes you developed yourself.

Q17. How do I detect a copy early?

Run a recurring search on your product name and on distinctive visual features, and monitor the channels where a copy would appear. Detection speed is what makes every other protection credible.

Protecting Waterproof Bag Designs in China: NNN Agreements and Beyond