How to Develop a Custom Product in China from Scratch

Developing a custom product in China from scratch means taking an idea through seven stages: concept validation, design and engineering, finding a development-capable factory, prototyping in rounds, tooling and molds, a pilot run, and finally mass production. For a moderately complex product, the journey typically runs three to nine months, and tooling alone can cost anywhere from a few hundred dollars to tens of thousands, which is why understanding the process before committing money matters.

Modifying an existing catalog product is faster and cheaper, but the moment you need something that doesn't exist yet, a custom shape, a new mechanism, a different material, you're in new product introduction territory. CN Ally manages this process for brands on the ground in China: finding factories with real development capability, coordinating prototypes, and keeping the handoff to mass production honest. Here's how the seven stages fit together, where the money goes, and where first-time developers get stuck.

What are the stages of developing a custom product in China?

The process runs through seven gates, each one a go-or-no-go decision:

Stage · What happens · Typical duration

• 1. Concept validation: Prove someone wants it and it can be made · 1–3 weeks
• 2. Design & engineering: CAD models, materials, design-for-manufacturing review · 2–6 weeks
• 3. Factory selection: Find a partner that develops, not just manufactures · 2–4 weeks
• 4. Prototyping: Proof of concept, then functional prototypes in rounds · 3–8 weeks
• 5. Tooling & molds: Build production tooling; the biggest upfront cost · 3–6 weeks
• 6. Pilot run: A short production run from final tooling · 2–4 weeks
• 7. Mass production: Scale up with quality control baked in · Ongoing

Durations are indicative, not guarantees, and skipping a stage, approving tooling before a functional prototype exists, is how development budgets get burned.

Validate the concept before spending a dollar on tooling

The most expensive mistake in product development isn't a bad factory. It's a bad product manufactured perfectly. Before engineering begins, answer two questions honestly.

First, market validation: does anyone want this, at a price covering manufacturing, shipping, duties, and your margin? Study what's already selling, read customer complaints in reviews, and check whether your idea solves a real one. If you can't name the buyer and the price they'd pay in one sentence, you're not ready.

Second, technical feasibility: can this be made with processes available to you? A product needing micro-precision machining has a different development path than a sewn textile or a molded plastic part. Talk to factories or an engineer at the concept stage, not after you've committed to a design. As one China-based engineering firm notes in their development series, prove functionality at the concept level before detailed engineering, even with a rough model built from whatever is at hand.

Turn the idea into engineering: CAD, materials, and DFM

Once the concept survives validation, it becomes engineering data. For most physical products, that means 2D drawings and a 3D CAD model defining every dimension, material, and tolerance.

If you can't produce CAD yourself, this is where many projects need their first outside help: an industrial designer or mechanical engineer. Factories can't reliably invent your product from a description. A hand sketch gets you quoted; proper CAD gets you quoted accurately.

Two decisions in this phase save more money than anything else later.

Material selection. Material determines the process, tooling cost, and regulatory burden: food-grade silicone faces different compliance testing than the same shape in ABS. Pick materials early and confirm availability at production scale.

DFM, or design for manufacturability. This is where an experienced factory reviews your design and flags trouble: walls too thin to mold, tolerances tighter than the process can hold, assembly steps a redesign could simplify. Hours of DFM review at the drawing stage replace weeks of tooling rework later.

Set a target unit cost now and design toward it, not after the mold is cut. Your first prototype will almost always come back too expensive, and a cost target lets you cut surgically instead of panicking.

Find a factory that develops, not just one that manufactures

This is the most consequential decision in the process, and the one importers get wrong most often. Many Chinese factories are excellent at running someone else's design. Far fewer have engineering teams that can co-develop a product, iterate on drawings, suggest material substitutions, and troubleshoot a prototype. Those are two different capabilities in two different factories.

In our guide to OEM vs ODM we explain the two models; for custom development you want an OEM-capable factory at minimum, or an ODM/JDM partner with its own engineering if you need design help.

How to vet development capability:

• A real development factory has engineers and rapid prototyping equipment (3D printers or CNC). A pure production factory has production lines and not much else.
• Ask how they handle engineering changes. You want to hear about revision control, updated drawings, and signed change approvals, not "just tell us what to change."
• Ask what happens when a prototype fails. Experienced development factories expect iterations and have a workflow for them. Inexperienced ones treat it as your problem and wait for instructions.
• Check their export history: factories that have developed products for overseas brands can usually describe the process, and our verification guide covers how to check these claims.

One more filter: ask for feedback on your drawings before quoting. Specific technical questions mark a development partner; a quote within hours without questions means they haven't understood your product. And if you're still deciding between full custom development and modifying an existing design, read our private label guide first; the hybrid path, a proven base with custom tooling for the parts that differentiate you, is often the smart move.

The prototyping cycle: expect rounds, not miracles

Prototyping runs in cycles, not straight lines. A typical progression:

1. Proof of concept. A rough model testing whether the core idea works. It can be wood, cardboard, and hot glue. Its only job: does this function at all?
2. Looks-like prototype. Focuses on appearance, size, and ergonomics, often 3D-printed. Lets you hold the product and judge whether people will pick it up.
3. Works-like prototype. Uses production-intent materials and mechanisms, where engineering problems surface: parts that bind, assemblies that fail, electronics that overheat.
4. Works-like-looks-like prototype. Combines both. This is the unit you approve as the production reference, and it becomes your golden sample (see our golden sample guide for managing it).

The first physical sample from your spec typically takes 6–10 working days, each revision round another 5–7 days, based on the published process of one China manufacturer we've reviewed. Two to three revision rounds is normal. Five rounds means vague drawings or a factory that's guessing; diagnose which it is.

Make every revision a written change order: what changed, why, what it costs. Verbal agreements are how a $2,000 prototype phase becomes a $10,000 one with nothing documented.

Tooling and molds: the point of no return

Tooling, molds, and dies are the hard tooling that turns prototypes into production parts, and usually the largest single upfront investment, the one most importers underestimate.

On cost, the range is enormous. One Chinese injection molding company, Zhongde, states publicly that its injection molds start from $500, while complex multi-cavity steel molds for large-scale production can exceed $100,000 (see their mold cost breakdown). A sourcing analysis of small-volume molding services puts typical tooling investment in the $1,000–$15,000+ band. The drivers: mold material (aluminum for prototypes vs. hardened steel for production), part size, complexity, cavity count. Get at least two quotes for the same spec; identical tooling quotes can vary dramatically between shops.

Three things to get right about tooling:

Who owns the mold. Many Chinese factories assume in-house tooling stays in-house. If you paid for it, the contract should say the mold is your property, you can take it elsewhere, and the factory maintains it. Losing mold ownership locks you into one supplier forever.

Prototype molds first. Cheaper aluminum or soft-steel tooling for the pilot run lets you confirm the product sells before investing in hardened production steel.

Tooling trials. A mold is done when it produces parts that meet spec, not when it's delivered. Budget two to three trial runs (T1, T2, T3) with your approval on parts from each.

The pilot run: a dress rehearsal for mass production

Between approved prototypes and full production sits the pilot run, sometimes called a trial order: a short run, often a few hundred units, made with production tooling on the production line.

The pilot run is not a formality. It catches what prototypes can't: how the line handles your product, whether assembly is clear, whether the defect rate at real speed is acceptable.

What to do with a pilot run:

• Inspect it like production: our initial production check guide covers evaluating the first units off the line.
• Test the units hard. Run them, drop them, ship a few to yourself overseas. Pilot units are your last cheap chance to find durability problems.
• Lock the spec after it. Once the pilot passes, the design is frozen; any later change is a new engineering change, not a casual tweak. Suppliers who "improve" things between pilot and bulk order without telling you cause the defect-rate problems in our defect reduction playbook.

If the pilot run reveals fundamental problems, go back. It costs weeks now; it costs a container later.

How long does custom product development take in China?

Nobody can give an exact timeline without knowing the product, but the stages add up predictably. Here's a realistic timeline for a moderately complex consumer product, say a small electronic device:

Phase · Weeks · Cumulative

• Concept validation: 1–3 · Month 1
• Design & engineering (CAD, DFM): 2–6 · Months 1–2
• Factory selection & quoting: 2–4 · Month 2
• Prototyping (2–3 rounds): 3–8 · Months 2–4
• Tooling build + trials: 3–6 · Months 4–5
• Pilot run + evaluation: 2–4 · Month 5–6
• Mass production ramp-up: 2–4 · Months 6–7

Simple products (a new shape in an existing category) can move through in three months. Electronics with firmware, certifications, and custom PCBs routinely take nine months or more. Chinese New Year shuts most factories for three to four weeks in January or February and can add a full month to any timeline it touches. Plan around it.

The calendar killer is usually decision latency on your side and revision cycles caused by vague specifications. A clear spec, quick approvals, and a factory that asked hard questions up front compress every phase after them.

Protecting your idea and your investment

Two separate risks, two separate answers.

Intellectual property. China is a first-to-file trademark system: register your trademark before sharing your product, because the first filer owns the rights, not the first user. For appearance, consider a Chinese design patent (外观设计专利), faster and cheaper than Western equivalents. Manufacturing agreements should include confidentiality and non-compete clauses naming your product. None of this is perfect protection, but brands copied with zero recourse are overwhelmingly the ones that registered nothing and signed nothing. For serious investments, consult an IP attorney who works in China; this isn't legal advice.

Commercial protection. Pay in stages tied to gates: a deposit for development, then releases at prototype approval, tooling completion, and pilot acceptance. Never pay full tooling cost before seeing a sample. Get everything in writing and keep the revision log. If the relationship ends, your drawings, mold ownership clause, and golden sample are what let you move production elsewhere.

Frequently asked questions

How much does it cost to develop a custom product in China?

It depends on complexity and tooling. Simple products can be developed for a few thousand dollars total; complex electronics with custom molds and certifications can run into tens of thousands before the first production unit. The main drivers are engineering time, prototype rounds, and tooling: injection molds range from under $1,000 for simple prototype tooling to tens of thousands for production steel molds. Get itemized quotes per stage so you can stop at any gate.

Can a factory in China develop my product from just a sketch?

Some development-capable factories can turn detailed sketches into engineering drawings, especially for simpler products. But the factory will make dozens of design decisions on your behalf, and you may not like the product that comes back. The better path: develop the design yourself or with an engineer first, then let the factory handle DFM and prototyping. If you truly have only a sketch, work with a sourcing agent or development partner on the ground.

Who owns the mold if I pay for tooling?

You should, but only if the contract says so. Put it in writing before paying: the mold is your property, the factory maintains it, and you can transfer it elsewhere. Without this clause, many factories assume tooling stays with them, locking you into that supplier.

How do I know if a factory can actually develop products, not just manufacture them?

Look for an in-house engineering team, rapid prototyping equipment, a structured revision process, and a history of developing products for overseas brands. During quoting, ask for technical feedback on your drawings: factories that ask sharp questions can develop; factories that quote instantly without questions usually can't.

What if development fails and the product doesn't work?

That's what the gates are for: failure should happen cheaply. If the proof of concept doesn't work, you've lost hundreds, not tens of thousands. Set kill criteria in advance: unit cost missed, core mechanism fails after two redesigns, timeline slips past your market window. Walking away from a $3,000 failed prototype is good business.

Should I develop from scratch or modify an existing product?

Modify when speed and cost matter more than uniqueness: most successful private label brands start from a proven design and customize the differentiating parts. Develop from scratch when the product can't be built from existing designs, or when owning the design matters strategically. The hybrid path, an existing base plus custom tooling for key components, is often the best of both.

Your next move: decide whether the product justifies the journey

The decision rule is simple: develop from scratch only if the product can't be built by modifying something that exists, or if owning the design is strategically worth the cost and time. If neither is true, start with customization and earn your way to full development.

If the journey is worth it, the highest-leverage move is assembling the right team before spending: clear specs, a realistic budget, and people on the ground to vet factories, manage prototypes, and keep tooling ownership where it belongs. CN Ally supports brands through this process, from factory selection to inspecting the pilot run and first mass production. If you want experienced eyes on your factory shortlist or development plan, write to hi@cnally.com and tell us what you're building.