Three ways to get a toy made in China, three completely different risk profiles. How to read where the margin sits, where the accountability stops, and which model survives contact with your second order.
Most importers choose between an agent, a trading company and a factory by comparing unit price. That is the wrong number, because the three models are not selling the same product.
They differ in who holds the contract, who carries the quality risk, who owns the tooling, and what happens on the morning something goes wrong. An agent who is 4% cheaper per unit can cost you a season. A factory that quotes 12% below a trader can leave you unable to produce a second SKU in the same category.
I have been on all three sides of this table for 21 years — as a buyer, as a trading company, and now as a manufacturer that also sources components for customers. This guide is what I would want a first-time importer to read before signing anything.
The labels get used loosely, so let us pin them down by who is legally and commercially responsible.
A sourcing agent works for you, not for the factory. You pay them a commission — typically a percentage of order value, sometimes a fixed monthly retainer — and they negotiate with factories on your behalf. The purchase contract is between you and the factory. The agent is a service provider.
A trading company buys from the factory and sells to you. Your contract is with the trader, your invoice comes from the trader, and your unit price includes their margin. You may never see the factory, and in many cases you are not told who it is.
A direct factory manufactures and sells to you. Your contract, invoice, tooling and quality liability all sit in one place. You talk to the people who own the machines.
The one question that identifies the model: if the toy is defective, who is legally obliged to make it right — the party you paid, or someone upstream of them? Everything else follows from that answer.
Importers compare the unit price on the quote. The real comparison is the full cost stack, because each model hides its margin in a different line.
| Cost line | Sourcing agent | Trading company | Direct factory |
|---|---|---|---|
| Where the intermediary margin sits | Visible commission line, usually negotiable and disclosed | Baked into unit price, usually undisclosed | None — factory margin is inside the unit price you already see |
| Typical structure | Commission on FOB value plus optional service fees | Markup on factory cost, sometimes 8–20% depending on volume and whether they carry stock | Straight factory quote, tiered by quantity |
| Tooling and moulds | Quoted separately by the factory; you can own it if you insist in writing | Often quoted by the trader and frequently held in their name | Quoted directly; ownership is a contract term, not a favour |
| Quality control | Third-party inspection, arranged and paid by you | Their in-house QC, quality of which varies enormously | Their QC plus your inspection; the factory has the machinery to rework |
| Consolidation of multiple SKUs | Possible but you manage the logistics | Their core strength — several factories, one container, one invoice | Only what that factory makes |
| Compliance paperwork | You or your lab drives it; agent chases documents | Usually handled for you, sometimes competently | Factory provides test reports and declarations if they already hold them |
| Payment terms | You pay the factory directly — deposit and balance as negotiated | Trader controls the payment schedule, may extend terms at a price | Standard 30/70 or negotiated against production milestones |
| Minimum order | Whatever the factory will accept; agents can split across factories | Often lower than the factory's own MOQ because they aggregate | The factory's MOQ, which is the real constraint |
Read that table downward and a pattern emerges: the agent gives you transparency and no leverage, the trader gives you convenience and no visibility, the factory gives you leverage and no flexibility.
A good agent earns their commission on the first three months of a product's life, when you need to talk to eight factories in four weeks and none of them should know who you are.
They are strongest when you are still developing a product: comparing processes, benchmarking quotes across a category, understanding why one factory's price is 30% below the others. They also protect you from being treated as a one-off buyer — a factory that sees an agent's volume behaves differently than one talking to a first-time brand.
What they cannot do is give you capacity priority. When a factory is full in September, they prioritise the customer who has been ordering for six years, and an agent cannot change that.
A trader is the right answer when your assortment is wide and your volume per SKU is thin. Twenty SKUs across plush, wooden, plastic and packaging, none of them hitting a factory MOQ, is exactly the shape of order a trader exists to fill.
They also absorb administrative load: one invoice, one payment, one set of export documents, one problem to call. For a retailer buying a seasonal mix rather than developing a product, that is a real service and worth paying a markup for.
The trade-off is that you never learn the true cost of your own product. That is fine while you are buying a catalogue. It becomes expensive the moment you want to invest in tooling, differentiate the design, or negotiate on a real cost basis.
At volume, the factory wins on price because there is no third party in the chain. More importantly, it wins on engineering: the people who can change a wall thickness, adjust a mould gate, or tell you that your 1.2 mm ABS shell will sink on the visible face are the ones standing next to the machine.
This is also the only model where tooling ownership is straightforward, where you can walk the line during production, and where a rework decision can be made the same day instead of after three emails and a phone call between parties who have never met.
What a factory cannot do is be everything. One factory makes one thing well. If your range crosses categories, a direct-only strategy means managing several suppliers with no one coordinating them.
The agent failure: the agent has no contractual liability for quality. If 4,000 units arrive with a paint defect, your recourse is against a factory in another country that has never spoken to you. A good agent will fight for you. The contract will not.
The second agent failure is quieter: split loyalty. An agent paid on commission has an incentive to place the order, not to place it well. If they also take a rebate from the factory — which happens — your interests and theirs have quietly diverged, and you will never see it on an invoice.
The trader failure: opacity compounds. You do not know the factory, so you cannot audit it. You do not know the cost, so you cannot negotiate. You do not own the tooling, so you cannot leave. Each of those is survivable alone; together they mean your product is not really yours.
The trader's other failure mode is spec drift. Because the trader is incentivised to protect margin, a substitution — a slightly thinner fabric, a cheaper battery holder, a different pigment — is easy to make and hard for you to detect without a retained golden sample and a real inspection.
The factory failure: narrow capability, and a MOQ that is a hard wall rather than a negotiating position. A factory that runs one moulding line for one product family will quote you beautifully on that family and badly — or not at all — on anything else.
Factories also tend to be weak communicators in exactly the phase where communication matters most. A spec question that an agent would escalate in an hour can sit unresolved for four days because the salesperson is also the export document clerk.
Here is the exercise I ask every importer to do before choosing a model. Take your target order value — say USD 40,000 FOB — and price it three ways, including the costs that only appear after the quote.
| Line item | Via agent | Via trader | Via factory |
|---|---|---|---|
| Product cost at factory | 40,000 | 33,000 (estimated, undisclosed to you) | 34,500 |
| Intermediary margin | 2,000–4,000 commission | 7,000 embedded markup | 0 |
| Your management time | Moderate — you still contract and pay the factory | Low | High — you own the relationship |
| Inspection | You pay, roughly 300–600 per man-day | Included, quality unknown | You pay, same rate |
| Tooling ownership risk | Manageable if written into the PO | Highest — often held by the trader | Lowest — you contract the tool directly |
| Realistic all-in delta | Baseline | +10% to +20% on unit cost | −5% to −15% on unit cost |
Those figures are illustrative, not a market survey — the point is the shape, not the decimals. On a 40,000 dollar order, the spread between the cheapest and most expensive model is often smaller than the cost of one bad production run. Which means the decision should be made on risk and capability, not on price alone.
If you have not yet built your landed-cost model, do that first — the numbers above only mean something once duty, freight and fees sit on top. Our landed cost guide walks through the full stack.
| Your situation | Best starting model | Why | Move to next when |
|---|---|---|---|
| First order, under USD 30k, one SKU | Trading company | You need a small mixed order, someone else's MOQ flexibility, and a single throat to choke | You reorder the same SKU twice and know your real sell-through |
| Developing a new product, no factory contacts | Sourcing agent | You need quote benchmarks across several factories without revealing your brand | You have chosen a process and a factory, and can contract directly |
| Brand with a defined hero SKU, 5k–20k units | Direct factory | Unit cost, engineering access, tooling ownership, capacity priority | You add categories the factory cannot make — keep them, add a second |
| Retailer with 40+ SKUs across categories | Trader plus one direct factory for the top sellers | Consolidation for the tail, margin and control for the volume drivers | When the tail accounts for under 15% of revenue |
| Investor-backed brand planning tooling | Direct factory, with an agent retained for benchmarks | You are buying an asset; own the mould and the contract | Never go back — but keep benchmarking |
| Amazon private label, 500–1,000 units per SKU | Trading company, then factory on the winner | Low volume, high SKU count, fast iteration | One SKU passes 3,000 units per run |
Notice that almost every row ends with "move to next". These models are stages, not religions. The importers who get hurt are the ones who pick a model for ideological reasons — "we only work direct" — and then force every product through it.
The vetting work is largely identical. The model changes who you are vetting, not what you check.
A practical test that costs nothing: send all three types of partner the same technical question — something specific, like the wall thickness at a snap-fit joint or the fabric weight tolerance on a plush panel. The speed, precision and willingness to say "I need to check" tells you more about the working relationship than any brochure.
After enough cycles, the structure that tends to survive looks like this:
Direct contracts with two or three factories for the core range — the products that carry your revenue, where you own the tooling and you have a named contact who answers within a day.
One trading partner for the tail — the accessories, the gift-with-purchase items, the seasonal novelties that will never justify their own factory relationship. You pay a markup and you should, because the alternative is managing nine suppliers for 6% of revenue.
No permanent agent, but a retained consultant or a periodic benchmarking exercise — someone who can get you three comparable quotes when a price increase lands, so you negotiate from data rather than from loyalty.
This structure costs slightly more than a pure direct model and delivers far more than a pure trading model. It also has a useful property: no single partner can hold your business hostage.
Choose the model that matches the order you are actually placing this quarter, not the business you intend to be in three years. Then write the exit into the contract before you need it: tooling ownership, spec sheets, golden samples, and the right to audit.
The difference between an agent, a trader and a factory is not really price. It is who is standing next to the machine when something goes wrong — and whether you have the paperwork to make them fix it.
Tell us your product, target volume and market. We will tell you honestly whether you should be talking to us as a factory, as a sourcing partner, or not at all — and what the first order should look like.
Talk to a Sourcing Specialist →