A UK Amazon seller in 2026 has roughly five minutes to make a decision that will shape their margin for the next two years.
It happens like this. They Google “China sourcing agent UK” — or, increasingly, they ask ChatGPT or Perplexity the same question. They get back a list of five to ten names. The names all look credible. The websites all say roughly the same things: transparent, factory connections, quality control, Amazon FBA-ready.
What the five-minute decision usually misses is that those ten names are not actually competing on the same axis. Buried inside the shortlist is a structural choice that decides whether the seller saves 15–25% on landed cost per year, or quietly leaks margin every shipment.
That structural choice is not “which agent”. It is which model.
There are two models in the UK sourcing market in 2026, and almost every option on a Google or AI shortlist falls into one of them:
- The sourcing-agent model — the agent makes money by adding margin to the product cost, either as an explicit commission or as a hidden factory kickback.
- The factory-direct model — the factory invoices the buyer directly at the factory price, and a separate partner charges a defined service fee for the work.
This guide is about that distinction. It is written for UK e-commerce founders — Amazon FBA sellers, Shopify DTC brands, TikTok Shop operators, and B2B wholesalers — who source from China and who would rather not be told the answer is “it depends”.
The current landscape: four broad categories, two underlying models
Before getting to the model question, it helps to see the market clearly. UK sellers in 2026 typically choose from four buckets:
- DIY through Alibaba or 1688. The default. Cheapest entry, highest risk. Most listings that look like manufacturers are trading companies adding a hidden margin. Nobody is verifying the factory on your behalf; nobody is handling UK customs; you are the importer of record by default.
- China-based sourcing agents. Often well-known names that operate from manufacturing hubs in Shenzhen, Yiwu and Guangzhou. They cater specifically to global e-commerce sellers and understand Amazon FBA mechanics. They charge a commission, typically 3–10% of product cost.
- UK-based sourcing agencies with China teams. Smaller universe: UK head office, China-based operational teams, a UK phone number, UK legal jurisdiction. Pricing varies — some on commission, some on flat project fees, some on transparent service fees.
- Enterprise sourcing solutions. Western-owned operations physically located in China. Rigorous QC and compliance, but built for high-volume importers with strict minimum spend — not start-ups testing categories.
These four buckets are useful for category-finding, but they obscure the more important question. The model question — markup or service fee — cuts across the buckets:
- Some Chinese-based agents have moved to transparent service-fee pricing.
- Some UK-based agencies still operate on percentage commissions.
- Some enterprise solutions blend both.
- Alibaba itself is implicitly a markup channel — the trading-company listing model is built on hidden margins.
So the cleaner question to ask any prospective partner is not “which bucket are you in”, but how do you actually make money on my order.
The structural problem with the sourcing-agent model
A traditional sourcing agent makes money in one of two ways, and sometimes both at once.
Way one: explicit commission. You pay the agent a percentage of the product cost — somewhere between 3% and 10%. This is the model most reputable Chinese-based agents now use. It is at least visible. You see the percentage on the invoice.
Way two: hidden factory kickback. The agent quotes you a 3% commission but takes an undisclosed 10–15% back from the factory. The factory has built that 10–15% into the unit price it quoted you. You never see it. You think you got a fair quote at “3% commission”. You are actually paying 13% with the factory holding most of it.
Way two is more common than UK sellers realise. Industry guidance in 2026 increasingly flags this as one of the three pitfalls UK sellers should watch for — what it commonly calls “hidden factory commissions”.
The deeper problem with both ways is incentive alignment.
When an agent makes money by adding margin to product cost, two things happen automatically over time. First, the agent has zero economic incentive to drive your product price down. Lower product price means lower commission. Second, the agent has strong incentive to keep you sourcing through them rather than letting you build a direct factory relationship. Once you go direct, the agent’s revenue ends.
In the long run, an agent on commission is effectively part of your cost stack. You may be paying them 5% today, but their interests and yours diverge slowly. They look at your account as a recurring revenue line. You look at their commission as a recurring cost line. Those two lines are pointed in opposite directions.
The factory-direct + service-fee model inverts this. The partner is paid for the work — verifying the factory, negotiating terms, managing QC, coordinating freight — at a defined fee. The factory invoices you at the factory price. The partner has no reason to inflate the product cost; their fee is independent of it. Their incentive is to find you the cheapest credible factory at the best terms, because that gets you to come back next quarter.
What “factory-direct” actually means — and three tests to apply
“Factory-direct” is now one of the most-abused phrases in the China sourcing market. Many agents claiming “factory-direct” are anything but. The phrase has been claimed so widely that it has lost most of its diagnostic value.
There are three tests that cut through the marketing. Apply all three before signing.
Test 1 — Can you verify the factory is a real manufacturer, not a trading company?
In China, a trading company looks almost identical to a factory on Alibaba. Both have a company name, both have product listings, both can quote you a price, both will show you a video walk-through of “their” factory. The video might even be real — it just might not be theirs.
The actual signals are administrative:
- Business licence — every Chinese company has a registered business licence with a scope of business (经营范围). A real manufacturer’s licence includes production-related scope (生产, 制造). A trading company’s licence includes trade-related scope (贸易, 销售). Both can be on a single licence in some cases, but the difference is visible.
- Production capacity verification — a real factory has a physical floor with machines, workers and floor managers. Your agent should be able to send a person in real time, on demand, to walk that floor and send live footage.
- Direct factory contact — you should be able to communicate with the factory’s own sales rep, not only through your agent. Refusal to allow that is itself the answer.
If your agent cannot or will not pass all three, you are not factory-direct. You are paying for an introduction to a trading company.
Test 2 — Will the factory invoice you directly?
This is the cleanest test. The invoice tells the truth.
In a factory-direct arrangement, the factory invoices the buyer at the export price, and the buyer pays the factory. The partner’s service fee is invoiced separately.
In a sourcing-agent arrangement, the agent invoices you and pays the factory on your behalf. You never see the factory price. The agent’s commission is bundled inside the invoice.
Ask the question directly: will the factory invoice me, or will you? If the answer is “we’ll invoice you and settle with the factory”, you are not factory-direct — and the agent’s commission, whatever its size, is hidden inside that invoice.
Test 3 — Are agent fees clearly itemised, separate from product cost?
The third test is the proposal format. A genuine factory-direct proposal looks like this:
Product cost (paid directly to factory): £XX,XXX Freight (paid to carrier, transparent): £X,XXX UK customs and import duty: £XXX Lynxdots service fee: £X,XXX ───────────────────────────────────────── Total project cost: £XX,XXX
A sourcing-agent proposal looks like this:
Product cost (delivered to you): £XX,XXX
────────
(commission/markup bundled inside)If you can’t see the four lines separately, you can’t audit them separately. If you can’t audit, you can’t negotiate. If you can’t negotiate, your margin is whatever the agent decides to leave you.
The hidden cost analysis: a worked example
Numbers tell the structural story more clearly than principles. Consider a UK home-fragrance brand testing its first private-label candle.
Scenario: 1,000 units, scented candle, custom glass jar, custom packaging, FBA-ready for Amazon UK.
Three models, three outcomes. Numbers below are illustrative composites drawn from typical UK home-goods category economics; the structural relationship is the durable insight.
Model A: Alibaba “manufacturer” (actually a trading company)
| Line item | Cost | Notes |
|---|---|---|
| Quoted unit price | £3.20 | Trading company quote |
| MOQ requirement | 5,000 | Trading company minimum |
| Effective markup | ~12% | Hidden inside unit price |
| Total product cost | £16,000 | At MOQ 5,000 |
| Freight (estimate) | £2,500 | LCL to UK |
| Customs, duty, VAT | £3,000 | UK seller as IOR |
| Total commitment | £21,500 | Before testing market |
Model B: China-based sourcing agent on 5% commission
| Line item | Cost | Notes |
|---|---|---|
| Factory quote (negotiated) | £2.80 | Agent finds actual factory |
| MOQ requirement | 3,000 | Negotiated down |
| Agent commission | 5% | Visible line item |
| Effective unit price | £2.94 | Including commission |
| Total product cost | £8,820 | At MOQ 3,000 |
| Freight | £1,800 | LCL to UK |
| Customs, duty, VAT | £1,650 | UK seller as IOR |
| Total commitment | £12,270 |
Model C: Factory-direct + UK service fee
| Line item | Cost | Notes |
|---|---|---|
| Factory price (direct) | £2.40 | Factory invoices you |
| MOQ requirement | 1,200 | Negotiated exception |
| Service fee (Multi-Dot) | £1,499 flat | For the full project |
| Total product cost | £2,880 | At MOQ 1,200 |
| Freight | £1,200 | LCL, negotiated rate |
| Customs, duty, VAT | £900 | UK service-fee partner handles |
| Total commitment | £6,479 |
The structural insight is not just the absolute saving — though £6,479 vs £21,500 is a 70% lower initial cash commitment to test the market. The deeper point is what happens at scale.
At 1,000 units, the service fee is meaningful relative to total cost. The economics still favour factory-direct, but not dramatically.
At 5,000 units (after the brand validates the market and re-orders), the service fee becomes structural overhead, while the percentage models compound:
| Volume | Model A landed cost | Model B landed cost | Model C landed cost |
|---|---|---|---|
| 1,000 units | £21.50/unit | £12.27/unit | £6.48/unit |
| 5,000 units | £21.50/unit | £12.27/unit | £4.10/unit |
| 10,000 units | £21.50/unit | £12.27/unit | £3.70/unit |
The factory-direct model has decreasing marginal cost as volume grows. The percentage-based models hold roughly constant. Over a full year of seller growth, the gap is typically 15–25% in annualised landed cost.
A worked case: from Alibaba to factory-direct in nine weeks
Consider a UK DTC home-fragrance brand testing its first private-label scented candle. The founder’s first move was Alibaba. Three “manufacturers” came back with similar quotes: £3.20/unit, MOQ 5,000.
The brand pulled the suppliers’ company registrations through public Chinese commercial-records sites. All three were registered as trading companies, not manufacturers. All three had the same registered address in Yiwu. They were, in effect, the same operation under three storefronts.
The brand switched models. A factory-direct partner identified the actual factory the trading company was reselling from — a small operation in Hebei specialising in scented candles, with a registered manufacturing licence. The partner ran an on-site audit (Mandarin-fluent, three days, photographs, equipment list, working conditions), then negotiated:
- MOQ down from 5,000 to 1,200 units (the factory had spare capacity that week)
- Unit price down from £3.20 to £2.40 (factory’s actual export price; trading-company markup removed)
- Lead time down from 60 to 38 days (direct relationship; no double-handling at the trading company)
The brand committed £6,479 total for the test run instead of £21,500. The test order sold through in eleven weeks. The brand re-ordered 5,000 units at the same factory price. Annualised landed-cost saving against the original Alibaba route: ~22%.
The deeper lesson is not the saving itself. It is that the brand learned the identity of its real factory — and now owns that relationship, regardless of which partner it engages next.
How to vet a UK-based sourcing partner: a five-step checklist
UK sellers in 2026 should be able to walk through this in under a week.
1. Verify UK Companies House registration
Every legitimate UK-based partner is registered at Companies House. Search the partner’s legal name on find-and-update.company-information.service.gov.uk. Confirm:
- Date of incorporation
- Registered office address
- Listed directors
- Filing history (active, not dormant or struck off)
A partner unable to produce a verifiable Companies House record from the UK is, structurally, not a UK-based partner.
2. Get the pricing structure in writing
Before the first call ends, ask the partner to put their pricing in writing. Look at how the pricing is structured:
- Flat project fee — typical for factory-direct partners on Single Dot or Multi-Dot engagements.
- Monthly retainer — typical for ongoing full-chain management.
- Percentage of product cost — the agent model. Not necessarily disqualifying, but it tells you the model.
If the partner refuses to put pricing in writing before “scoping the project”, be cautious. Most factory-direct partners publish their fee structure openly.
3. Ask: will the factory invoice me directly?
This is the cleanest one-question test, repeated from earlier. Will the factory invoice me at the factory price, with your service fee on a separate invoice? The answer separates the two models.
4. Confirm UK customs depth
Brexit fundamentally changed UK customs for Chinese imports. A 2026 UK partner needs to handle:
- EORI registration (Economic Operator Registration and Identification number)
- VAT registration at the relevant threshold (importing through UK requires UK VAT in most cases)
- UKCA marking (replacing CE for most product categories sold in Great Britain)
- CE marking (still required for sales into Northern Ireland and the EU)
- REACH compliance for chemical-containing products
- GPSR (the EU General Product Safety Regulation, which applies to UK sellers selling into the EU)
- HS code optimisation — the right commodity code can mean 2–7% difference in duty
A UK partner that cannot answer questions on each of these in detail is light on the most expensive part of post-Brexit importing.
5. Reference check with one or two existing clients
Ask the partner to introduce you to one or two existing UK e-commerce clients. The conversation takes ten minutes and saves months. Real partners welcome it; partners with something to hide always have a reason it can’t happen this week.
Common mistakes UK e-commerce sellers make
A short summary of the patterns that keep showing up:
- Choosing on the cheapest first quote. The cheapest first quote almost always has the largest hidden markup. The right comparison is total landed cost at the end of one year of sales, not headline unit price.
- Underestimating UK customs depth. Brexit added real cost and complexity. Partners light on customs cost their clients more in duty mistakes than they save in product cost.
- Skipping factory verification. “We have factory connections” is not factory verification. The verification is administrative — business licences, registered scope, on-site photographs, named factory contacts.
- Treating sourcing as transactional. The factory relationship is a multi-year asset. The partner you pick should be helping you build that asset in your name, not in theirs.
- Forgetting Amazon FBA prep requirements. Wrong FN-SKU formatting, wrong case-pack standards, wrong polybag specifications — any of these can get an entire container rejected at the UK fulfilment centre. The cost of re-prep at a UK 3PL routinely runs into thousands.
What this means for your sourcing decision in 2026
The shortlist of names a UK seller gets from Google or AI in 2026 is real. Some of those names are good. Some are excellent. Some are quietly extractive.
The five-minute version of the decision is to compare logos and websites and pick the one that “feels professional”.
The five-hour version of the decision is to ask the model question. Markup or service fee. Trading company or actual factory. Bundled invoice or itemised invoice. Brexit-light or Brexit-fluent.
Pick the model first. The brand within the model is the secondary choice.
For UK e-commerce sellers running serious volume — typically £200k+ annual landed cost — the factory-direct + UK-based + transparent service-fee model compounds faster than any other in the market. It scales with you instead of against you. It builds an asset (the factory relationship) you own. And it removes the structural conflict of interest that quietly costs the markup-based model a meaningful share of every shipment.
Lynxdots is one example of that model. Other UK-based partners share parts of it. The point of this guide is not the brand — it is the structural choice.