A founder in Seattle launched a jewelry brand with $12,000 of startup capital. She picked the cheapest Alibaba quote, paid a 50% deposit on a 200-piece order, and waited. The samples arrived six weeks late and did not match the reference. When she asked for a refund, the supplier went quiet. She relaunched the second time by running eight checks before she paid anyone. That second order shipped on time, and the brand is still going. Choosing a jewelry manufacturer for a small business is a skill, not a negotiation. Here are the eight checks we ask every first-time buyer to run on us, and on whoever else they are talking to.

The eight checks that matter

  1. Sample before PO. Any factory that will not send a sample before a bulk PO is a factory you should not bulk-order from. A sample costs $40-$150 and tells you more than 20 messages ever will.
  2. Outgoing defect rate in writing. Ask for the last 12 months' defect rate on the metal you plan to buy. A credible number is under 0.5% on silver, under 0.3% on gold. A factory that hedges ("very low") is hiding a number.
  3. Pull-test and plating-thickness spec. Ask what pull-test force they use on prong settings and what rhodium thickness they plate to. If the answer is "we check by eye," walk.
  4. Export documentation track record. Ask for the last three DDP shipments to your country. A factory that has never shipped to the US or EU will teach you customs paperwork on your dime.
  5. Tooling ownership clause. Confirm in writing who owns the CAD file and the rubber mold after final payment. "We keep it" is a dealbreaker for a growing brand.
  6. RMA window and process. Ask how defective pieces are handled: replacement on next PO, credit, or refund. Get the number of days in writing. We accept 30 days from carton receipt.
  7. Communication cadence. Email, WeChat, WhatsApp? A factory that takes 48 hours to reply on a sample question will take 48 hours to reply when a prong fails on 200 pieces.
  8. Referenceable customers. Ask for two references that are not on their marketing page. A factory that will not introduce you to an existing buyer has something to hide.

The signs of a factory that will stall

Three patterns show up again and again on failed first orders. None are illegal. All are predictable.

  • Quotes that look 30% below everyone else. The number is not wrong. It is missing lines: tooling, plating, certificates, packaging. The landed cost catches up later.
  • No spec sheet process. They ask for your design and start modeling the same day. A factory that does not write a spec sheet with you is a factory that will not defend a sample that goes wrong.
  • Upfront deposit above 50%. The industry norm is 30-50% deposit, balance against shipping photos. A factory that demands 70% up front is either cash-poor or worried you will walk.

What a small buyer actually needs on day one

If you are spending under $20,000 on a first order, you do not need the biggest factory in the city. You need the factory that will treat a 50-piece order as a real order. That means:

  • A per-order mix MOQ, not a per-style MOQ that forces you to buy 200 pieces of one SKU.
  • A sample process that takes under 10 working days.
  • Willingness to plate to 0.5µm rhodium even on a small run.
  • A real email thread, not a chat window that disappears.

Decision guidance by startup stage

Pre-launch (under $5k budget): Do not cut new CAD. Buy existing catalog pieces, private-label them with a logo and upgraded plating, and sell through a small Shopify run. You learn the customer, not the factory.

Launching ($5k-$25k): Run 2-3 hero SKUs at 30-60 pieces each. Use a factory that opens at 30 pieces per style and that sends a sample before bulk. Budget 10% of your order for rework and shipping surprises.

Proven ($25k+ per season): Move to a rolling forecast. You have sell-through data; use it to lock casting trees in advance and drop unit cost by 12-18%. This is the inflection point where a small brand becomes a wholesale operation.

Frequently Asked Questions

How much should a first wholesale order cost?

A realistic first order for a small brand is $3,000-$10,000, covering 2-3 hero SKUs at 30-60 pieces each. Going above $15,000 before you have sell-through data is a common startup mistake.

Should I pay 50% deposit before bulk production?

The industry norm is 30-50% deposit, balance against shipping photos. Avoid any supplier that demands 70%+ up front or refuses to send shipping photos before balance payment.

How long should I expect a first order to take?

From signed PO to carton: 35-50 working days for a first-time custom style, 20-30 days on existing tooling. Add 5-7 days for DDP air freight to the US or EU.

Can I switch factories after a bad first order?

Yes, if you own the CAD file and the spec sheet. This is why tooling ownership is check #5. If you do not own the file, you restart from zero at the new factory.

Do I need a lawyer to review a manufacturing agreement?

For orders under $25,000, a short written agreement covering MOQ, price lock, tooling ownership, and RMA terms is usually enough. Above that, a lawyer familiar with China manufacturing agreements is worth the fee.

Key takeaways

  • Sample before PO. Numbers before deposit. Numbers before trust.
  • Ask for the outgoing defect rate, pull-test spec, and plating thickness in writing. Hedging answers are data.
  • Lock tooling ownership before you pay. It is the clause that lets you leave if the relationship fails.
  • Spend under $10k on a first order. Prove the customer before you scale the factory.

Run the eight checks on us; we will send a sample, our last 12 months' defect rate, and two buyer references within 48 hours. See the production steps, browse the moissanite catalog, read the buying guides, or review our QC standards.

For HS classification of jewellery in customs and export documentation, the WCO Harmonized System nomenclature is the authoritative nomenclature.

Authoritative References

For buyers who need to verify the technical and regulatory basis of this guide, the following independent institutions publish authoritative reference data: