Where suppliers actually live

Start where the factories are, not where the ads are. Marketplaces (Alibaba, Global Sources, Made-in-China) are directories with a chat box — useful for finding factories in a category, useless for judging them without the vetting in Lesson 3. Search by product plus process (“grip socks silicone printing”), not by product alone, and you will surface specialists the generic listings bury.

Trade shows and industry directories are the slow, high-signal route: a factory that exhibits spends real money to be found by professionals, and association directories in the big producing regions (for socks, the Zhuji cluster in China is the world's center of gravity) list manufacturers, not storefronts. Referrals outrank everything: a non-competing brand owner who has shipped two clean orders from a factory is worth more than a hundred platform badges. Ask in founder communities who actually made their product — the question people dodge tells you as much as the one they answer.

Two traps to walk past. Trading companies posing as factories: they add a margin, slow every answer, and cannot control the production line they rent. They are not always a bad buy — small orders sometimes only make sense through them — but buy knowingly, priced accordingly. And the too-eager supplier: instant replies, any spec, any price, no questions. Real factories ask about your artwork, your sizes, your target quantity. A supplier with no questions has not read your pack, and will not read it later either.

Build a longlist of eight to twelve, because attrition is the plan: half will fail vetting, a few will quote wildly, and you want three to five serious quotes standing at the end. Sourcing feels slow exactly once — the first time. Your vetted list, with notes, is an asset every future product reuses.

Specialist vs generalist, domestic vs overseas

The single strongest predictor of a good outcome is embarrassingly simple: the factory already makes products like yours, for buyers like you, every week. A specialist owns the right machines, stocks the right materials (which lowers your effective MOQ), has solved your product's common defects before you meet them, and can show you production samples in your category on the first call. A generalist can make anything, the way a diner can cook anything on a twelve-page menu.

Ask the specialist question directly: “What share of your production is grip socks?” and “Show me three current styles you produce in my category.” Specialists answer with specifics and photos of lines you can verify. Generalists answer with the factory tour video and a catalogue of everything from umbrellas to underwear.

Domestic versus overseas is a trade, not a morality play. Overseas specialists (for most soft goods, Asia's clusters) bring lower unit costs, deep material supply chains, and higher MOQs, paid for with longer lead times, freight exposure, and communication overhead. Domestic makers bring speed, tiny minimums, easy visits, and painless reorders, at unit costs that often double or triple the overseas quote. Run both through the same landed-cost calculator from Module 1: domestic wins more often than founders expect on small, fast, high-margin products, and loses predictably on price-led ones.

The grown-up answer is frequently staged: develop and launch domestically or with a sample-run specialist, prove the product, then move volume production to the overseas specialist with a tech pack and a golden sample that survived contact with customers. Your documents travel; your education compounds.

Vet before you chat: the supplier scorecard

Vetting happens before the friendly chat, because chat is where judgment goes to be charmed. Score every longlisted supplier on the same five evidence points, from sources that are not the supplier's own listing copy.

One — category proof. Current products in your category, shown in production, not just in a catalogue PDF. Two — business reality. A verifiable business license, years operating, and a factory address that exists; platform verification services and a video walk-through of the floor with your product's machines running both count. Three — export record. They ship to your market routinely and can name the paperwork without blinking. Four — communication quality. Your RFQ (next lesson) is the test: complete answers, real questions, numbers that add up. Five — references and certifications where your category needs them — material safety certifications like OEKO-TEX for textiles, audit reports like BSCI or ISO 9001 — as current documents, not logo soup on a banner.

Score each 0–2, require a minimum total, and keep the sheet. The supplier who fails on evidence but wins on personality is the most expensive supplier you will ever meet, and the scorecard is how you decline him with a clear head. It also becomes your reorder insurance: when your chosen factory has a bad season, the runner-up on a scored sheet is a phone call, not a new research project.

Red flags that end vetting on the spot: prices that move between messages with no spec change, refusal to do a paid sample (every real factory charges for samples — free samples are a sales cost recovered somewhere you cannot see), pressure to pay full production money before a pre-production sample exists, and any request to move payment off agreed channels to a personal account.

• Category proof seen: current products like mine, in production
• Business license, years operating, and factory address verified
• Ships to my market routinely; export paperwork named without hesitation
• RFQ answered completely, with real questions about my spec
• Needed certifications produced as current documents
• No red flags: stable pricing, paid samples, no off-channel payment pressure

The RFQ: one message that gets real quotes

An RFQ — request for quotation — is the tech pack plus five questions, sent identically to every vetted supplier. Identical is the operative word: the moment each factory quotes a different version of your product, you are holding five prices for five products, and comparison is theater.

The message itself is short, because the pack does the talking: “We are developing [product] for [market]. Attached is our tech pack, revision [date]. Please quote: (1) unit price at 500 / 1,000 / 3,000 units, per the pack's BOM; (2) what the MOQ is per design and per size; (3) sample cost and sample lead time; (4) production lead time after PP sample approval; (5) your price basis (EXW or FOB, and which port). If any spec in the pack raises cost sharply, tell us which and what you would change.” That last line is a gift to good factories — it invites the cost engineering that specialists are best at, and their answers teach you where your spec is expensive.

Judge the replies on three axes. Completeness — every question answered, or an honest “we cannot do line X” (a factory that declines part of a spec is telling the truth at cost to itself; trust it more). Internal consistency — price breaks that make sense as quantity rises, lead times that match the process. Cost intelligence — the supplier who writes back “your terry cushion sole adds $0.40; a half-cushion gets you the same feel at $0.15” has just auditioned as a development partner, whatever their rank on price.

Put every quote into one table — supplier, unit price at your quantity, MOQ structure, sample cost/time, lead time, terms — all converted to landed cost with the Module 1 calculator. The cheapest unit price loses to a higher one with half the MOQ more often than founders expect, because cash parked in slow variants is a cost too. Choose on the table, not on the chat.