What Must Exist Before You Send a Single RFQ

The Minimum Viable Technical Package

Before any outreach to find a manufacturer, four documents need to exist in a form a factory engineer can act on without guessing.

The first is a drawing or 3D file with tolerances. A sketch or a concept render does not count. The factory needs to know what they are actually making — STEP files are the standard for mechanical parts, Gerber files for PCBs. Without these, the supplier is quoting their imagination.

The second is a bill of materials with part specifications. Not a list of component categories, but a list of actual components with key specifications flagged. If there is a proprietary sensor or a specific connector form factor that is non-negotiable, that needs to be in the document. Factories substitute components by default when specs are absent. This is not a quality failure; it is how they manage margin. The substitution changes the product.

The third is a quantity band, not a single number. Founders often ask for pricing at their "target" volume without specifying what pilot quantities and scale quantities look like. This matters because many factories will not quote accurately for 500 units if they only want to work with buyers at 10,000 units. Showing your quantity range surfaces this mismatch before you invest three weeks in email chains (which, once you've been through it, you'll never want to repeat).

The fourth is a test standard or acceptance criteria. What does a passing unit look like? What failures are grounds for rejection? Without this, the factory defines quality themselves — and their definition is optimized for throughput, not for your customer's experience.

The practical implication here is not bureaucratic. It is strategic. The moment you send a complete technical package, you radically cut the number of factories willing to quote — and the ones who do are signaling actual capability rather than aspirational interest.

The Stage-Fit Question Nobody Asks

Even with a complete technical package, there is a second dimension that most founders skip: does this factory's business model fit your current production stage?

A factory optimized for mass production at 50,000 units per month has a cost structure that makes small runs actively unprofitable for them. They will quote you, take your order, deprioritize your production, and deliver late. Not malice. Incentive alignment. Their machine utilization model does not have room for you.

The question to ask is not "can you make this?" Almost every factory will say yes. Ask instead: "What is your typical order size, and where does our volume fit in your production queue?" That question produces honest answers and, more importantly, reveals whether the relationship is viable before the PO is signed.

Channel Selection Is a Signal, Not a Search

How Different Sourcing Channels Produce Different Supplier Types

When founders are told to "find a manufacturer," the instinct is to open Alibaba. Alibaba is a legitimate tool, but it is a search engine for suppliers who have paid to be discoverable by Western buyers. That is a selection effect. The factories on page one are optimized for inquiry volume, not for technical complexity — they have English-speaking sales teams, flexible MOQ claims, and fast response rates. These are marketing capabilities, not manufacturing capabilities.

The channel used to find a factory predetermines the type of factory found. Referrals from other hardware founders surface factories that have already survived someone's pilot run; that is a real data point. Trade shows like Canton Fair put a face to capability claims and allow for a physical audit of sample quality. Engineering networks in manufacturing clusters like Shenzhen produce introductions to factories that never bother with online marketing because they have been running at capacity with existing relationships for eleven months straight, year after year.

None of these channels is universally superior. Each suits a different buyer profile. A founder at the prototype stage needs a different factory than a startup preparing for Series A volume commitments. The sourcing channel needs to match the production stage.

Choosing Alibaba by default is a product decision disguised as a search decision.

Local Partners as Structural Intermediaries

There is a category of sourcing support that sits between "find a factory yourself" and "hire a full sourcing agent": local partners with production relationships in China who can filter the supplier set before you see it.

This model works when the partner's incentives are aligned with yours rather than with the factory's. A commission-based sourcing agent earns more by placing orders, not by finding the best fit. A partner paid for China-side production management has an incentive structure aligned with production quality and timeline reliability, because their reputation is staked on the outcome, not the transaction.

The distinction matters because the sourcing process is not complete at contract signing. Factory fit is a dynamic condition that degrades when production pressures shift, when your volumes grow faster than the factory expected, or when a larger customer displaces you in the production queue.