Most hardware crowdfunding campaigns don't fail at the fundraising stage. They fail in the four to eight months that follow — specifically in the transition from "working prototype" to "factory-ready design." A sourcing agent with an engineering background changes this equation because they catch design-for-manufacture (DFM) issues before tooling is cut, before the money is committed.

This guide covers what actually happens between a Kickstarter closing and a shipping container leaving Shenzhen — the stages, the costs, what can go wrong at each, and where an experienced hand makes a measurable difference.

The three stages most hardware startups don't plan for

The hardware startup mental model is usually: build prototype, run campaign, ship product. The factory side is treated as an implementation detail.

In practice there are three distinct engineering stages between a working prototype and a shipping product, and each one involves a different factory relationship and different costs.

Stage 1: Prototype → Engineering Validation (EV)

This is the stage founders are familiar with. The goal is to prove the product works. EVs are usually hand-built or contract-assembled in small quantities (1–10 units). The factory relationship, if there is one, is loose — often a prototype house rather than a production factory. Cost at this stage is dominated by NRE (non-recurring engineering) and component procurement at retail or near-retail prices.

Stage 2: Engineering Validation → Design Validation (DV)

This is where most hardware startups get surprised. DV means proving the design can be manufactured reliably and repeatedly. The factory takes the EV design and builds 20–100 units using production-like processes. Defects in this stage are design defects — PCB layouts that cause soldering issues, housing tolerances that make assembly too slow, components that fail at production-line temperatures.

Fixing these problems requires design revisions, which may require new tooling. DV is iterative and expensive, and founders who didn't budget for it either delay or compromise on quality.

Stage 3: Design Validation → Production Validation (PV)

PV is the bridge from "we can make it" to "we can make it at volume." A pilot run of 200–500 units is assembled, inspected, and used to set yields, cycle times, and QC pass rates. The data from PV is what the factory uses to price the mass production run and commit to a timeline.

Founders who budget for Stage 1 and get surprised by Stages 2 and 3 are the ones who miss their shipping estimates by four months and go back to backers asking for patience.

What a sourcing agent does at each milestone

Before your campaign launches

The work that happens before a campaign is often the most valuable and the least visible. A sourcing agent with manufacturing experience can:

• Survey the supplier landscape for your product category — who makes it, what the typical MOQ looks like, what certifications are achievable and at what cost
• Produce a preliminary Bill of Quantities (BOQ) cost estimate based on your design, before you've committed to any factory
• Flag DFM issues in your prototype design that would require expensive tooling revisions later
• Identify which certifications (FCC, CE, RoHS) you need and give you a realistic timeline — this is important because certification timelines frequently conflict with shipping promises made during crowdfunding campaigns

The cost of this work: typically $500–1,500 for a preliminary manufacturing review and supplier landscape report. The value: knowing before you campaign whether your $45 landed price target is achievable, and whether your "ships in June" promise is realistic.

Campaign funded: RFQ and factory selection

Once you have funds committed, the first formal step is getting quotes from qualified factories. A well-run RFQ process involves:

• Sending a complete RFQ to 5–8 pre-qualified factories (not 50 — factories that see a mass RFQ give it minimum effort)
• Completing a factory audit for the top 2–3 respondents — you want to visit before you place any money
• Signing NDAs with the shortlisted factories before sharing your full design files

The NDA step is one that founders consistently skip because they're eager to move fast. Don't. Your design files contain your manufacturing IP. A factory that receives your Gerber files, BOM, and housing drawings without an NDA has no legal obligation not to copy them or share them with a competitor. The NDA doesn't guarantee IP safety — nothing does in China manufacturing — but it creates a legal paper trail and signals that you're a serious buyer who knows how the process works.

Factory audits for electronics startups focus on specific capabilities: SMT line specifications (which nozzle sizes, what pick-and-place machines), in-house testing capability (aging chambers, functional test fixtures, RF test), and quality certifications relevant to your product. Use the factory audit checklist to know what to look for.

Engineering samples (ES): 5–20 units

After signing a manufacturing agreement with your chosen factory, the first formal deliverable is engineering samples — typically 5–20 units built by the factory using production-like processes. The timeline is usually 4–8 weeks from order to delivery.

Engineering samples are not the same as your prototype. The factory makes them using their tooling (if tooling exists at this stage), their component suppliers, and their assembly process. This is when you first see what your product actually looks like when made by someone else, at scale.

Evaluate ES units against a written evaluation checklist, not against your gut feeling. Areas to check:

• Mechanical fit and finish: housing tolerances, button feel, seam alignment
• Electrical function: all features working per spec
• RF performance: Bluetooth, WiFi, or cellular operating at spec across temperature range
• Battery runtime: within 10% of target
• Charging: full cycle with the correct charger

Document every deviation from spec in writing. This document becomes the basis for the DV revision discussion with the factory.

Pilot run (PR): 50–200 units

After ES revisions are incorporated and the factory signs off on a validated design, the pilot run produces 50–200 units using the production process — the same assembly line, the same component reels, the same test fixtures that will be used in mass production.

The pilot run is where manufacturing yields first become visible. If the factory's SMT line has a solder paste stencil that's slightly off spec, you'll see a consistent reflow defect pattern. If the injection molding tool has a sink mark issue, you'll see it in every unit. A 5% yield loss on a pilot run of 200 units is annoying. The same 5% yield loss in a mass production run of 10,000 units is a $50,000 problem, assuming a $100 unit cost.

For your first production relationship with a new factory, run a pilot run. The cost is higher per unit than mass production (because they're not at full line efficiency), but the data it generates is worth more than the savings from skipping it.

Mass production (MP): 500+ units

By the time you reach mass production, the factory should have:

• A validated design with approved component substitutions documented
• A test fixture calibrated to your spec
• An AQL sampling plan agreed in writing
• A delivery timeline with milestone dates

Quality inspection in mass production follows three stages:

Pre-production (PP) inspection: Before manufacturing starts, verify the factory has received the correct components. Check BT module part numbers, battery cell model, PCB revision, housing tooling revision. Specification drift starts here — factories sometimes substitute cheaper components without notifying the buyer. Catching it before a run starts is free; catching it after 5,000 units are assembled is not.

During-production (DUPRO) inspection: When 10–20% of the batch is complete, inspect the first finished units. Problems found at this stage affect only the units already assembled; the remaining 80–90% of the run can be corrected. This is the highest-value QC intervention in mass production.

Pre-shipment inspection (PSI): When 100% of production is complete and at least 80% is packed, perform an AQL sample check. For consumer electronics, AQL 2.5 is standard — on a 5,000-unit batch, this means checking 200 units with a pass/fail threshold of 10 defects.

DFM: the most expensive review to skip

Design for Manufacturability review is a systematic check of your design against the capabilities and constraints of the production process. For an electronics product, a DFM review typically covers:

• PCB trace spacing and via sizing: pick-and-place machines have minimum component size requirements; some component footprints look fine in CAD but cause soldering defects on a production reflow oven
• Component footprint accuracy: a 0402 resistor with a poorly-matched footprint will have inconsistent solder joints; multiply that by 50 components per board and 5,000 boards
• Housing draft angles: injection-molded housing parts need draft angles (typically 0.5°–3° depending on finish) to release from the mold; zero-draft surfaces cause mold sticking and part deformation
• Battery connector accessibility: if the battery connector is in a location that requires removing three other sub-assemblies to access, labor cost per unit goes up and assembly consistency goes down
• Test point accessibility: ICT (in-circuit test) fixtures need physical access to test points; a board where critical test points are under components or on the wrong layer will require either a custom fixture or manual testing

The cost of catching a DFM issue before tooling is cut: a DFM review typically runs $500–1,500 and takes 3–5 days.

The cost of catching it after: a mold revision in Dongguan typically costs $3,000–15,000 depending on the revision complexity, and adds 4–6 weeks to the timeline. A PCB layout revision after the first production run requires a new Gerber order, new stencil, and scrapping the boards already assembled. On a 200-unit pilot run, that's $4,000–12,000 in wasted material depending on component cost.

Request a DFM review before any tooling is cut. For an OEM/ODM project, the factory should provide DFM feedback as part of the engagement — but don't assume they will without asking, and don't assume their feedback is complete. For the PCB side specifically, our PCB assembly sourcing page covers the factory evaluation criteria — SMT line specifications, IPC class requirements, and inspection coverage — that determine whether a factory can actually produce your board to spec.

The tooling decision

Injection molding tooling is the largest single capital cost in most consumer hardware products, and it's the one founders are least prepared for.

Prototype tooling (soft tooling): Aluminum molds, faster to machine, lower cost ($2,000–8,000 for a simple 2-cavity mold), shorter lifespan (typically 5,000–20,000 shots). Used for early DV and pilot runs. The advantage is that modifications are cheaper — you can still add material to an aluminum mold via welding, whereas a steel mold modification is significantly harder and more expensive.

Production tooling (hard tooling): Hardened steel (typically P20 or H13), slower to machine, higher cost ($5,000–50,000 depending on part complexity and number of cavities), rated for 500,000–1,000,000+ shots. Required for mass production runs above 10,000–20,000 units.

Who owns the tooling: You do, or should. Tooling paid for by the buyer belongs to the buyer — this should be stated explicitly in the manufacturing agreement. A factory that holds your tooling without a clear ownership agreement has significant leverage over your supply chain. If you switch factories, tooling they "own" doesn't move with you.

Where tooling is stored: Typically at the factory that uses it. Get explicit written agreement on storage, maintenance responsibility, and the right to retrieve it on 30 days' notice.

For PCB tooling: the Gerber files and pick-and-place program are your production IP. Keep copies. Panelization — arranging multiple PCBs on a single production panel — is something the factory handles but you should understand: it affects SMT line efficiency and cost per board.

Certifications: timing matters more than you think

For most consumer electronics products sold in the US and EU, you need:

• FCC (Part 15): Required for US market, covers intentional and unintentional RF emitters. Certification process: 6–12 weeks, cost $5,000–15,000 for a typical Bluetooth device. Voluntary testing labs (A2LA accredited) in China or US.
• CE marking: Required for EU market, covers multiple directives (RED for radio, LVD for low-voltage, ROHS). Timeline: 4–12 weeks depending on directives involved. Can often run in parallel with FCC.
• RoHS: EU restriction of hazardous substances — requires declaration from component suppliers, verified against your BOM. Usually bundled with CE process.
• UN 38.3: Required for lithium battery transport, regardless of market. Without it, your finished goods can't be air-shipped.

The crowdfunding timing trap: Kickstarter and Indiegogo campaigns routinely promise delivery timelines that don't include certification time. A campaign that closes in January and promises August delivery is implicitly assuming that FCC and CE certification are already underway or complete — which they almost never are for a first-time product.

The realistic buffer: start FCC/CE certification in parallel with your pilot run, at minimum. That means certification is running while you're still validating the manufacturing process. If the certification test fails (common — first submissions fail roughly 40–50% of the time for products with RF components), you have time to address the findings before mass production is complete.

The module approach: Using a pre-certified BLE or WiFi module (Nordic Semiconductor, Espressif, u-blox, Murata) significantly reduces your certification scope. Instead of certifying the entire radio system, you certify only the product-level integration — typically shorter timeline and lower cost. For a first product, using a certified module is almost always the right call.

Common milestone mistakes

Locking in a factory before IP agreements are signed. Some factories will not work without an NDA; some will. Regardless of what they say, sign one before sharing Gerbers, BOM, or housing drawings. The factory that copies your design two years later is often not the one that seemed obviously untrustworthy — it's the one you thought was your partner.

Starting tooling before DFM review. This happens when founders are impatient and the factory is eager to take the tooling payment. Tooling is the point of no return. Whatever DFM issues exist in the design at that point become very expensive to fix. Do the DFM review, get written sign-off, then cut tooling.

Paying 100% deposit before production is validated. Standard payment structure is 30% deposit on order confirmation, 70% before shipment — after passing PSI. Factories that insist on 50% or more upfront before you've seen any product are a risk signal. A founder who pays 100% upfront has no leverage if quality problems emerge during production.

Skipping the pilot run to save time. The argument is usually "we need to ship in four months and the pilot run takes six weeks." The reality: a pilot run that reveals a 12% yield loss will cost you far more than six weeks if you discover the same problem during mass production. The factories that are most confident you don't need a pilot run are the ones you should be most concerned about.

Not identifying a backup supplier before mass production starts. Single-source dependency on a factory is manageable when everything is going well. It becomes a serious business problem when the factory has a fire, a labor dispute, or simply decides your order isn't a priority. Maintain a relationship with at least one qualified backup factory after your first production run.

Realistic timeline and cost

The following assumes a consumer electronics product with custom injection-molded housing, PCB with BLE module, and battery — targeting US and EU markets at 5,000-unit initial production run.

The total elapsed time from "campaign funded" to "goods at US warehouse" is realistically 9–14 months for a first-time product. Campaigns that promise 6-month delivery from close are banking on everything going right on the first try — which almost never happens.

The total capital required before the first dollar of revenue (excluding campaign funds for the production order itself): $20,000–80,000, depending on product complexity, tooling requirements, and certification scope. This is often a shock to founders who assumed the campaign proceeds would cover everything.

The sourcing agent's role: what it changes

The process above is manageable. It's also the process where most things go wrong, and where having experienced representation on the ground in China changes the outcome.

Specific ways a manufacturing-experienced sourcing agent changes the equation:

• Pre-campaign manufacturing review: identifies whether your target cost and timeline are achievable before you promise them publicly
• DFM review: catches issues before tooling is cut, not after
• Factory audit: verifies capability and quality history directly, not from a sales presentation
• Production management: someone watching the line who can read a production spec, not just translate "it looks fine"
• Inspection management: coordinates PP, DUPRO, and PSI with a qualified inspector, reviews reports before you see them

For the EU startup Bluetooth speaker case, this end-to-end support covered five months from factory selection to delivery of 5,000 units with a 0.4% defect rate. For the US smartwatch startup, it meant navigating concurrent FCC and CE certification while the factory ran the pilot run — shaving six weeks off the timeline by running processes in parallel rather than sequentially.

Both of those outcomes are available in the consumer electronics category at volumes in the 2,000–10,000 unit range. They're not exceptional — they're what competent manufacturing management produces.

What to do next

If you're at the crowdfunding stage and want to understand what production will actually cost and take, get in touch. We scope this out before you commit to a factory — a preliminary manufacturing review and supplier landscape report typically takes 5–7 business days and gives you numbers you can put into a real production budget.

If you're past the campaign stage and already working with a factory, the questions to ask before you cut tooling: Has the design been through a DFM review? Is there a signed NDA? Have you audited the factory, or taken their word on capability? The answers to those three questions predict most of what happens next.

For more on the factory evaluation side of this process, the factory audit checklist covers the 47 specific points worth verifying before you commit to a supplier.