The campaign page is the easy part. The hard part starts the day the money clears and you discover your render has a 0.8mm wall where the factory needs 1.5mm — and that the battery you promised cannot legally fly.
I have watched this fail from both sides of the table: as the buyer with backers waiting, and as the person Chinese factories call when a foreign startup arrives with a funded campaign and no manufacturing file. The pattern is always the same. The campaign succeeds, the founder books a factory visit, and tooling gets quoted in three days because quoting is free. Six weeks later T1 samples arrive with a light leak, a battery door that will not latch, and an enclosure two millimetres wider than the PCB allows. The delivery date slips a quarter, then another.
None of that is dishonesty. A campaign is a marketing artefact; manufacturing runs on a different document set.
The short version: money raised is not a production plan. Between "funded" and "shipped" sit four gates — DFM review, tooling and T1 approval, certification, and pilot run. Each can cost you a quarter if skipped, and each is cheaper than the slip you cause by skipping it.
1. A campaign freezes a promise, not a design
When you launch, you commit to a price, a specification, a delivery window, and implicitly to regulatory obligations in every market you ship to. What you have at that moment is a working prototype, a CAD file and a render — not a manufacturable product.
The first thing to do after funding is not negotiate price. It is to hand the factory a complete data package and pay them to tell you what is wrong with it. A serious DFM review covers wall thickness and draft angles, parting lines and gate locations, achievable tolerances, snap-fits, thermal paths, battery placement, assembly sequence and test access. Skip it and you are not saving money — you are buying tooling twice.
2. Where the money leaks: promise vs production reality
These mismatches come up over and over. None is exotic, and all are cheaper to fix on paper than in steel.
| What the campaign promised | What production requires | The fix, and when |
|---|---|---|
| A slim, seamless enclosure | Minimum wall thickness and draft angle; a parting line somewhere; room for screws or clips | Redesign before tooling — adding 0.5mm later means a new mold |
| Anodised aluminium unibody | CNC cost per unit far above molded plastic; finishing adds steps and yield loss | Decide in DFM whether the look earns the unit cost, or use coated plastic with a metal insert |
| A custom PCB on a chosen chip | Component availability, reel minimums, and a second source for anything critical | Lock the BOM with the factory's purchasing team, not your engineer's wish list |
| Six weeks of battery life | Sleep current, radio duty cycle, and the cell you can actually source at volume | Measure on the pilot build, not the dev board |
| Ships worldwide | Different certification regimes per market, plus battery transport rules | Pick target markets before launch; certify the pilot build |
| Delivery in March | Tooling, T1/T2, testing, pilot run, mass production, ocean transit, customs, 3PL intake | Build the timeline backwards and add buffer, not optimism |
Every row above is a decision made before the mold is cut. After that, each one becomes a change order.
3. Tooling: the largest irreversible cheque you will write
Tooling breaks crowdfunding budgets because it is quoted as one number and paid as a deposit before you have seen a single part. An honest framing: a simple two-plate mold for a modest plastic part is commonly quoted in the low thousands of US dollars in China; a multi-cavity tool for a cosmetic part, or anything with sliders and lifters, runs well into five figures. The ranges are wide on purpose — size, cavity count, steel grade and finish drive it, and a firm number without a DFM review is a guess.
Three things matter more than price:
- Tool ownership. The PO should name the tool, state it is your property, and define release to you or a nominated third party on request. "The mold stays with us" is normal practice, not a conspiracy — but ownership and release must be written.
- T1 before balance. Tie most of the tooling cost to sample approval, not mold completion. A tool that produces rejected parts is not finished.
- Change orders in advance. Ask what a post-T1 modification costs — steel-safe, slide change, full re-cut. The answer predicts the relationship.
For the mechanics, see the injection mold guide.
4. Your MOQ is not the factory's MOQ
Factories quote a minimum for their own reasons: machine time, material minimums, setup labour. You should care about a different number — the quantity you can move without strangling your cash. Take net campaign proceeds after platform and payment fees, subtract tooling, certification, packaging, freight and duty, then divide by landed unit cost. If that is below the factory's MOQ, you have four options:
- Raise price or volume while the campaign is live — the only moment you have leverage over your own backers.
- Change the process — CNC, vacuum casting or 3D printing for run one, at a higher unit cost.
- Find a factory sized for you — the workshop that takes 500 units and grows with you, not the plant that treats your order as a distraction.
- Order the MOQ and sell the excess — legitimate only with a credible retail channel and a clear view of storage and aged-inventory cost.
Option four is where founders quietly go broke. Excess units are not inventory; they are a bet on a demand curve you have not tested.
Watch duty and freight on small runs. A 500-unit first run shipped by air can cost more per unit in freight than the factory charges to make the product. Model landed cost, not FOB cost, before committing to a volume.
5. Certify before you launch, not after you ship
Crowdfunding platforms are not customs. A product can be funded, made and delivered in some markets without anyone asking for a certificate — and then be unsellable through retail, Amazon or any serious distributor. Requirements depend on product and market, and they change. The shape of the problem does not:
- Radio and electrical products generally need market-specific approval before sale — FCC in the US; CE marking under the applicable directives in the EU. A supplier's certificate covers a model, a factory and a configuration; it does not automatically cover your variant.
- Battery-powered products face a second layer: lithium cells need UN38.3 test documentation before they can be offered for transport, and mode (air vs sea, state of charge) changes both paperwork and cost — see shipping lithium battery products from China.
- Children's products carry the heaviest testing and labelling burden. If a product could reasonably be used by a child, treat it as one until a lab says otherwise.
- Materials and coatings face restricted-substance rules that differ by market and get updated; declarations must match the actual resin and coating.
None of this is legal advice. Confirm the requirements against the official text of each market's regulator, then have your factory's compliance contact and testing lab confirm them in writing for your specific model. The sequence that works: decide markets, identify standards, test the pilot build, then print labels. A certificate attached to a hand-built prototype is not a certificate for the thing you ship.
6. Work backwards from the delivery date
Backers forgive a good product that is late. They do not forgive silence, or a product that arrives broken. Protect both with a timeline built backwards from the promise using realistic durations, not the factory's optimistic verbal estimate.
| Stage | What must be true at the end | Typical duration |
|---|---|---|
| Data package and DFM review | Manufacturable design signed off; open issues closed in writing | 1–3 weeks |
| Tooling build | Mold cut, first shots pulled, dimensional report issued | 3–6 weeks |
| T1 sample approval | Fit, finish and function accepted; change list agreed | 1–3 rounds, days each |
| Certification testing | Reports in hand for the pilot build and target markets | 2–6 weeks (lab-dependent) |
| Pilot run | Line proven, yield measured, packaging validated | 1–2 weeks |
| Mass production | Full quantity built and inspected against AQL | 3–6 weeks |
| Inspection and release | Third-party inspection passed; balance released | 3–7 days |
| Freight and clearance | Cargo at destination, entry filed, duty paid | Sea 30–45 days; air 5–10 days |
| Fulfilment intake | Units received, counted, shelved, ready to pick | 3–10 days |
Add it up honestly and a first hardware run is four to six months before freight, when nothing goes wrong. Campaigns promising ninety-day delivery are not lying on purpose — they are pricing a timeline they have never run. The two durations founders underestimate are certification and the T1 loop: lab queues are not yours to control, and first shots are almost never right.
7. Where the budget actually goes
Founders budget the product and forget the wrapper. A realistic first-run split, as planning ranges rather than quotes:
| Line item | Share of gross raise | Notes |
|---|---|---|
| Platform and payment fees | ~5–10% | Deducted before you see the money; model the net figure |
| Tooling and NRE | ~10–25% | Front-loaded, non-refundable, hardest to recover if the design changes |
| Unit production cost | ~30–45% | Scales with quantity; quoted per unit at a stated volume |
| Certification and testing | ~3–8% | Per model, per market; retesting after a design change costs again |
| Packaging and inserts | ~3–7% | Often quoted separately and easily forgotten |
| Freight, duty and clearance | ~8–15% | Sensitive to volume, mode and destination — the biggest swing factor |
| Fulfilment and backer shipping | ~8–15% | Per parcel, not per container; kills "free shipping" promises |
| Contingency | ~10% | Not optional. This absorbs T2 and the re-test |
The ranges overlap deliberately: the split depends on whether you are making a molded plastic accessory or a battery-powered device with a screen, and tooling, certification, freight and fulfilment together commonly consume a third or more of the raise.
8. Pick a factory that fits your order size
The most expensive mistake in crowdfunded hardware is choosing a factory by capability when you should choose by attention. A large plant with brand-name clients will quote your 2,000-unit order, because quoting costs nothing. It will also assign it to whoever is free, put your tool behind a client ordering 200,000 units a month, and answer your emails when someone remembers. You will not be cheated. You will simply be unimportant — expensive when backers are counting days.
What you want is a factory for whom your order is meaningful: a mid-sized workshop with engineering capacity, a named project engineer, and a track record of first runs. Ask these before sending money, and read the answers rather than the brochure:
Supplier verification is a discipline, not a vibe — see factory audits and supplier verification.
9. The six clauses that decide the outcome
You will not negotiate a fifty-page supply agreement as a first-time buyer, and you do not need to. Get these into the proforma invoice or a short purchase contract, signed.
- Specification by revision. Reference the approved T1 sample and drawing revision. "As per sample" is worthless without a sample both sides sealed and dated.
- Payment tied to milestones. Deposit on PO, tooling tranche on T1 approval, balance on inspection pass. Never pay a balance against a promise to ship.
- Inspection standard and remedy. Name the AQL level and inspecting party, plus what happens to a failed lot — rework, replacement or credit, and who pays freight either way.
- Tool ownership and release. Ownership, tool number, and release conditions.
- Late delivery consequence. A defined per-week deduction or a defined right to cancel. A delivery clause with no consequence is decoration.
- IP and exclusivity. Who owns design files, tooling and custom firmware; whether the factory may sell your design to anyone else. Put it in writing even when the relationship feels good — especially then.
10. Fulfilment: the part backers judge you on
Backers never experience your manufacturing process — they experience a parcel, and review how it arrived, how long it took, and whether a customs bill came with it.
- Who fulfils. A partner that handles crowdfunding backer lists beats a generic 3PL, because backer data contains address exceptions standard warehouse workflows choke on.
- Sea or air. Air reaches backers in days and costs multiples of ocean freight per unit. Common compromise: air for early birds and reviewers, sea for the bulk.
- Who pays duty. A "free shipping" promise shipped DDP without modelling duty and import VAT lands the bill on you or your backer — see DDP shipping from China.
11. The pre-launch checklist, in order
The Bottom Line
Crowdfunding rewards a story; manufacturing rewards a document set. The founders who ship are not the ones with the best campaign — they are the ones who treated the funded moment as the start of an engineering project rather than the end of a marketing one.
Fix the design before the steel. Tie money to milestones instead of promises. Certify the build you will actually ship. And pick the factory that answers your email in week six, because week six is when it matters.
Build the product once, correctly
The China Sourcing Guide covers the full sequence for a first production run — supplier verification, prototyping and tooling, sample approval, inspection standards and landed cost. Start with the free guide and see how the pieces fit.
Get the free sourcing guide