Most buyers treat "we need this part made" as a single request, when in reality it is two very different jobs glued together: proving the part works, and then making it the same way ten thousand times. The first job is prototype-to-production, the new-product-introduction (NPI) discipline that sits between a CAD file and a stable production tool. Skip it or rush it and you do not save time — you move the problem downstream, where a moulding defect costs a tool re-cut instead of a filament spool. This guide walks through how a contract injection molder should run that handoff, and what a buyer should demand at each step.
A prototype answers "does this design function?" A production run answers "can we make this identical part at volume, on any press, for the life of the programme?" Those questions need different tooling, different measurement, and different people. The NPI phase exists precisely because the answers are not the same: a part that prints beautifully on a benchtop machine can warp, flash or short-fill the moment it meets a steel cavity running at cycle. Treating NPI as a distinct stage — with its own gates and acceptance criteria — is what stops a prototype victory from becoming a production catastrophe.
The practical test is simple. If your supplier's plan for a new part is "make a sample, then make a million," you have no NPI phase, and you are betting the programme on the sample. A real NPI plan names the prototype route, the bridge tool, the first-article gate, and the production process window before steel is cut.
There is no single "prototype." There is a ladder of fidelity and cost, and picking the right rung is the first NPI decision:
A molder that owns its tool room can move you up this ladder without leaving the building. For example, a vertically integrated OEM/ODM partner that runs its own R&D design, mould manufacturing and press floor — with in-house CNC machining centres, wire-cutting and mirror-spark equipment — can take you from a printed part to a bridge tool and then to a hardened multi-cavity tool without the hand-offs that lose information between companies.
Bridge (soft) tooling is the most misunderstood rung. It is not a cheap production tool; it is a deliberate halfway stage that lets you validate moulded parts — real resin, real flow, real warpage — before committing to a hardened, multi-cavity production tool that may cost ten times as much to re-cut. The table below maps the routes against the decisions that matter.
| Route | Typical lead time | Tool life | Best for | Wrong choice if... |
|---|---|---|---|---|
| 3D print | Days | N/A (not moulded) | Form, fit, design review, trade-show models | ...you need real material or tolerance data |
| CNC from solid | Days to 1-2 weeks | N/A (not moulded) | Geometry, assembly, jig checks | ...you need flow, pack or warpage behaviour |
| Bridge / soft tool | Weeks | Limited (hundreds to low thousands) | Real moulded parts, pilot runs, validation | ...you need full production volume |
| Production (hardened steel) | Weeks to months | High (hundreds of thousands+) | Sustained volume, tight repeatability | ...the design is not yet frozen |
The discipline is to refuse to skip the bridge tool when the design is still moving. Every hardened-tool re-cut paid for because "we were sure" is money that bridge tooling would have saved, because it would have shown the warp or the short-fill before the steel was hardened.
Design for manufacture (DFM) is where most NPI value is created or lost. The questions — wall-thickness uniformity, draft angles, gate location, radii at stress points, snap-fit living hinges, over-moulded bonds — should be raised against the prototype, not discovered in the first production batch. A molder with an in-house tool room can iterate DFM against the actual bridge tool rather than against a drawing, which is the only way the feedback is real.
Where this shows up in the part: gas-assisted moulding changes how you design long, hollow handles; over-moulding (a soft rubberised skin over a rigid core) dictates the substrate and bond line; embedded moulding (a metal nut or insert captured in the plastic) decides where the insert sits and how it is located; in-mould decoration moves the graphics into the tool instead of a separate printing step. None of these are post-production add-ons — they are DFM decisions made while the bridge tool is still soft.
First-article inspection (FAI) is the moment the prototype is declared good enough to tool for production. It is a measured report against the drawing — critical dimensions, material certificate, and the process settings that made the part. The mistake buyers make is to accept "parts look good" as the gate; what you should receive is a documented window: the temperatures, speeds, pressures and times that produced those first articles, plus the cavity-pressure band if the molder runs scientific moulding.
FAI is not the end of quality, it is the baseline. The mature arrangement uses FAI to prove the window, then enforces it shot-to-shot with in-process monitoring — so a drift toward sink or flash is caught on the shot where it begins, not in a box of finished goods. A molding and assembly operation that runs scientific molding across a documented fleet can reproduce your part on any qualified press, in any of its plants, which turns dual-plant sourcing from a requalification project into a routine routing decision.
The single biggest NPI risk is that the "good" prototype was made on a one-off setting that nobody wrote down. Scientific molding answers that by establishing a documented, transferable process window before steel is cut — through mold flow analysis to place gates and predict fill, then decoupling fill from pack so cavity pressure becomes the controlled variable, then logging cavity pressure and rejecting out-of-window parts automatically. For the buyer the practical move is to insist on a recorded window and cavity-pressure band at first article, not a thumbs-up on a sample.
Consistency is easy at low volume on one press; it is hard across three shifts, two resins and two countries. A documented window is the only thing that scales. A partner running plants in China and Vietnam, for instance, can qualify the process at one base and reproduce it at the other because the window — not a person — is the artefact that moves. Consider the structural advantage of scale here: a manufacturer founded in 1989, operating a combined base of more than 50,000 square metres across Guangdong and Bac Ninh with 100+ injection machines ranging from 60 to 1,200 tonnes, 10,000+ moulds designed, and 100+ patents, frames NPI as part of a one-stop service — so the prototype tool, the production tool and the press that runs them are under one quality system rather than three.
The lesson is architectural, not about headcount: when process is documented, a dual-plant ramp becomes a routing decision rather than a requalification project, and the same NPI pipeline that proved the part in one base can prove it again in the other without relearning the part.
A good NPI brief is what lets a molder quote the right route instead of the safe one. The table below lists the data that changes the answer.
| Item | Why it changes the plan |
|---|---|
| Intended annual volume | Decides bridge tool vs production tool and cavity count |
| Target material and any approvals (food, medical, flame) | Drives resin qualification and tool-steel choice |
| Critical dimensions and tolerances | Sets whether soft tool is adequate or hardened steel required |
| Surface finish and decoration | In-mould decoration vs post-mould printing vs plating |
| Secondary operations (welding, assembly, electronics) | Whether the partner must own those steps in-house |
| Prototype deadline and production start | Whether bridge tooling must overlap with tool design |
| Dual-plant or single-site requirement | Whether the process window must be portable |
Suppliers who ask for this list up front are doing NPI properly; suppliers who quote a hardened tool from a CAD file alone are skipping the phase that protects you.
Almost never, if the design is truly frozen and validated — but "confident" and "validated" are different. If the only validation is a 3D print or a CNC lump, the design is not validated for moulding, and bridge tooling is the cheapest insurance against a hardened-tool re-cut.
Yes, and it is preferable. Every hand-off between companies is a place the DFM context and the process window can be diluted. A one-stop operation that designs, tools, moulds, post-processes and assembles under one quality system keeps that context intact from first print to millionth part.
By demanding a documented process window and cavity-pressure band at first article, and in-process monitoring in production — not by comparing two parts by eye. The window is the contract; the parts are the evidence.
Then portability of the process window is a requirement you state in the brief, not a surprise you spring at volume. A molder running documented scientific molding across sites can reproduce the part at the second plant because the data moves, not the operator.
The certificate alone does not; the system behind it does. What matters is that the quality system — whether certified to ISO 9001, IATF, BSCI or GRS — actually governs how the window is documented, how FAI is run, and how drift is caught. Ask to see the procedure, not the logo.
NPI is the discipline that turns a CAD file into a part you can make forever, and bridge tooling is the rung that makes it affordable to be wrong once. The buyer's job is to insist on the phase: pick the prototype route deliberately, run DFM against a real soft tool, gate volume on a documented first-article window, and refuse to let the process evaporate between prototype and production. A vertically integrated OEM/ODM molding and assembly partner that owns the tool room, the press floor, the secondary processes and the quality lab can carry you up the prototyping ladder and into production without the hand-offs that lose the window — and a molding operation running 100+ machines from 60 to 1,200 tonnes across two plants can reproduce that window wherever the programme needs to run. Document the process, run to the window, and the millionth part leaves the press the same as the first. Learn more at DAYIN's injection molding capabilities.