About 70% of the geometry on a design changes before the part is ready to mold. That number lands hard for inventors who assume a finished CAD model means the manufacturer will press a button and start making parts. The CAD that communicates an invention and the injection-molded production part are two different things governed by two different sets of physics. The bridge between them is design for manufacturing. Since 2010, working with inventors out of Champlin, Minnesota, Enhance Innovations has watched most of them try to skip that bridge at least once.
Here is what happens between a license-ready design and the first production run, and where the money gets lost.
First, where most inventors actually are
Plenty of inventors reading this do not need to cross this bridge themselves at all. If you are pursuing a license deal, the company that takes the license usually owns manufacturing. They tool the part, they run production, they manage the supply chain. What you bring to that conversation is a virtual prototype: photorealistic renderings, a CAD model, and, when motion matters, a short animation. Companies evaluate and license off that package.
This article matters in two cases. You are self-manufacturing rather than licensing, which makes you a small-business owner with resources from the U.S. Small Business Administration behind you. Or a licensee has asked for production-intent detail before signing. In both cases the design still has to be made manufacturable, and the steps below are the same. The point to hold onto: a clean, production-aware CAD model is the asset, whether a licensee takes it to their factory or you take it to yours. If you are still weighing those two routes, the choice between licensing and manufacturing shapes everything that follows, and the step-by-step path from prototype to a finished product sets the wider context.
Why the design is not the production part
A 3D printer adds material in 0.1mm to 0.4mm layers, in any geometry you draw, with no concern for how plastic flows under heat and pressure. An injection mold pushes molten plastic through a gate, around the cavity, into every corner, in milliseconds, then opens and ejects the cooled part. The two processes care about different things, and so does the CAD behind each.
Things a rough concept model ignores that the mold cannot.
Wall thickness uniformity. Variation in wall thickness causes warping, sink marks, and short shots in molded parts.
Draft angles. Without 1 to 3 degrees of draft on every vertical face, the part will not eject from the mold cleanly.
Undercuts. A design can show any undercut you want. A molded part that needs an undercut requires a side action or a lifter, which adds cost to the tool.
Gate location. The point where plastic enters the cavity. Bad gate placement causes weld lines on visible surfaces, jetting, or fill issues.
Ejector pin marks. Molded parts carry small circular marks where the ejectors pushed them out. You design these into a non-cosmetic surface.
Tolerance achievable. A printer often hits +/- 0.2mm on a small part. A mold can hold +/- 0.05mm with the right tooling, or struggle to hit +/- 0.3mm with the wrong tooling.
Shrinkage. Plastic shrinks as it cools. ABS shrinks about 0.5%. POM shrinks about 2%. The CAD has to be drawn larger than final to account for it, with the percentage varying by resin.
This is why DFM (design for manufacturing) review exists. The DFM pass is a methodical look at every feature of the part to confirm it can be made by the chosen process at the chosen volume.
DFM review: what it checks
A real DFM review takes a CAD model and a target manufacturing process and asks the same question of every feature: can this be made, at the quantity needed, at the cost needed, with the quality needed?
The checklist for an injection-molded plastic part includes.
Wall thickness. Uniform and within the resin's recommended range (1 to 3mm for most consumer-product resins). Variations tapered, not stepped.
Draft. Every vertical face needs draft. Standard is 1 to 2 degrees. Textured surfaces need more.
Radii on inside and outside corners. Sharp internal corners cause stress concentration. Sharp external corners are hard to fill.
Bosses (cylindrical features for screws). Need draft, ribs, gussets. Wall thickness around the boss should be 60% of nominal wall to avoid sink.
Ribs. Used to add stiffness without adding material. Should be 60% of wall thickness, with draft, and rounded.
Gate location and parting line. Decided by the tool maker, but the designer should have a preference for where they go because they leave visible marks.
Ejector pin layout. Pins push the part out of the cavity and leave round marks. Plan where they land without affecting cosmetics or function.
Undercuts. Each undercut needs a tool action, a lifter, or a workaround. Each adds tool cost.
Threads. Molded threads are doable but expensive. Most products use threaded inserts (heat-staked or ultrasonic) instead.
Fits with mating parts. Slip fits, snap fits, press fits. Each requires a specific tolerance band the mold can hold across thousands of shots.
For a moderate-complexity consumer product with five to fifteen parts, a thorough DFM review takes one to three weeks. The output is a marked-up CAD package, a list of recommended changes, and an estimated cost-per-part at the target volume. This is the work an engineering firm does, and it is the difference between a CAD model that looks finished and one a factory can actually run. Enhance handles this through its engineering and prototyping service.
Tooling decisions: the money question
Once DFM is clean, the decision is how to make the parts. The choice depends on volume, part size, material, and budget. These are generic industry ranges, not Enhance pricing.
The numbers move with complexity, geometry, material, and where the tool is built. A mid-complexity housing tooled overseas ranges from $5,000 to $15,000 per cavity. The same tool built in the US costs $15,000 to $40,000.
Most inventors going from design to first production run choose between aluminum soft tools (for runs of a few thousand to ten thousand) and full steel production tools (for tens of thousands or more). Aluminum tools are cheaper, faster, and wear out sooner. Steel tools cost more, take longer to build, and run for a million-plus shots.
A common path: build aluminum tools for the first production run while you validate the market, then commit to steel tools when reorders justify the cost. Selecting the right factory at this stage is its own discipline, and finding a manufacturing partner deserves the same care as the tooling choice.
MOQ realities
Minimum order quantity (MOQ) is what a manufacturer requires you to order to make the run worth their setup time. MOQs are not negotiable in most cases. They are the floor below which the math does not work for the manufacturer.
If the product needs five different molded parts, an electronics assembly, packaging, and labels, the per-component MOQs stack. A first production run might require 1,000 to 3,000 finished units even when each molded-part MOQ is only 1,000.
Inventors get burned here when they assume "I will start with 100 units" and discover the supplier minimum is 1,500. They either pay for 1,500 or do not get the run. There is no path to 100.
Sample approval, golden sample, first-article inspection
Before the production run, the manufacturer makes samples. These go through a structured approval process. Skipping any step here is how inventors end up with 5,000 defective units.
T0 sample (or T1, depending on the tooling vendor). The first part out of the new mold. The point is to confirm the tool fills, ejects, and produces a part shaped like the CAD. Cosmetic finish is rough. Color is approximate. Function is the focus.
Engineering verification sample. The tool has been adjusted for fill issues, sink, and warpage. Cosmetic finish is closer to final. The engineering team measures critical dimensions, runs functional tests, and signs off or sends back changes.
Golden sample. The reference part. Once an engineering verification sample is approved as the standard, the manufacturer keeps a copy as the golden sample. Every production part is compared against it. If a production part deviates, the golden sample is the arbiter.
First-article inspection (FAI). Before full production, the manufacturer produces a small batch (5 to 20 parts), measures them against the spec, and submits a report. The engineering team approves the FAI report before the full run starts.
Production run. The manufacturer makes the rest of the order. Quality inspection happens during and after the run.
Skipping the FAI is how inventors find out their full 10,000-unit run has a 0.4mm offset on a critical dimension after the last carton has shipped.
PPAP for regulated products
If your product is a medical device, an automotive part, an aerospace component, or anything else with regulatory or safety constraints, the manufacturer will produce a PPAP (Production Part Approval Process) document. PPAP is a structured packet of evidence proving the part is being made the way it was specified.
Standard PPAP includes 18 elements. Process flow diagram. Failure mode and effects analysis (FMEA). Control plan. Measurement system analysis. Capability studies. Sample products. And a dozen others.
PPAP submission levels run 1 through 5. Most consumer products do not need PPAP. Anything regulated or safety-critical does.
If your product is in this category, plan for an additional 4 to 12 weeks and $5,000 to $25,000 in PPAP-related work on top of the base tooling and sample approval timeline.
The 70% rule and what it means for your timeline
The "70% of the design changes during DFM" number is real, and it shapes the timeline.
A typical injection-molded consumer product timeline from a frozen design to the first production unit.
These numbers are for a single part. A product with multiple molded parts runs each on its own timeline, and the integration step that pulls them together adds weeks.
Inventors used to 3D printing timelines (a part in a week) often underestimate this. The first production run of an injection-molded consumer product is seldom under four months from a frozen design. Six to nine months is normal.
Where inventors lose the most money on this bridge
Five recurring patterns Enhance sees.
Picking a manufacturer before DFM. The inventor signs with a manufacturer based on per-part quotes built from a CAD that has not been DFM-reviewed. The manufacturer accepts the file, then comes back six weeks later with $12,000 in tool changes that were obvious to anyone reading the file carefully. The inventor pays. The lead time slips.
Assuming the concept CAD is the production CAD. It is not. Production CAD has different wall thicknesses, different draft, different fillets, different tolerances. The concept model is the starting point for the production CAD, not a substitute for it.
Underestimating tooling cost. The inventor budgets $8,000 for tooling and the actual quote is $35,000. They cut corners with a single-cavity tool with a shorter lifetime, then pay for two tools instead of one when the first wears out at 30,000 cycles.
Skipping samples to "save time." The inventor approves the production run off the T0 sample because the timeline is tight. The full run reveals a sink mark on the most visible surface. They ship anyway, take the customer service hit, fix it on tool revision two.
Trying to negotiate MOQs to zero. The inventor pushes for "a small first run, just 200 units." The manufacturer agrees and quotes $42 per unit. The math never works at retail. The same product costs $4.20 per unit at MOQ.
How an integrated firm handles this bridge
When an inventor brings a design toward manufacturing, the early weeks of the engagement tend to look the same. A complete DFM pass on every part of the assembly. The CAD rebuilt for production. Tolerances documented. Each DFM-driven change mapped back to the inventor with the reason it matters. Tooling quoted from domestic and overseas tool makers. Manufacturer selection based on the tool design, the resin, the volume, and the inventor's risk tolerance.
Enhance Innovations runs design, CAD and engineering, renderings, marketing materials, manufacturing sourcing, and licensing representation under one roof. That matters on this bridge because the production CAD, the sourcing decision, and the manufacturer handoff are not separate projects handed to separate freelancers who each restart from someone else's files. They are one continuous chain of work.
The bridge is the most expensive part of getting an invention to market. Skipping any piece of it does not save money. It moves the cost downstream where it costs more. If you are not sure yet whether you are headed for a license deal or self-manufacturing, the $399 patent search is the low-friction first step, it confirms whether the idea is clear before any tooling money is on the table. Reviewing the USPTO patent process overview first helps you understand where that search fits.
FAQ
How long after my design is "done" until I have units to sell?
For an injection-molded consumer product, plan for four to nine months from a frozen design to the first production units in hand. Add buffer for shipping, customs, and any quality issues that need correction.
Do I have to cross this bridge myself?
Not if you license. A licensing company usually owns manufacturing and crosses this bridge with their own supply chain. You cross it yourself when you self-manufacture, or when a licensee asks for production-intent detail before signing.
Can I keep using 3D-printed parts for the first batch instead of tooling?
For very small quantities (under 100 units) and non-cosmetic parts, sometimes yes. Multi Jet Fusion and SLS produce parts that can pass for short-run production in industrial applications. For consumer products with cosmetic surfaces and tight tolerances, the finish and tolerances of 3D printing seldom match what retail buyers expect. Most products move to molding by 500 to 1,000 units.
What is the cheapest way to get my first 1,000 units made?
For most injection-molded products, the cheapest path is a single-cavity aluminum soft tool plus a 1,000-unit run. Generic tooling cost lands in the $5,000 to $15,000 band plus per-part cost. For 1,000 units, the all-in is often $12,000 to $25,000 depending on complexity. Cheaper paths exist (urethane casting, low-volume 3D printing) but the per-part cost is higher and the finish is usually worse.
Should I tool in the US or overseas?
US tooling runs 2 to 3 times the cost of overseas tooling, with shorter lead times, easier oversight, and stronger IP protection. Overseas tooling is cheaper and faster to a low-cost steady state but adds shipping, communication, and IP risk. Many inventors split the difference: tool overseas, run production there, and stage final assembly and inspection in the US.
What is a "tooling deposit" and how much should I expect to pay?
A tooling deposit is what the manufacturer charges to start building your tool. The standard is 50% of the tool cost upfront, with the balance due when the tool produces an approved sample. Deposits of 100% upfront are a red flag unless you have a long relationship with the supplier.
Who owns the tool when it is built?
This is in the contract. If you paid for the tool, you should own it, with rights to move it to a different manufacturer if needed. Some manufacturers try to retain ownership and only license its use to you. Read the contract. Get tool ownership in writing.