A prototype that works is a genuinely great day. It is also where most first-time founders badly misjudge how much is left, because the mental model is that the hard part is over and the rest is "just manufacturing." In practice, a hand-built unit that works on a bench is a quarter to a third of the way to a product you can box, ship to a stranger, and stand behind for two years.
The gap is not one big task. It is thirty small ones, each invisible until you hit it. This article lays them out so you can budget against reality instead of optimism.
The core difference: one that works vs. thousands that work
A prototype has to work once, in your hands, under conditions you control. A product has to work every time, in anyone's hands, under conditions nobody controls — including a user who drops it, a factory operator who assembles it slightly differently, a component lot that runs at the edge of its tolerance band, and a warehouse in Arizona in August.
Everything below follows from that single change in requirement.
1. The design gets rebuilt for manufacturing
Prototype parts are usually machined, printed, or hand-modified. Those processes have no draft angles, no parting lines, no shot-to-shot variation, and no tooling cost. Production processes have all four. Converting a printed housing into an injection molded one is a genuine redesign: wall thickness gets equalized, ribs replace solid sections, bosses get repositioned so the tool can eject, and snap features get re-engineered because molded material behaves differently than printed material. This is the discipline of design for manufacturing, and skipping it is the single most common reason a first tooling attempt has to be recut.
Expect the mechanical redesign alone to run several weeks and to change the part's look slightly. Founders attached to the prototype's exact geometry find this stage painful.
2. Components get sourced, not scavenged
Prototypes run on whatever was in stock at a distributor. Products run on parts with a defined manufacturer part number, a second source where possible, a lead time you have verified, and a lifecycle status that is not "not recommended for new designs." Every specialty part becomes a schedule risk. Every part with a sixteen-week lead time silently becomes your launch date.
The BOM becomes a controlled document
A prototype bill of materials is a spreadsheet you update when you remember. A production one is versioned, has approved manufacturer part numbers, and cannot change without a paper trail. That transition is where a lot of chaos gets removed from a project.
3. Tolerances stop being decorative
On a one-off, you file the part until it fits. In production you cannot. Every dimension that matters needs a tolerance, every stack of mating parts needs to be checked at both extremes, and every fit needs to be specified rather than assumed. Assemblies that closed beautifully on the prototype start binding at one end of the range and rattling at the other. This is ordinary and fixable, but only if it is done deliberately before tooling rather than discovered afterward.
4. It has to survive being tested to failure
A prototype demonstrates function. A product has to demonstrate durability: drop testing, vibration, temperature and humidity cycling, ingress protection, cycle life on every moving element, and battery behavior at the extremes of its rated range. Formal reliability testing including HALT and life testing exists to find the failure you would otherwise find in the hands of a thousand customers.
Budget for units that get destroyed. That is the point of the exercise.
5. Certification enters the picture
Nothing about a prototype is regulated. A product for sale usually is. Depending on what it does, that can mean FCC emissions testing for anything with a radio or a fast clock, UL or an equivalent NRTL listing for anything on mains power, CPSC requirements and third-party testing for children's products, FDA pathways for medical claims, and UN 38.3 for lithium cells before a carrier will fly them.
The relevant list is product-specific, and working out which certifications a new electronic product actually needs early changes design decisions — shielding, creepage distances, plastic flammability ratings — that are expensive to retrofit. Certification also consumes calendar time you cannot compress with money: labs have queues, and a failed scan means a redesign and a retest.
6. Documentation replaces knowledge in your head
A prototype lives in the memory of whoever built it. A product lives in files: dimensioned drawings with tolerances and finishes, Gerbers and pick-and-place data, an approved BOM, assembly instructions with torque values, test procedures with pass criteria, packaging drawings, and label artwork. That whole bundle is the manufacturing data package, and a factory cannot quote accurately or build repeatably without it.
Founders consistently underestimate this stage because it produces no visible object. It is nonetheless the difference between a contract manufacturer building your design and a contract manufacturer improvising it.
7. Production needs its own hardware
Series assembly requires things the product itself does not include: assembly fixtures that hold parts in the right relationship, a test fixture that verifies every unit electrically and functionally, programming jigs for firmware, and gauges for critical dimensions. A production test fixture is not optional at any real volume; without one you are shipping units nobody checked.
8. The cost model changes shape
Prototype cost is dominated by labor and one-off parts. Product cost is dominated by material, cycle time, and amortized tooling. A part that costs $180 machined may cost $2.40 molded — and require $28,000 of tooling first. The reverse also happens: a prototype that was cheap by hand becomes expensive at volume because it needs a secondary operation nobody accounted for. If your numbers do not work at volume, the fix is design change, not negotiation, which is the whole subject of what to do when your prototype costs too much to manufacture.
9. Somebody has to make the first hundred
Between the last prototype and mass production sits a build that is neither: a pilot run on production tooling, with production processes, at a fraction of full volume. This is where assembly time gets measured, where operators find the step that is ambiguous, and where the first real yield number appears. Running a proper pilot production run is cheap insurance compared to discovering a systematic defect across ten thousand finished units.
10. The product includes everything around it
Packaging that survives parcel shipping. A manual someone can follow. A serial number scheme. A warranty policy and returns path. Spare parts. A support inbox. None of it is engineering, all of it is required before the first customer, and it adds a month nobody scheduled.
A realistic picture of the gap
| Dimension | Prototype | Finished product |
|---|---|---|
| Quantity that must work | One | Every unit, for years |
| Part making | Printed, machined, hand-fitted | Tooled, with fixed tolerances |
| Documentation | Sketches and memory | Controlled drawings and BOM |
| Testing | Does it function | Does it survive abuse and testing to standards |
| Typical added time | — | Six to eighteen months |
The step that gets skipped most often is deciding what kind of prototype you needed in the first place. A demonstration unit and a production-intent unit are different objects with different budgets, which is why it pays to be clear on whether you need a looks-like or a works-like prototype. The broader path is covered in our guide to going from prototype to production.
If you have a prototype that works and no clear map of what stands between it and a shippable product, Projects House does exactly this transition — DFM, documentation, testing, certification planning, and factory transfer. Send us photos and a description through our contact form and we will tell you what is actually left.