From a working prototype to a product a factory can repeat

The process that made the prototype is not the process that will make the product

Prototypes are made by methods chosen for speed: machined instead of moulded, printed instead of cast, hand-soldered instead of reflowed, assembled by the person who drew it. Each substitution changes behaviour. Moulded material shrinks, warps and has different strength along different directions; a cast surface finishes differently; a reflowed board behaves unlike one built with an iron; an adhesive cured on a bench differs from one cured to a takt time. So parts that fitted perfectly stop fitting, and a mechanism that felt right becomes stiff. Expect to revalidate function on parts made by the intended process, not merely to check dimensions.

Repeatability is a design property, not an assembly skill

A designer building the tenth unit compensates unconsciously: aligning a part by eye, choosing the screw that goes in easily, adjusting until it works. A production operator building many units per shift cannot and should not do any of that. Features that can only be assembled one way, connectors that will not seat backwards, fasteners of a single length, self-locating parts and tests that catch the errors people actually make all belong in the design. If assembly depends on judgement, output quality will follow whoever is on shift, and the yield you achieved in the workshop will not survive contact with a production line.

The bill of materials stops being a list and becomes a supply chain

Parts bought singly from a catalogue become parts bought in quantity on lead times, from vendors with their own qualification status, minimum quantities and lifecycle plans. Somebody has to check that every item is still in production, has an approved source, tolerates the shipping and storage conditions, and does not sit at a single supplier with no alternative. Components chosen during prototyping because they arrived quickly are frequently the wrong ones for volume. This review is unglamorous and repeatedly deferred, and it is the most common reason a first production run stalls with everything ready except one item.

Freezing the design, and what a soft freeze really costs

Tooling, test fixtures, packaging, documentation and supplier qualification all depend on a stable definition, and each late change ripples through every one of them. Teams resist a freeze because something can always be improved, so they declare one and keep changing the design underneath it. The cost lands as scrapped tooling inserts, reprogrammed fixtures, obsolete inventory and a manufacturer who stops believing your revision numbers. Better to hold a short, deliberate change window with a named approver, batch the changes into defined revisions, and be honest about which improvements will wait for a later build.

Gates between the first build and steady output

The transition works best as a sequence with evidence at each step rather than a single leap. Engineering builds prove that the design can be assembled at all. Design validation confirms the product does what it claims on representative parts. Process validation shows the intended equipment and settings produce conforming output consistently. A pilot run then exercises the whole system, including planning, documentation, packaging and people. Skipping a gate does not remove the work; it moves it into the launch, where discovering the same issue costs far more and is watched by customers rather than by engineers.

Frequently asked questions

How do we know we are ready to commit tooling?

When the features the tool will define are settled and validated on parts made the intended way, when the material is chosen and available, and when the remaining open items are ones that live outside the tool. Committing earlier buys schedule and risks paying for the tool twice. Where pressure is severe, consider tooling only the stable portion, or accepting a temporary tool for early builds while the final one waits, provided everyone understands what that temporary route cannot prove.

Should the shop that built our prototypes also make production?

Not automatically. Prototype specialists are optimised for speed, variety and hand skill, and many are honest that volume is not their business. Some do transition well, and the continuity is genuinely valuable because the tacit knowledge stays in one building. Judge them on the same evidence as anyone else: equipment for the intended process, capacity for your rhythm, documentation practice and measurement capability. Familiarity with your product is worth something, but it does not substitute for capability at volume.

What most often causes the first production run to fail?

Rarely the headline technology. It is usually a missing component, a fixture that was never made, an inspection method nobody agreed, a work instruction written the night before, packaging that has not been tested, or a design change that reached engineering but not the floor. In other words, the supporting system rather than the product. That is exactly what a pilot run exists to expose, which is why compressing the schedule by deleting it tends to be a false saving.