From Prototype to Production: What Actually Changes

A prototype has to work once, in your hands, under conditions you control. A production product has to work thousands of times, in a customer's hands, built by people who have never met you, using parts that might come from a different supplier next quarter. Those are different engineering problems, not the same problem at a bigger scale.

Take a founder who mills a phone stand prototype from a single block of aluminum on a friend's CNC machine. It looks great, it holds a phone at the right angle, and it took one afternoon to make. Quote that same design for injection molding, the process that starts to make sense once you need five thousand units, and the shop calls back with a problem: the internal ribs that made the CNC part strong create undercuts a mold cannot release from without side action tooling that adds thousands of dollars to the mold cost. Nothing about the part was wrong for a prototype. Everything about it assumed a process that does not scale to volume.

That is the gap this lesson is about: the design, sourcing, and process decisions that are invisible while you are hand building units one at a time, and that turn into real cost or real failure the moment volume shows up.

None of these problems show up when you are building one unit at a time, because you are the one quietly compensating for them without noticing. You sand the part that does not quite fit. You remember which orientation the connector goes in. You call the one supplier who happens to have the part in stock this week. At volume, none of that compensation is available, because you are not the one doing the building anymore.

Design for manufacturing (DFM) means designing the product to be easy and cheap to build correctly, not just possible to build once. It has its own lesson in this section, so this is only the preview: DFM review should start well before you feel ready for it, because late stage design changes are dramatically more expensive than early ones.

Go back to the phone stand. If the DFM conversation happens after the mold is already cut, fixing the undercut means cutting new tooling, often ten to twenty thousand dollars and several weeks of lead time, for a part that could have been redesigned for free on a laptop before any steel was touched. The same pattern shows up everywhere in physical products: a wall thickness that is fine for 3D printing but causes visible sink marks in molded plastic, a snap fit that needs hand pressure to seat but was specified to release at a defined force, a label applied by hand but expected to survive an automated pick and place machine. None of these are exotic problems. They are the ordinary cost of finding out what your process actually requires after you have already committed to it instead of before.

There is no universal signal that tells you it is time, and any founder who claims otherwise is describing their own product, not yours. But these are the signs experienced founders point to most often, and they tend to arrive in a specific order: real validation first, a repeatable process second, a known cost target third.