Pre-production prototype, real.
Also called: Pilot build · Near-production prototype · Process-proving unit · Pre-tooling sample
A unit built the near-production way, in the real materials and process, to flush out what only making it for real reveals.
Build one the way you intend to make the run: real materials, real assembly sequence, near-final parts. It de-risks the process, not just the design. Skip it and you discover the surprises after committing to tooling and a batch, when they cost the most.
What a pre-production prototype is
A pre-production prototype is a unit made as close to the way you intend to manufacture it as you can manage before you commit to the full tooling and the run. The point is not to prove the design works, you should already know that. The point is to prove the process works: that the real materials behave, the assembly sequence holds together on a bench, and the performance survives the move from a hand-built lab unit to something made the production way.
This is the gap that catches teams out. A breadboard or a 3D-printed shell tells you the idea is sound. It tells you almost nothing about what happens when the shell is fired ceramic that shrinks unpredictably, or when the parts arrive from three suppliers and have to go together in a fixed order. In my experience the design is rarely the thing that bites at this stage. The process is.
So you make one for real, or as near to real as the budget allows. Soft tooling instead of hard tooling. A near-final board instead of the dev board. Then you watch it like a hawk and write down everything the process tells you, because it will tell you things the CAD never could.
Here is what that unit looked like for the proofing box we ran through the framework, so you can see the shape of a real pre-production build rather than a generic template.
The pre-production unit earned its keep in three lines of the table. None of those surprises were design faults. They were the things only the real fired ceramic could tell us, and finding them now costs a re-cut, not a scrapped tool.
- Commit to hard tooling off the back of a lab prototype.
- Discover the ceramic shrinkage on the first production batch.
- Find out the assembly order doesn’t work with 200 units on the bench.
- Pay to re-cut a tool you already paid for.
- Soft-tool one shell, fire it, see what the material actually does.
- Build one unit in the intended order and watch where it fights you.
- Catch the shrinkage and the sequence faults on unit one.
- Re-cut allowances cheaply, before the run is committed.
How it fits the bigger picture
Pre-production prototype is activity 08.10.05 in the framework, inside Stage 08 Develop. It builds on the earlier development prototypes that proved the design, and it feeds directly into the production prototype (08.10.06), the unit made on the actual production tooling that signs the process off for the run.
What it can do
It surfaces the process faults that hide until you make the thing for real: material behaviour, tolerance stack on fired or moulded parts, and whether the assembly sequence holds on a bench. It lets you fix them while a fix is still cheap, before tooling and a batch are committed.
What it can’t do
It can’t fully replicate the production line; soft tooling behaves differently to hard tooling, and one hand-built unit won’t expose the faults that only appear at volume. That is what the production prototype and the first pilot run are for. This stage de-risks the process; it doesn’t sign it off.
See the full 10-stage process →
Try it yourself
List every part of your product that changes when you make it the production way, not the prototype way: a moulded shell instead of a printed one, a real board instead of a dev board, parts from real suppliers. Build one unit using those, in the order you intend to assemble the run. Write down every place it fights you. Those notes are the whole point.
Want the engineering approach mapped to your product first? Start the Free Sprint → and the GPT will help you frame the build before you spend on tooling.
Your pre-production checklist
Project notes: what the kiln told us
▸ From the notebook · optional reading
One soft-tooled shell from a Stoke-on-Trent pottery, and the millimetre of shrinkage that would have scrapped a hard tool.
3 min read · click to open
The design was signed off. The thermals worked on the dev unit, the board layout was final, and Dan wanted to commit the ceramic tooling and order a first batch. I asked for one thing first: “Let’s make one the real way before we spend on a hard tool.”
The soft-tooled shell
We worked with a small Stoke-on-Trent pottery who could cast a shell from a soft plaster tool, fire it in their kiln, and hand us a piece of the actual production ceramic for a fraction of what hard tooling costs. The maker was blunt about it: “It’ll move in the firing. They always do. The question is how much.”
It moved about a millimetre across the long wall. Enough that the PCB mounting bosses, dimensioned off a 3D-printed mock-up, fouled the inside of the fired shell. The lid sat proud. On the printed prototype everything had dropped together perfectly. On the real fired ceramic it didn’t.
What it saved
If we’d gone straight to a hard tool, that shrinkage would have been cast into a steel mould we’d already paid for, and the fix would have meant re-cutting the tool. Instead we re-cut the shrinkage allowance on the plaster, moved the bosses 2mm inboard, and re-sequenced the bench assembly so the board went in before the lid seal, not fighting it after.
We fired a second shell, assembled it in the new order, and re-ran the thermals. Still 26°C, clearances still inside the BS EN 61010 margins on the real layout. Then we committed the tooling.
The whole exercise cost one soft-tooled shell and about a week. A re-cut hard tool would have cost a good deal more, and a delayed first batch on top. The kiln told us something the CAD never would have, and it told us while it was still cheap to listen.
— Develop stage, project notes, 2026
— Next in Develop → Production prototype