Production prototype, signed-off.
Also called: First-off-tool · Pilot run · Pre-production sample · Process sign-off
The first units made on the real production tooling and process, checked against the design and quality bar before the full run is committed.
Make a handful of units on the actual tooling, not a one-off bench build. Check fit, finish, assembly time, function and reject rate against the spec. Fix process issues now. Sign off the process before you order 500 to 1,000. Get this wrong and every unit carries the fault.
What a production prototype is
A production prototype is the first batch made the way every unit will be made: real tooling, real process, real line. Not a hand-built sample, not a 3D print of the housing, not the engineer carefully assembling one perfect unit on the bench. The point is to prove the process, not the design. The design was proven earlier. What you are testing now is whether the process can repeat it.
This is where a clean design meets a messy reality. The fired ceramic comes out a fraction of a millimetre off the CAD model. The lid that seated perfectly on the prototype binds on the production part. Assembly that took the engineer four minutes takes the line worker eleven, because the fixturing was never specified. None of these are design faults. They are process faults, and they only show up when you run the actual process. So you run it, in small numbers, and you fix what breaks before the numbers get large.
The decision at the end is binary: sign off the process, or fix it and run another batch. I have never regretted holding a sign-off back a week. I have regretted releasing one early every single time.
- The prototype worked, so the production parts will too.
- Skip the pilot batch and order the full run to save time.
- Treat first-off rejects as bad luck, not a process signal.
- Sign off on one good unit the engineer built by hand.
- Run a small batch on the real tooling and line first.
- Measure the reject rate against a number you set in advance.
- Fix the process, then repeat the batch to confirm the fix held.
- Only commit the big run once a repeat batch hits the bar.
How it fits the bigger picture
Production prototype is activity 08.10.06 in Stage 08 Develop. It takes the signed-off design and proves the process can build it repeatably. Once the process is signed off, the next activity is the focus group test (08.20.01), which puts real production units in front of real users to confirm the thing holds up in their hands, not just on the line.
What it can do
It catches the gap between a design that works and a process that repeats, while that gap is still cheap to close. It gives you a hard reject number to decide on, and it stops a quiet tooling fault from being multiplied across an entire run of 500 to 1,000 units.
What it can’t do
It can’t fix a bad design; if the prototype was wrong, the production prototype just builds the wrong thing reliably. And it can’t tell you whether buyers want the product. That is the focus group test, and ultimately the market.
See the full 10-stage process →
Try it yourself
Before you commit a production order, run a small batch on the actual tooling and the actual line. Set your acceptable reject rate before you start, not after you see the results. Check fit, finish, function and build time against the spec. Fix every process fault, repeat the batch, and only sign off when a fresh batch clears the bar. Then order the run.
Want the engineering process behind a clean sign-off? Start the Free Sprint → and we will map your route to a production-ready product.
Your production-prototype checklist
Project notes: the lid that didn’t seat
▸ From the notebook · optional reading
The first 20 units off the Stoke-on-Trent tooling, the 25% reject rate that stopped the run, and the week we spent earning a sign-off.
3 min read · click to open
Dan was ready to order the first 1,000 the moment the tooling was signed off in Stoke-on-Trent. The prototype was lovely. I asked for 20 units off the real tooling first, assembled on the line, before a single production order went out. He thought it was belt-and-braces. It was not.
What the first 20 told us
Five of the twenty failed. Three lids would not seat cleanly on the body, because the production ceramic shrank a touch more on firing than the prototype had, and the mating faces fouled. Two had glaze pinholing on the visible top face, which on a £149 counter object is not something a baker forgives. Assembly was the other shock: the line build took eleven minutes a unit, against the four the engineer managed on the bench, because nobody had specified a jig.
What we fixed
A small relief skim on the lid mating face fixed the seat. A slower glaze-firing ramp cleared the pinholing. And we worked up a simple assembly jig that brought build time down to six minutes and took the human judgement out of the alignment. None of these touched the design. All three were process.
Then I made the call Dan did not love: hold the sign-off a week and run a second batch of 20 to prove the fixes held. The second batch came in at one reject in twenty. That was the number we had agreed in advance, so I signed off the process and we committed the first run of 500 to 1,000. Had we ordered the thousand off that first batch, a quarter of them would have been scrap or returns. The week cost us seven days. It saved us roughly 250 bad units.
— Develop stage, project notes, 2026
— Next in Develop → Focus group test