A pilot run is not a small production run. It exists to prove the process, the fixtures, and the documentation.
The common misreading
Most pilot builds are treated as a small batch of sellable units. Fifty boards, built carefully, shipped to a customer trial. The engineering team watches the yield and calls the pilot a success if the number is high enough.
That reading wastes the run. A high yield built by the two most experienced operators on the floor, working from tribal knowledge and a half-finished work instruction, proves almost nothing about what happens when the same board is built in volume by a different shift.
What the run is actually for
A pilot exists to prove three things that only a real line can prove. That the process window holds — print, place, and reflow, at the real panel and the real thermal mass. That the fixtures and test rigs work at pace rather than on the bench. And that the documentation is sufficient for someone who was not in the design meetings.
The product is incidental. The process is the deliverable.
If your best operator is what made the pilot succeed, you have not proven the process.
How to run it so it proves something
Build it with the operators who will build the volume, not the ones who designed it. Have an engineer present to observe rather than to help — the moment an engineer touches a unit, that unit stops being evidence.
Log every deviation, every hesitation, every point where the operator had to ask. Those are not defects; they are the specification for the next revision of the work instruction.
Closing the run properly
A pilot closes with document changes, not a yield number. Updated work instructions, updated fixture drawings, updated inspection criteria, and a written list of what remains unproven and how it will be proven.
If the run produced no document changes, it was not a pilot. It was a small production run with optimistic paperwork.
Written by the manufacturing delivery team. Published articles carry a named author once attribution is confirmed.
