It finds bad parts. It doesn't stop them from being made.
Quality inspection is the oldest quality tool there is. For most of manufacturing history, the way shops kept bad parts from reaching customers was to put someone at the end of the line with a gauge and a clipboard. That model still exists in many shops, and it is not wrong, but it is not enough. Inspection is a sorting activity. It separates good from bad after the fact. It does nothing to change what happens on the next part.
"Inspection tells you what you made. It doesn't tell the process to make it right."
At its core, inspection is comparison. You have a requirement, written on a print or a spec sheet, and you have a part. You measure or observe the part and decide whether it meets the requirement. The mechanics vary widely. A dimensional check uses calipers, micrometers, gauges, or a CMM. A visual check uses a sample board or a written standard. A functional check runs the part through its intended use, like cycling a switch or pressurizing a fitting.
There are three common positions for inspection in a shop. Incoming inspection checks material from suppliers before it enters production. In-process inspection checks parts between operations, often at the operator's bench. Final inspection checks finished goods before shipment. Each catches different problems. Incoming catches supplier issues. In-process catches operation-specific problems and prevents bad parts from consuming more value. Final catches anything the first two missed.
Inspection can be 100 percent or sample-based. 100 percent inspection looks at every part. Sample-based inspection looks at a defined subset, often using AQL tables or process capability data. Sampling is cheaper but accepts a statistical risk that defects get through. 100 percent is more thorough but expensive and, ironically, less reliable for repetitive visual checks because attention drifts.
Picture a 30-person contract machine shop running small batches of stainless brackets for a medical device company. The customer requires a final QA sign-off on every shipment, and the shop has one inspector running a CMM in a small enclosed area off the production floor. Parts pile up at her door waiting for first-article approval and end-of-run checks. The bottleneck is real: jobs sit for hours waiting on her sign-off.
A lean rebuild would push most of the check work back to the operators. First piece off setup, the operator runs three datums themselves at the bench with calipers and a height gauge against the print, signs the inspection sheet, and starts the run. The inspector now only sees first articles when geometry is unusual and the final shipment audit. The CMM goes from being a bottleneck to being a verification step. The same volume of parts gets checked, but the queue at the inspector's door disappears.
Inspection sits at the bottom of the lean quality hierarchy. The lean alternatives that catch problems earlier and cheaper are self-inspection, where the operator checks their own work at the bench, and successive-inspection, where the next operator catches the previous step's problems. Earlier still is source-inspection, which checks the conditions that produce defects before the defect can happen. When inspection does happen as a formal checkpoint, it is usually a quality gate at a stage boundary, and the lean preference is to make those gates rare by building quality in upstream.
The questions we hear most about this term.
Long-form guides that pick up where this definition leaves off, written for manufacturers running Arda today.
Same-day setup. No distributor lock-in. Zero stockouts. Top teams double revenue in 9 months.
.svg)
.svg)
%201.svg){kind=logo}
.png)
.svg)
.svg)
