Quality Planning Best Practices
7 min read · Updated 2026-08-17
Nothing here is novel. All of it is routinely skipped under launch pressure, which is exactly when it matters.
1. Define each characteristic once
A characteristic should be created once — normally when the drawing is read — with its nominal, tolerance, classification and measurement method. Every document afterwards references it. This single practice removes most of the reconciliation work in a quality department, and its absence is the root cause of most document disagreement.
2. Control the process, not just the product
Measuring the finished feature tells you what already happened. Controlling the parameter that produces it stops it happening. A control plan consisting entirely of product measurements describes an inspection department. Aim for process controls first, product verification as confirmation.
3. Prevention beats detection, and it is not close
A prevention control reduces the chance of the cause occurring. A detection control finds the problem afterwards. Prevention is cheaper in every dimension: no scrap, no rework, no containment, no customer notification. When an FMEA action is chosen, prefer the one that changes occurrence over the one that improves detection — even when improving detection is easier, which it usually is.
4. Build the three process documents on one operation list
Process flow, PFMEA and control plan describe one process. If they are built on one operation list they cannot disagree. If they are three files they will, and the disagreement will be found by an auditor. This is a structural choice, not a matter of diligence.
5. Make revisions cheap
Here is the uncomfortable part. If updating a document set takes two days, people avoid updating it. They make the change on the line, promise to update the paperwork, and then the next urgent thing arrives. The documents are now wrong, and the organisation has quietly learned that the quality system is overhead rather than a tool.
Nobody solves this with more discipline. You solve it by making the update take ten minutes.
6. Record the reason, not just the change
A revision history saying "updated control plan" is nearly worthless. One saying "increased inspection frequency at Op 20 after tool wear caused three oversize bores in July" is a genuine organisational asset. Two years on it is the difference between understanding a decision and reversing it by accident.
7. Treat capability as a gate, not a report
Capability studies are often produced because a submission requires them, then filed. Their real use is deciding whether a process is ready — whether pre-launch inspection frequency can be relaxed, or whether a characteristic needs more control before volume. Establish statistical control first, then measure capability; the reverse order produces numbers that describe an unstable process and predict nothing.
Two things quality systems commonly get backwards
Writing the FMEA after the control plan. It happens under deadline: the control plan is what the shop needs, so it gets written, and the FMEA is reverse-engineered to justify it. The analysis then contains no thinking — it is a rationalisation. If the FMEA is a formality, the organisation has lost the only step in the process that was supposed to find problems before they occurred.
Treating audit preparation as a project. If evidence has to be assembled for each audit, the records were not organised in the first place. Evidence should be retrievable by scope — this customer, this part, this process, this date range — as a query rather than an expedition. When that is true, audit preparation stops being an event.