Why SPC data cannot answer an auditor
A Cpk number in a spreadsheet is not process capability evidence. Without the characteristic, the specification limits, the sample window and the revision behind it, the number cannot answer the question an auditor actually asks.
An auditor stops at one line on the control plan, a bore diameter with a tight tolerance, and asks a plain question. Show me the process capability for this dimension over the last quarter. Now the hunt starts. Someone remembers the SPC lives in a spreadsheet on the machine at cell four, except that operator is on nights. Someone else opens the standalone SPC package that only three people have a login for. A third person produces a capability study, and it turns out to be the one run at part approval eighteen months ago and never touched since. Three sources, three answers, and the clock is running while the auditor watches you look for your own data.
A Cpk of 1.4 answers nothing on its own
A capability index is a relationship, not a headline number. Cpk of 1.4 only means something once you can say which control plan characteristic it belongs to, which specification limits it was measured against, which part revision was in force at the time, how many samples went into it, and over what window they were taken. Strip that away and 1.4 is a rumour. Change the revision and the tolerance moves, so the same measurements give a different Cp and Cpk. Shorten the sample window and the number swings. The test an auditor is really applying is whether the same characteristic, fed the same samples against the same spec limits, comes out to the same figure today that it did last quarter. If your SPC cannot reproduce its own number, the number was never evidence.
What breaks when SPC lives apart from capture
- Readings scatter into per-machine spreadsheets and a standalone SPC app, none of them tied back to the released control plan, so nothing proves which characteristic or revision a column of numbers belongs to.
- The capability study was run once at part approval and never recomputed, so the Cpk on file describes a process that has drifted, worn tooling and all, since the day it was signed.
- Every site keeps its SPC its own way, in its own format, so a cross-site review turns into a translation exercise and the comparison stops being trustworthy.
- An out-of-spec reading shows up as a point on a chart and stops there, with no record of who caught it, what they did, or whether the part ever came back inside tolerance.
- Sample counts and windows differ from file to file, so two capability figures that look alike were never calculated on the same footing.
Capability that falls out of the capture itself
The fix is not a better spreadsheet. It is capturing the measurement against the released control plan characteristic in the first place, so the context is welded to the reading instead of remembered later. When an operator records a sample that way, the sample already carries its part, its revision, its operation, its specification limits and its method. Process capability then computes off that same stream, Cp and Cpk against the correct characteristic and the correct spec limits, with the sample count and window sitting underneath as plain fact rather than a claim. There is no separate study to keep in step, because the capability is a reading of the capture, not a document that ages next to it.
In VoraControl, operators capture against a released control plan characteristic, so every sample carries part, revision, operation, specification and method. VoraControl trends those measurements against the specification limits and computes Cp and Cpk directly on the capture stream, tied to the right characteristic. When a reading falls out of specification it hands off to Resolve, so the response lives with the signal that raised it instead of dying on a chart. And when an auditor asks for capability on one characteristic, VoraAudit surfaces the live Cp and Cpk for it to review on screen or print, drawn from the same samples the floor recorded. A characteristic that has too few samples yet is shown with that status stated plainly, not padded and not hidden, because an honest gap answers an auditor better than a borrowed number.
Released control plans, inspection capture and SPC trends in one governed flow.
Talk to someone who understands manufacturing control and audit evidence. We do not do generic demos.