Upload a CSV, map your measured value and its spec limits, and get Cp, Cpk, Pp, Ppk, the defect rate in parts per million, the process sigma level, and an honest normality check. Free.
Free analyses run on up to 10,000 rows. Larger files are randomly sampled to that size — sign up to analyze your full dataset.
Running your capability study...
Sent to — Cp, Cpk, Pp and Ppk with a confidence interval, the distribution drawn against your spec limits, the defect rate in parts per million, the process sigma level, a normality check, the R code, and AI insights.
Analyze another fileThe analysis estimates the process spread twice. Within-process sigma is the short-term spread: the pooled within-subgroup standard deviation when a subgroup column is mapped, otherwise the average moving range between consecutive measurements divided by 1.128. Overall sigma is the total standard deviation across the whole record. Cp is the specification width divided by six within-process sigma, and Cpk replaces the width with the distance from the mean to the nearer limit divided by three sigma, so it penalises a process that runs off-centre. Pp and Ppk repeat both on overall sigma. The defect rate is reported twice, once as the share of measurements actually outside the limits and once as the tail area of a fitted normal curve, and the process sigma level is the normal deviate matching that modelled rate. Normality is tested with Shapiro-Wilk and with an Anderson-Darling statistic computed directly, and when either rejects, the report states plainly that the indices should not be trusted as stated.
Use it when you have a run of measurements of one characteristic and a specification it has to meet, and you need to know whether the process fits inside that window, how often it will not, and whether the problem is spread, centring, or drift between subgroups.
Not for judging stability — that is a control chart's job and should come first, since capability assumes a stable process. Not for pass/fail attribute data, where a proportions analysis applies instead. And not to be taken at face value when the normality test in the report fails.
Built for: Quality engineers, manufacturing and process engineers, and anyone who has to certify that output meets a tolerance
Typical data source: A spreadsheet or CSV with one measured value per part, sample or batch, plus the spec limits from the drawing or the process sheet
One row per measured part, with the specification limits carried as constant columns. For example, a machined shaft diameter:
Minimum 20 rows · Best with 100-5,000 measurements, ideally in subgroups of 3-10
Standard-library analysis: a full Cp / Cpk capability study of one measured characteristic against its specification limits. Computes the short-term indices (Cp, Cpk, CPU, CPL) from within-subgroup variation and the long-term indices (Pp, Ppk, Cpm) from total variation, estimates the defect rate in parts per million both as observed and under the normal model, reports the process sigma level, draws the distribution against the spec limits, and tests normality — because the indices assume a normal distribution and mislead badly when it fails. Works on any dataset: map the measured column and supply the limits.
Your measurements drawn against the spec limits — the picture the single Cpk number is summarising, and where the headroom actually is.
Every capability index side by side with a confidence interval on Cpk, so you can see whether the shortfall is spread or centring.
How many parts per million fall outside the limits, both as counted and as the normal model predicts, plus the process sigma level.
The assumption the indices stand on, tested rather than assumed — with a plain statement when the indices should not be trusted.
Short-term spread versus drift between subgroups — how much capability is recoverable without touching the machine.
Plain-English interpretation — what the numbers mean, what's significant, and what to do next.
Is this process capable of holding the tolerance?
Map your measured characteristic and its spec limits. You get Cp, Cpk, Pp and Ppk with a confidence interval on Cpk, the distribution drawn against the limits, the defect rate in parts per million both observed and modelled, and an explicit normality test so you know whether those indices can be trusted at all.
See our FAQ for details on pricing, data privacy, and how the analysis works. Every report includes a Methodology section showing the statistical test, assumptions checked, and diagnostics run.
Run any analysis on your own data — validated R analyses, interactive reports, AI insights, and PDF export.
Try Free — No Credit CardTell us what went wrong, in your own words. We capture the page you're on automatically, so no need to describe where you are.