Measure capability from samples or summary statistics. Review centering, spread, yield, sigma, and risk instantly. Turn specification data into actionable improvement insights for teams.
Use raw measurements or summary statistics to estimate Cp, Cpk, Pp, Ppk, Cpm, yield, sigma level, and expected nonconformance.
This example represents five consecutive subgroups with four measurements each. It is useful for checking that your page calculations and subgroup summaries behave as expected.
| Subgroup | Reading 1 | Reading 2 | Reading 3 | Reading 4 | Subgroup Mean |
|---|---|---|---|---|---|
| 1 | 10.02 | 9.98 | 10.05 | 10.01 | 10.015 |
| 2 | 9.97 | 10.03 | 10.04 | 10.00 | 10.010 |
| 3 | 9.99 | 10.01 | 10.06 | 9.96 | 10.005 |
| 4 | 10.02 | 10.01 | 10.03 | 9.98 | 10.010 |
| 5 | 10.00 | 9.97 | 10.04 | 10.02 | 10.008 |
Capability indices compare specification width with process variation and centering. The page computes within and overall capability separately so you can distinguish short-term performance from long-term performance.
Cp = (USL - LSL) / (6 × σwithin) Cpk = min[(USL - Mean) / (3 × σwithin), (Mean - LSL) / (3 × σwithin)] Pp = (USL - LSL) / (6 × σoverall) Ppk = min[(USL - Mean) / (3 × σoverall), (Mean - LSL) / (3 × σoverall)] Cpm = (USL - LSL) / [6 × √(σoverall² + (Mean - Target)²)] Yield = Φ((USL - Mean) / σoverall) - Φ((LSL - Mean) / σoverall)In raw-data mode, within sigma is estimated by pooled subgroup standard deviations when complete subgroups are available. If subgrouping is not practical, the page falls back to the average moving range method.
1. Choose raw measurements when you have actual observations, or summary statistics when mean and sigma are already known.
2. Enter LSL and USL carefully. Add a target if your process has a nominal center that matters for Taguchi-style loss thinking.
3. In raw-data mode, provide subgroup size if the data were collected in rational subgroups. Consecutive values will be grouped automatically.
4. Click calculate. Review Cp, Cpk, Pp, Ppk, yield, and the chart. Export results to CSV or PDF for reporting.
It measures how well a stable process fits customer or engineering limits. The indices compare process variation and centering against the specification window.
Cp measures potential capability using spread only. Cpk also accounts for how centered the mean is between the limits, so it drops when the process drifts.
Cpk uses within-process variation, which reflects short-term capability. Ppk uses overall variation, so it captures longer-term performance, drift, and instability.
Many teams treat 1.33 as capable and 1.67 as strong, but the right target depends on risk, cost, compliance needs, and customer requirements.
That pattern means the process spread is tight enough, but the average is not centered. Shifting the mean toward target can improve Cpk without changing variation.
Use raw measurements when possible, because they allow subgroup estimation, previews, and charting. Use summary mode when your study already produced validated means and sigma values.
Yes. Yield, ppm, and the plotted curve rely on a normal approximation. For highly skewed or non-normal data, transform the data or use non-normal capability methods.
You can still interpret the relevant one-sided index, such as Cpu or Cpl. This page asks for both limits so the full two-sided report remains consistent.
Important Note: All the Calculators listed in this site are for educational purpose only and we do not guarentee the accuracy of results. Please do consult with other sources as well.