MCManufacturing Compass

Cp and Cpk Calculator

Explore process capability and distribution using measurements and specification limits.

No signup · Calculated in your browserInput data is not sent to servers, analytics or external APIs.

The file contains your input data. It is not sent to a server. Results are not saved. Maximum file size: 2 MiB.

Input data

Showing example data
Input method

Low variation, with the mean shifted toward the upper limit.

Use one value per line, or comma/tab separators. Use a decimal point; omit thousands separators, headings and units.

40 measurements recognized.

For a one-sided specification, enter only the applicable limit.

Results

Example data
Ppk1.331

Approximately equal to the commonly used reference value of 1.33.

1.33 is a commonly used reference, not a universal acceptance rule. Follow your customer requirements and internal standards.

The mean is shifted toward the upper limit

Variation is relatively small, but the process mean is shifted from the midpoint toward the upper limit. Check equipment offsets, tool wear and initial settings.

Raw data uses the overall sample standard deviation (n−1), so these results are Pp and Ppk.

Distribution

Histogram of measurementsDistribution of 40 measurements. Mean 10.0525, standard deviation 0.0319.061319259.7734–9.8353: 0 measurements9.8353–9.8972: 0 measurements9.8972–9.9591: 0 measurements9.9591–10.0209: 8 measurements10.0209–10.0828: 25 measurements10.0828–10.1447: 7 measurements10.1447–10.2066: 0 measurementsLSL: 9.8LSLUSL: 10.18USLMean: 10.0525MeanSpecification midpoint: 9.99Specification midpoint9.773410.2066
MeasurementsSpecification limitsMeanSpecification midpoint
Select a bar to inspect its interval and count. Solid lines show specification limits, the dashed line shows the mean, and the dotted line shows the specification midpoint.
Additional statistics and checks

Additional statistics

Pp
1.984
Ppu
1.331
Ppl
2.636
Mean
10.052
Standard deviation
0.032
Measurements
40
LSL
9.800
USL
10.180
Specification midpoint
9.990
Offset from midpoint
0.062

What to check next

  • Equipment offsets
  • Tool wear
  • Initial settings and condition changes
  • Process stability over time
  • Measurement system
  • Sampling method

Understanding your results

Check which standard deviation and specification limits the calculation uses.

Cp and Cpk versus Pp and Ppk

Cp and Cpk use a within-process standard deviation, often estimated from variation within rational subgroups. Pp and Ppk use overall standard deviation. Here, raw data uses the sample standard deviation with denominator n−1. The tool does not infer subgroup structure or a time horizon from pasted values.

In the mean-and-standard-deviation mode, supply your own appropriate short-term estimate. The tool labels these results Cp and Cpk; it does not estimate the within-process standard deviation for you.

How the indices are calculated

Cp = (USL − LSL) / (6σ). Cpu = (USL − mean) / (3σ), Cpl = (mean − LSL) / (3σ), and Cpk is the smaller of Cpu and Cpl. For Pp, Ppu, Ppl and Ppk, this tool uses the overall sample standard deviation in the same expressions.

USL and LSL are specification limits, not control limits. With only one specification limit, the corresponding one-sided index is shown; a specification midpoint and a two-sided potential index are undefined.

Read 1.33 as a reference, not a pass/fail rule

The appropriate capability requirement depends on the product, process stage, customer and internal criteria. A displayed value cannot guarantee product quality. A negative Cpk or Ppk means the mean is beyond the specified limit; zero means the mean is on that limit.

Check stability and the measurement system separately

Examine time-ordered data with an appropriate control chart, and consider measurement variation, sampling, material lots and the shape of the distribution. Capability indices alone do not demonstrate statistical control or establish a cause for poor performance.

Do not infer an actual nonconforming rate from these indices without checking the distributional assumptions. The learning curves are theoretical normal distributions, not fitted evidence of normality in your measurements.

Using the calculator and learning mode

The calculator starts with a fixed illustrative dataset. Select another example, paste your own measurements, or supply summary statistics. Use a decimal point, separate values with newlines, commas or tabs, and omit thousands separators and units. Click Calculate after editing. Invalid tokens are counted; correct them before relying on the result.

The histogram is available only for raw data. Copy results to take the numerical summary and explanation into your notes. If clipboard access fails, select and copy the read-only text manually.

In Explore and learn, change the mean, standard deviation or limits while comparing with a saved baseline. Presets update the current conditions; Reset restores both sets of values. Moving a specification limit changes the evaluation, not the process itself.

Privacy and limitations

Calculations run in your browser. Measurements, specification values, results and clipboard text are not sent to a server, analytics service or external API, and are not saved by this tool. Reloading the page or switching language clears the working state.

Site analytics record fixed interaction categories, such as input mode, sample ID and language. This tool does not replace a validated capability study or your organization's decision process.

Frequently asked questions

Why does raw data produce Pp and Ppk?

The raw-data calculation uses the overall sample standard deviation of all supplied measurements. It does not estimate within-subgroup variation, so the output is labeled Pp and Ppk.

Can I use a one-sided specification?

Yes. Supply only USL for an upper-sided index (Ppu or Cpu), or only LSL for a lower-sided index (Ppl or Cpl). Pp and Cp require both limits and are not shown for a one-sided specification.

Does Cpk of 1.33 guarantee a good process?

No. 1.33 is a commonly used reference, not a guarantee of quality or stability. Consider customer requirements, internal standards, the measurement system, sampling and the process distribution.

Sources and edition dates

English edition updated . Translation date: 2026-09-21.

Reviewed by RYO on 2026-09-21. Published 2026-09-21.