New to this topic?
We recommend reading these guides first to get the most out of this one:
Six SigmaStatistical Process Control
Cp
Can It Fit?
Cpk
Does It Fit?
1.33
Minimum Acceptable
1.67+
World Class

What Is Process Capability?

Process capability measures whether your process can consistently produce output within customer specifications. It compares the voice of the process (natural variation) to the voice of the customer (specification limits). A capable process has variation that comfortably fits within the spec range.

This is the foundation of Six Sigma quality. If your process is not capable, no amount of inspection will fix it — you need to reduce variation or change the process.

Cp vs. Cpk: What They Mean

Cp (Capability Potential)

Can the process fit within the spec? Cp measures the width of the spec range relative to the width of the process spread (6 standard deviations). It assumes the process is perfectly centered.

Cp Formula

Cp = (USL – LSL) ÷ (6 × σ)

Where USL = Upper Spec Limit, LSL = Lower Spec Limit, σ = process standard deviation. A Cp of 1.0 means the process spread exactly equals the spec width — barely fits. A Cp of 2.0 means the spec is twice as wide as the process — comfortable room.

Cpk (Capability Performance)

Does the process actually fit? Cpk accounts for centering — how close the process mean is to the center of the spec. A process can have high Cp (narrow spread) but low Cpk (off-center, producing defects near one spec limit).

Cpk Formula

Cpk = min[(USL – μ) ÷ (3σ), (μ – LSL) ÷ (3σ)]

Where μ = process mean. Cpk takes the smaller of the two distances (mean to nearest spec limit) and divides by half the process spread. If Cpk = Cp, the process is perfectly centered. If Cpk < Cp, the process is off-center.

Cp is the same. Cpk is not. The same process, aimed 0.15 mm high: identical spread, identical Cp of 1.33 — and 99 times the defect rate. Centring is a machine setting, not a project.

Both curves are a normal distribution with σ = 0.1 against a specification of 9.60 to 10.40, so 6σ = 0.60 in a 0.80 window and Cp = 1.33 for each of them. Centred on 10.00 the process is symmetric in the window, Cpk = 1.33, and about 63 parts per million fall outside — the page's "Good" band. Aimed at 10.15, one and a half standard deviations high, the nearer limit is only 2.5σ away, Cpk falls to 0.83 and roughly 6,210 parts per million fall outside — the page's "Poor" band. Nothing about the variation changed. The tail areas are drawn to scale, which is why only the shifted curve has a visible one.

Interpreting the Numbers

Cpk ValueRatingApprox. DPMOSigma LevelWhat It Means
Below 0.67Not capable45,500+< 3σProcess cannot meet spec. Major redesign or variation reduction needed.
0.67 – 1.00Poor2,700 – 45,5003-4σSignificant defects. Process improvement urgent.
1.00 – 1.33Marginal63 – 2,7004-4.5σProducing some defects. Improvement needed for critical features.
1.33 – 1.67Good0.6 – 634.5-5σIndustry standard minimum for most manufacturing. Acceptable.
1.67+Excellent< 0.65+σVery capable. Variation well within limits. World class.
2.00Six Sigma3.46σSix Sigma target. Near-zero defects.

Visual: Capable vs. Not Capable

Cpk = 1.5 (Capable)
LSL
USL
Room to spare ✅
Cpk = 0.7 (Not Capable)
LSL
USL
Spilling over limits ❌
The bell curve (process spread) must fit comfortably within the spec limits (LSL to USL)

How to Improve Capability

Center the process (fix Cpk first)If Cp is good but Cpk is low, the process is off-center. Adjust the process mean: change the machine setting, calibrate the tool, adjust the recipe. This is often a quick win.
Reduce variation (improve Cp)If Cp itself is low, the process is too variable. Identify sources of variation using fishbone diagrams: material lot-to-lot, machine wear, operator technique, environmental factors. Then attack the biggest sources.
StandardizeStandard work reduces operator-to-operator variation. TPM reduces machine-to-machine variation. Incoming material specs reduce lot-to-lot variation. Standardize everything that affects the output.
Error-proofPoka-yoke devices prevent the conditions that cause variation. Fixtures that enforce position, sensors that verify settings, interlocks that prevent wrong material.
Monitor with SPCOnce capable, use Statistical Process Control charts to monitor the process over time. Detect shifts and trends before they cause defects. React to special cause variation immediately.
✅ Capability-Driven Quality
  • Cpk measured for every critical feature
  • Process centered and variation reduced before production
  • SPC charts monitored during production
  • Capability studies required for new products/processes
❌ Inspection-Driven Quality
  • Cpk unknown or never measured
  • Rely on 100% inspection to sort good from bad
  • No SPC — process drifts until defects appear
  • New processes launched without capability study

🎯 Key Takeaway

Process capability tells you whether your process can meet spec consistently, not just whether today's parts are good. A Cpk of 1.33+ means you have breathing room; below 1.0 means you are producing defects by design. Measure capability for every critical feature, center the process, reduce variation, and monitor with SPC. It is the difference between hoping for quality and engineering it. Use the DPMO/Sigma calculator to connect capability to defect rates.

Interactive Demo

Adjust the process mean, standard deviation, and specification limits to see how Cp and Cpk change. Watch the distribution shift relative to the spec window.

⚡
Try It Yourself
Process Capability Simulator
▼
Adjust the specification limits and process parameters. Watch Cp and Cpk change as you shift the mean or increase variation. Cpk < 1.0 means your process can't consistently meet spec.
Specification Limits
10.5
912
9.5
811
Process Parameters
10
8.511.5
0.15
0.050.5
1.11
Cp
1.11
Cpk
858
Total PPM
Marginal
Status
Cpk between 1.0–1.33: Marginal capability. Process may produce 858 PPM defects. Reduce variation or center the process.
LSLUSL
Ready for the full knowledge check? Test your understanding with guided scenarios and data export.
PROTake the Pro Knowledge Check →
Keep Learning
📖
Guide
Cost of Quality
🧮
Calculator
DPMO & Sigma Calculator
🧮
Calculator
First Pass Yield
🎓 Part of the Quality & Measurement track — view all 7 guides →
Free forever · Every feature included

Stop reading, start modeling

Model your process flow, run simulations, optimize staffing with TOC math, and test your knowledge with 107 interactive checks — all in one platform.

Open Workbench → Pro Knowledge Check

Take this to a room

The running order

For quality and process engineers. They should leave able to say whether a low Cpk on their line is a centring problem or a variation problem.

7 beats · 10 min
  1. 1

    Two questions, two numbers

    Cp asks can it fit. Cpk asks does it fit. The difference between them is where the process is sitting.

    • Cp is spec width over six sigma of process spread. It assumes perfect centring.
    • Cpk takes the worse of the two sides, so it punishes being off-centre.
    • Cp good and Cpk poor means the process is off-centre, not too variable.

    Ask the room Do we know Cp and Cpk for our critical features, or only whether today's parts passed?

  2. 2

    What the numbers mean in defects

    Capability is abstract until you put it on the same axis as scrap.

    • Below 0.67: not capable. 45,500 defects per million or worse.
    • 1.00 to 1.33: marginal - producing defects, needs work on critical features.
    • 1.33 to 1.67: the industry minimum most customers expect.
    • 1.67 plus: very capable, variation well inside the limits.
  3. 3

    Below 1.0 is by design

    This is the sentence that changes how people hear the number.

    • A Cpk under 1.0 means the process is producing defects by design, not by accident.
    • No amount of inspection changes that. It only sorts the output.
    • Today's parts being good is not evidence of capability.
  4. 4

    Centre first - it is the quick win

    If Cp is fine and Cpk is not, you are not fighting variation. You are fighting an offset.

    • Adjust the mean: machine setting, tool calibration, recipe.
    • Often an afternoon of work for a large Cpk gain.
    • Do this before anyone proposes buying equipment.
  5. 5

    Then reduce variation

    If Cp itself is low, the spread is the problem and you have to find where it comes from.

    • Fishbone the sources: material lot to lot, machine wear, operator technique, environment.
    • Attack the biggest source first, not all of them.
    • Standard work removes operator-to-operator spread. TPM removes machine drift.
  6. 6

    Then hold it

    Capability is a snapshot. SPC is what keeps it true next month.

    • Once capable, chart it and watch for shifts and trends.
    • React to special causes immediately, leave common cause alone.
    • Recalculate capability after any real process change.
  7. 7

    What we do next

    Pick the critical features and run a study on each. Require one for every new process.

    • List the three to five features that drive complaints or scrap.
    • Run a capability study on each - centre, then reduce.
    • Make a capability study a launch requirement for new products.

    Ask the room Which feature do we start with, and what is its current Cpk?

Related Guides

📊 Six Sigma
The quality methodology
🔢 DPMO Calculator
Convert capability to defect rates
🛡️ Poka-Yoke
Reduce variation at the source