Your supplier sends over a capability report. It shows a Cpk of 1.71 across all critical dimensions. Is that good? Should you approve the parts? Does it tell you anything about how the process will behave six months into production?

These are questions that engineers and product teams face regularly, yet the statistics on a capability report are rarely explained in plain terms. Cpk and Ppk are among the most referenced quality metrics in manufacturing, but also among the most misread. A number alone never tells the full story — the context around it determines whether a process is genuinely ready for production or whether it just looked good on a test run.

This guide breaks down what Cpk and Ppk actually measure, how to read them on a supplier report, what benchmarks apply across different industries, and what questions to ask when something doesn't add up. Whether you're evaluating a supplier for plastic injection molding, CNC machining, or any other production process, understanding capability data is a core part of making confident sourcing decisions.

What Is Process Capability?

Process capability is a statistical measure of how consistently a manufacturing process produces parts within the specified tolerance limits. It answers a fundamental question: given the natural variation in this process, how likely is it that every part will meet specification?

Every manufacturing process has variation. A CNC spindle doesn't stop in exactly the same position twice. An injection mold heats and cools slightly differently cycle to cycle. A die casting shot changes subtly as the die warms up over a production run. Capability indices quantify that variation relative to engineering tolerances, giving engineers a numerical basis for evaluating whether a process is fit for production — or whether it's operating too close to the edge of the tolerance band to be reliable.

The key thing to understand is that capability indices are not quality scores. They are statistical descriptions of process behavior. A high Cpk doesn't mean every part is perfect; it means the process, as measured, has enough margin that defects are statistically unlikely. That distinction becomes critical when you're deciding whether to sign off on a supplier's production readiness.

The Four Indices: Cp, Cpk, Pp, and Ppk

Most capability reports include four related indices: Cp, Cpk, Pp, and Ppk. They work as a set. Understanding how they relate to each other tells you far more than any single number on its own.

Cp: Potential Capability

Cp compares the width of the engineering tolerance (the distance between the upper and lower specification limits) to the natural spread of the process (six standard deviations of variation). A Cp of 1.0 means the process spread exactly fills the tolerance band. A Cp of 1.33 means the process spread occupies about 75% of the tolerance, leaving some margin.

The critical limitation of Cp is that it assumes the process is perfectly centered between the specification limits. In practice, processes drift. A CNC lathe might produce parts that consistently measure toward the upper end of a tolerance band. Cp won't capture that — it only looks at spread, not position. This is why Cp is rarely used alone as a production acceptance criterion. It tells you whether the process could be capable, not whether it actually is.

Cpk: Actual Capability with Centering

Cpk builds on Cp by also measuring how well the process is centered between the specification limits. Rather than looking at the full tolerance width, Cpk calculates the distance from the process mean to the nearest specification limit, then divides by three standard deviations. By always using the nearest limit, Cpk automatically reflects the worst-case risk of producing out-of-spec parts.

This makes Cpk the most practical short-term capability index. If a process drifts toward one specification limit, Cpk drops — even if Cp stays constant. For example, in pressure die casting, a critical bore dimension might have a tight tolerance. If the core pin wears slightly and dimensions shift toward the upper limit, Cpk will decline before any parts are actually out of specification, giving you early warning. It uses within-subgroup variation, meaning it reflects the process behavior during a single controlled production run rather than across weeks of manufacturing.

Pp and Ppk: Long-Term Performance

Pp and Ppk follow the same mathematical structure as Cp and Cpk, but they use the overall standard deviation calculated from the entire production dataset rather than within-subgroup variation. This is the crucial difference. Because overall variation is always equal to or greater than short-term variation, Ppk will typically be equal to or lower than Cpk for the same process.

Where Cpk describes a process under controlled, stable conditions, Ppk describes how that process actually performed across real production — including tool wear, material batch changes, operator handoffs, machine warm-up cycles, shift changes, and minor process adjustments. For this reason, Ppk is often the index used during production validation and supplier qualification: it reflects what customers will actually experience, not just what the process looked like during an optimized test run.

Cpk vs. Ppk: The Difference That Matters

In plain language: Cpk measures what the process is capable of. Ppk measures what the process has actually delivered. Both account for centering. The only mathematical distinction is which standard deviation is used — within-subgroup for Cpk, and overall for Ppk.

That single change in calculation produces a fundamentally different engineering question. Cpk asks: "When this process is running smoothly and under statistical control, how capable is it?" Ppk asks: "Across real production conditions over time, how has this process actually performed?" A process can have an excellent Cpk and a disappointing Ppk — which is a meaningful finding, not a contradiction.

Consider an injection molding process producing a critical sealing surface. During process validation, a supplier measures 50 consecutive parts immediately after mold optimization and reports a Cpk of 1.74. Weeks later, after running across multiple shifts with different resin lots, minor press adjustments, and routine tool maintenance, a follow-up study reports Cpk of 1.71 but Ppk of 1.34. The underlying process capability is largely unchanged — but routine production introduces more variation than the controlled validation run captured. Neither number alone tells the full story; their relationship does.

Reading the Gap Between Cpk and Ppk

When you receive a capability report, the relationship between Cpk and Ppk is one of the most informative things you can examine. A small gap between the two values suggests a stable, well-controlled process. A large gap raises questions worth asking before production approval.

A significant gap doesn't automatically disqualify a supplier. It's a signal to dig deeper. Common causes include tool wear exceeding planned limits, inconsistencies between mold cavities in a multi-cavity tool, operator-to-operator variation, material batch differences, or inadequate preventive maintenance intervals. In some cases the explanation is benign; in others, corrective action is warranted before production launch.

Here's a practical illustration. Suppose two suppliers submit capability reports for the same part:

Supplier B appears stronger based on Cpk alone. But Supplier A's tightly aligned Cpk and Ppk indicate a process that maintains its capability under real production conditions. Supplier B's capability study may have been conducted under highly controlled conditions that don't reflect day-to-day manufacturing. For production risk, Supplier A may actually be the safer choice.

What Do the Numbers Mean? Benchmark Values

Capability indices have no inherent pass/fail threshold — the right target depends on the application, the feature's criticality, and the customer's requirements. That said, widely used benchmarks exist across manufacturing industries and provide a useful starting reference.

A Cpk of 1.33 has become the widely-used standard benchmark because it provides enough margin for the process to absorb normal variation without generating defects — assuming the process is stable, centered, and measured with a capable inspection system. It corresponds roughly to a 4-sigma separation between the process mean and the nearest specification limit, which translates to a very low defect rate under normal operating conditions.

What these benchmarks don't capture is the context in which the data was collected. A Cpk of 2.05 based on 30 parts measured immediately after machine setup is far weaker evidence than a Cpk of 1.56 based on 300 consecutive production parts measured across three shifts. Always look at sample size, production conditions, and the timeframe of the study alongside the index value itself.

Capability Expectations by Manufacturing Process

Capability targets aren't one-size-fits-all, and they vary meaningfully across manufacturing processes. Different processes have different inherent levels of variation, and the industry context of the part drives the required threshold.

• CNC Machining: CNC processes typically achieve high capability values on well-maintained equipment. Cpk values of 1.67 or higher are achievable on most critical dimensions. Key variables to watch are tool wear, fixture repeatability, and thermal drift over long runs.
• Plastic Injection Molding: A Cpk of 1.33 is the general production floor for most injection molded features. For critical-to-function dimensions in automotive or medical programs, 1.67 or higher is commonly specified at the PPAP stage. Multi-cavity tools require per-cavity capability data — a blended average can mask individual cavity drift.
• Pressure Die Casting: Die casting involves more inherent variation than machining due to thermal cycling, metal flow dynamics, and die wear. Achievable Cpk values depend heavily on the feature location and tolerance band. Critical machined surfaces on cast parts may be assessed separately after secondary operations.
• Sheet Metal Fabrication: Formed features like bend angles and flanges are more susceptible to springback variation. Capability targets should reflect the tolerance achievable by the process, and overly tight tolerances on formed features may require secondary sizing or restrike operations to achieve acceptable Cpk.
• Liquid Silicone Rubber (LSR) Molding: LSR parts introduce measurement challenges due to material flexibility. Capability data is still meaningful, but the measurement system and fixturing must be validated to ensure the measurement itself doesn't introduce significant variation.

The process context is important not just for setting targets, but for interpreting results. A Cpk of 1.40 on a tight-tolerance CNC feature may warrant investigation if you'd expect 1.67, whereas the same value on a complex cast geometry may represent excellent process control for that feature type.

Red Flags to Watch for in a Capability Report

Not all capability reports are created equal. A supplier can submit a statistically impressive-looking report that doesn't genuinely reflect production readiness. Knowing what to look for — and what to challenge — is an important part of supplier qualification.

• Small sample sizes: A capability study based on 25–30 parts measured immediately after setup tells you very little about production performance. Look for studies based on at least 100 consecutive production parts, ideally covering multiple shifts or batches.
• No control charts included: Capability indices assume a statistically stable process. Without control charts showing the process was in control during the study, a Cpk value has little predictive value. Stability must be confirmed before capability can be interpreted.
• Cpk and Ppk not both reported: A report that shows only Cpk, without Ppk, may be concealing long-term production variation. Both indices should be reported together for a meaningful picture.
• No measurement system validation: If the measurement equipment isn't appropriate for the tolerance, or if Gage R&R hasn't been completed, the capability values are unreliable. Measurement error inflates apparent process variation and can make a capable process look incapable — or mask a real problem.
• Study conditions don't match production: A capability study run on a freshly calibrated machine using hand-selected material, operated by an expert technician, may not reflect what production operators will achieve day to day. Ask whether the study was conducted under normal production conditions.
• All dimensions reported at exactly the same value: Suspiciously uniform Cpk values across all characteristics can indicate that data has been rounded, manipulated, or estimated rather than measured. Real production data shows natural variation across different features and dimensions.

Capability Is One Part of the Picture

A supplier who reports a Cpk of 1.67 across all critical dimensions has given you useful evidence — but not the whole story. Capability indices describe how a process was performing during the measurement period. They don't guarantee future performance, and they can't substitute for a complete quality system.

Capability data is most meaningful when it's part of a broader qualification package that includes process stability evidence (SPC charts), a validated measurement system (Gage R&R results), a control plan showing how the process is maintained, and documentation of how variation is monitored and responded to over time. In formal programs, this is what the Production Part Approval Process (PPAP) is designed to verify — capability is one element of that submission, alongside material certs, process FMEAs, and the control plan.

For teams moving from prototyping to volume manufacturing, capability planning should begin early. Tolerances that are feasible on a CNC-machined prototype may present genuine capability challenges when transferred to an injection molded production part. Processes like vacuum casting used in low-volume bridge production have different inherent capability characteristics than the injection mold tooling that replaces them at volume. Understanding those differences early — and designing tolerances that each process can reliably hold — is how you avoid capability problems at production launch.

Working with a manufacturing partner that understands capability across the full production lifecycle means tolerance and process decisions are made with real data behind them, not assumptions. From low volume production through high volume manufacturing, the goal is a process that doesn't just pass a capability study — it sustains that performance at scale.

Frequently Asked Questions

What is the difference between Cpk and Ppk?

Cpk measures short-term process capability using within-subgroup variation, reflecting how the process performs under stable, controlled conditions. Ppk measures long-term process performance using overall variation across the entire production dataset, capturing the effects of tool wear, material changes, and shift-to-shift differences. Cpk describes what the process can do; Ppk describes what it has done.

What is a good Cpk value?

For most production applications, a Cpk of 1.33 or above is considered capable. For critical characteristics in automotive, medical, or safety-related programs, 1.67 is commonly specified — and some programs require 2.00 or higher. The appropriate target depends on the feature's function, the industry standard, and the specific customer or program requirements.

Can Ppk be higher than Cpk?

In theory, yes, though it is uncommon. Because overall variation is typically equal to or greater than within-subgroup variation, Ppk is usually equal to or lower than Cpk. If Ppk appears higher than Cpk on a report, it's worth reviewing the study methodology, subgroup definitions, and calculation method before drawing conclusions.

Does PPAP require Cpk or Ppk?

PPAP submissions often require both, but the specific requirements are customer-defined. Many programs require Ppk at the initial submission stage (commonly Ppk ≥ 1.67) because it reflects observed production performance. Cpk is often used for ongoing monitoring once a process has stabilized in mass production. IATF 16949 automotive programs typically specify both metrics with defined thresholds.

What does it mean if Cpk is much higher than Ppk?

A large gap between Cpk and Ppk indicates that additional variation is entering the process during normal production that wasn't present during the controlled capability study. This could be caused by tool wear, material batch-to-batch variation, operator differences, or process drift over time. It doesn't automatically mean the process should be rejected, but it does warrant investigation before production approval.

Putting It All Together

Cpk and Ppk are not scores to be accepted or rejected in isolation — they are windows into how a manufacturing process behaves. Cpk tells you about potential, Ppk tells you about real-world performance, and the relationship between them tells you about process stability. Together with supporting context — sample size, study conditions, control charts, and measurement system validation — they give you a well-grounded basis for production decisions.

The right question when reading a capability report isn't simply "is this number high enough?" It's "what does this data tell me about how this process will perform once it's running at volume?" That shift in perspective turns capability analysis from a compliance checkbox into a practical engineering tool — and it leads to better supplier decisions and fewer surprises after launch.