Statistical Process Control (SPC): A Practical Guide to Reducing Variation

Leave a comment

Statistical Process Control, almost always abbreviated to SPC, is the systematic use of statistical methods to monitor and control process behaviour. It is older than Six Sigma by several decades and underpins almost every modern quality system, from automotive manufacturing to pharmaceutical production to clinical laboratories. If you cannot tell whether your process is stable, you cannot meaningfully discuss whether it is capable, and SPC is the discipline that gives you that answer.

This article explains what SPC is, where it came from, how to deploy it inside a Lean Six Sigma project, and the most common pitfalls we see in real-world implementations at ILSSI partner organisations.

Origins and Why SPC Still Matters

SPC traces back to Walter Shewhart at Bell Laboratories in the 1920s. Shewhart’s insight, later expanded by W. Edwards Deming, was that all processes show variation, and that variation falls into two categories. Common cause variation is the inherent random behaviour of a stable process. Special cause variation is the result of a specific, identifiable disturbance: a tool that has shifted, a material lot that is out of spec, an operator who was poorly trained on the new procedure.

The point of SPC is to tell these two types of variation apart. A stable process should only be improved through systemic redesign, not through adjustment of individual outputs. An unstable process should be investigated for special causes before any redesign is attempted. Confusing the two leads to the management behaviour Deming called ‘tampering’, which makes processes worse rather than better.

The Core Tool: The Control Chart

Every SPC implementation revolves around the control chart. The chart plots a process metric over time, with control limits calculated from the process itself rather than from customer specifications. Points within the limits, with no unusual patterns, indicate a stable (in-control) process. Points outside the limits, or showing one of the recognised special cause patterns, indicate that something has changed.

The choice of which control chart to use depends on the data type and sampling scheme. The most common chart in service industries is the I-MR (individuals and moving range) chart, because data often arrives one observation at a time. In manufacturing with rational subgroups, the Xbar-R chart is standard. For attribute data, the p, np, c, and u charts apply, depending on whether you are tracking defectives or defects, and whether the area of opportunity is constant.

For a complementary deep dive on the underlying ideas, the ASQ control chart resource offers a thorough introduction with worked examples.

Deploying SPC Inside DMAIC

SPC contributes to every phase of a DMAIC project, but its weight shifts as the project moves through the cycle.

In Measure

Build a baseline control chart of the current process. This serves two purposes: it confirms whether the process is stable enough to characterise meaningfully, and it produces the before-and-after comparison you will need at the end of the project. If the chart is wildly out of control, the early focus may shift to special cause elimination before any process redesign is even attempted.

In Analyse

Use SPC to test hypotheses about root causes. A change of supplier, a change of shift, a change of machine: each one is a potential special cause. By stratifying the chart on the suspected factor, the team can see whether the variation behaves differently between groups.

In Improve

Run pilot changes inside a control chart with stages, so the limits recalculate after the intervention. This makes the effect of the change visually obvious and resistant to dispute.

In Control

The control chart from the Measure phase becomes the long-term monitoring tool, often handed over to the process owner along with a written control plan. Modern manufacturing execution systems automate this monitoring; in service settings, a daily review of a printed or digital chart is often sufficient.

Common Cause vs Special Cause: A Practical Test

If you remember nothing else from this article, remember this. When a point falls outside the control limits, investigate. When a point falls inside the control limits, do not adjust the process. The former action is appropriate; the latter is tampering and will increase variation.

The Western Electric rules, embedded in Minitab and most other SPC software, extend the basic out-of-limits test with seven additional patterns that signal special cause variation: nine points on the same side of the centre line, six consecutive points trending up or down, fourteen points alternating, two of three points beyond two sigma on the same side, four of five points beyond one sigma on the same side, fifteen points within one sigma of the centre line (over-control), and eight points beyond one sigma on either side. Modern teams usually enable the most important three or four of these and review the others case by case.

SPC in Service and Transactional Processes

A persistent myth is that SPC only applies to manufacturing. This is incorrect. Loan processing times, call centre handle times, invoice error rates, surgery turnaround times, and patient wait times all behave statistically and benefit from SPC. The challenges in service settings are different: the data is often discrete and binary, the volumes per category may be lower, and the cultural resistance to ‘industrial’ statistical methods can be stronger.

The way around the cultural resistance is the same in every sector: lead with the chart, not the theory. Once a team sees a real signal that they recognise from their daily experience, the philosophical objections evaporate.

For sector-specific case studies, the ILSSI research papers archive includes peer-reviewed applications of SPC in healthcare, finance, and public services.

Process Capability: The Next Step

SPC tells you whether a process is stable. Process capability analysis, expressed through indices like Cp, Cpk, Pp, and Ppk, tells you whether the stable process is actually good enough to meet customer requirements. The two are sequenced: stabilise first, then assess capability, because capability indices calculated on an unstable process are meaningless. We cover capability in detail in a separate article in this series.

Tools and Software

Minitab remains the dominant tool in formal Six Sigma deployments because it implements all the standard chart types correctly and produces output the certification bodies recognise. JMP, Statgraphics, and SigmaXL serve similar functions. For organisations on a budget, Excel can produce respectable control charts with a little setup, though the discipline of doing it correctly is harder to enforce without dedicated software.

Practitioners working towards Black Belt certification will use these tools intensively. The ILSSI Black Belt programme covers advanced SPC, including EWMA and CUSUM charts for time-correlated data, multivariate control charts, and the design of statistical control plans.

Common Mistakes in SPC Deployment

  • Calculating limits on insufficient data. Twenty to twenty-five subgroups is the conventional minimum.
  • Using specification limits as control limits. They are conceptually unrelated and the comparison is misleading.
  • Treating control charts as a reporting exercise rather than a daily management tool.
  • Failing to recalculate limits when the process has been deliberately and successfully changed.
  • Allowing operators to adjust the process based on individual readings inside the limits.

Final Thoughts

SPC is one of those topics where the technical material is straightforward but the cultural change required to apply it well takes years. The organisations that succeed are the ones that build SPC into the daily rhythm of work, not the ones that produce monthly reports nobody reads. The practitioner’s job is to make the charts visible, the language consistent, and the response disciplined.

To begin your formal SPC training within an accredited Lean Six Sigma framework, explore the ILSSI Green Belt certification or contact info@ilssi.org for partner referrals in your region.