Statistical Process Control in Practice: A Realistic Scenario
Statistical Process Control (SPC) is one of the most practical tools in the Lean Six Sigma toolkit. It does not ask you to guess at whether a process is performing well -- it tells you, mathematically, whether the variation you are seeing is normal (common cause) or a signal that something specific has changed (special cause). Here is a realistic scenario showing how SPC was applied at a city permits department and what they found.
The Problem: Unpredictable Processing Times
The Development Services department at a mid-sized Canadian city had been receiving complaints from applicants about permit processing times. The department manager knew that processing times varied, but could not tell whether the variation was random or whether something specific was driving it. Some weeks, applications were turned around in three business days. Other weeks, it stretched to nine or ten. Applicants planning construction projects found it impossible to schedule contractors because they could not predict when permits would be issued.
The manager enlisted a process improvement advisor to help. The first step was to collect data. The team pulled the processing time -- from application received to permit issued -- for every permit processed over the previous 16 weeks: 288 individual records, grouped by week and by intake officer.
Building the X-bar and R Chart
The advisor constructed an X-bar and R chart (also called an average-and-range chart). The X-bar chart plots the average processing time per subgroup (in this case, per week) and allows the team to see whether the process average is stable over time. The R chart plots the range within each subgroup, showing whether the variation within a week is consistent. Control limits were calculated from the data -- not from targets or specifications, but from the actual process behaviour.
The chart told a clear story. For 13 of the 16 weeks, both the average and the range fell within the calculated control limits. The process was in statistical control during those weeks -- variation was predictable and consistent with a common-cause system. But three weeks showed points outside the control limits, and those three weeks also showed elevated range values, meaning not only were the averages high, but the variation within those weeks was unusually large.
Finding the Special Cause
The team stratified the data by intake officer. The pattern became immediately obvious. The three out-of-control weeks coincided with a week when one specific intake officer -- a newer employee hired eight months earlier -- was processing the majority of applications. Her individual processing times were not just slower on average; they were erratic, ranging from four days to fourteen days on similar application types.
The root cause turned out to be straightforward: this officer had never received training on the city's new permit management software, which had been rolled out seven months earlier. The other five intake officers had attended a two-day training session during the rollout. She had been on sick leave during that session and had never been rescheduled. She had been learning the system on her own, which explained both her slower times and the inconsistency -- some application types she had figured out, others she had not.
This is a textbook example of a special cause: a specific, identifiable, assignable factor (lack of training) that produced a specific, detectable pattern in the data. It had nothing to do with the process design itself. Addressing the common-cause system would not have fixed it. Only addressing the specific cause could.
The Fix and the Follow-Up Chart
The solution was simple: the officer completed a three-day accelerated training programme with a colleague who had used the system since rollout. The advisor re-ran the X-bar and R chart for the following eight weeks. All points fell within the control limits. The average processing time across the department settled at 4.2 business days, with a predictable range. Applicants received more reliable estimates. Complaints dropped.
Lessons learned: SPC does not tell you what the problem is -- it tells you there is a problem and where to look. The chart pointed to three specific weeks and a specific officer. The investigation found the root cause.
Separate common cause from special cause before intervening. If the manager had responded to the high weeks by pressuring the whole team to go faster, she would have introduced more variation, not less.
Control charts require ongoing use. A one-time chart is a diagnostic. A sustained chart is a management system. The department kept the X-bar and R chart running as part of their monthly operations review.
The data you need usually exists. The 288 permit records were already in the system. The insight came from organising them into a chart, not from collecting new data.
XNM supports public-sector clients with Lean Six Sigma process improvement and statistical analysis. Reach out to XNM's strategic advisory team to discuss how SPC and other process improvement tools can reduce variation and improve service delivery in your organisation.