Control limits vs. Specification limits: What to spot in supplier SPC charts

As a supplier quality engineer, there is always a constant trade-off between auditing speed and analytical rigor. Skimming static SPC chart images can leave critical issues undetected, but manually recalculating control limits and auditing raw subgroup data can drain hours you simply don’t have. 

This guide provides a rapid framework for telling specification limits and control limits apart, helping you catch what a quick scan can miss. 

Difference that matters: specification limits and control limits 

Every rigorous PPAP audit requires you to separate these two mathematically independent boundaries: 

• Specification limits (voice of the customer): set externally by design engineering and customer requirements; they define the allowable tolerance spread for a component. Establishing what the product must do to be functional and acceptable. 
• Control limits (voice of the process): calculated internally from actual production data, they represent the natural variation of a process under common-cause conditions. It defines what the manufacturing line is actually doing. 

It is important to note that specification limits in no way influence control limits. Dr. Donald Wheeler’s Four Process States frames this as well: a process can be in full statistical control while producing out-of-spec scrap (the threshold state) or out-of-control while temporarily staying within customer tolerance (the brink of chaos).  

The most common PPAP mistake comes from manually setting control limits in order to match specification limits instead of being calculated from the process. Making the chart look more capable than the process truly is. Any chart like this should go back for recalculation before you sign off. 

3 red flags to spot when auditing specification limits and control limits

During PPAP audits, you must look beyond overall pass/fail results and examine these three patterns in the raw subgroup data: 

1. Miscalculated limits

Control limits calculated from the total sample standard deviation of the overall dataset rather than within-subgroup variation come out wider than they should, masking instability. 

• How to catch it: Recalculate limits using subgroup ranges or subgroup standard deviations. If reported control limits are wider than the within-subgroup spread warrants, the chart understates variability. 

2. Over-adjustment and inconsistent entries 

Continuous machine tweaks or inconsistent manual entries distort distribution’s natural shape. 

• How to catch it: Unnaturally tight clustering, points sitting closer to the mean that the process allows, usually means rounded or incomplete data. Also, be on the lookout for sawtooth patterns, cyclic swings that point to an operator adjusting machine offsets after each reading. 

3. Unflagged run-rule violations

No point crossing the control limits doesn’t mean the process is stable, specifically if multiple consecutive points fall on one side of the center line. 

• How to catch it: Apply a recognized rule set, such as the Western Electric or Nelson rules, and look for sustained runs. 

Framework selection matrix: Quick engineering reference

To streamline reviews, you can use this reference matrix to quickly identify data anomalies and take immediate corrective action:

Stop trusting static chart screenshots

A clean-looking chart image can still hide a process that isn’t really in control. Validating the raw data behind it, not just the picture, is what protects your line. 

Ready to eliminate auditing fatigue and secure total visibility across your supply chain? Schedule a demo with Kiuey today to automate statistical verification across all your supplier submissions.