Key takeaways
Short answer: In control and in spec sound similar but measure completely different things, and confusing them is a classic quality error. In control means the process is statistically stable and predictable — its variation is all common cause, staying within control limits derived from the process itself. In spec means the output meets the tolerance the customer specified. The two are independent: control limits describe how the process behaves, specification limits describe what the customer needs. A process can be one without the other. For the variation underneath this, see common cause vs special cause variation.
A process is in control — in statistical control — when it is stable and predictable, exhibiting only common cause variation. Its output varies, but within a consistent band, and you can predict the range future output will fall within. Control limits define that band, and crucially they are calculated from the process's own historical behaviour — they describe how the process actually varies, not what anyone wants it to do. A point beyond the control limits, or an unnatural pattern, signals a special cause: the process has become unstable. Being in control is a statement about stability and predictability only. It says nothing, by itself, about whether the predictable output is actually good enough for the customer.
A process or product is in spec when its output meets the specification — the tolerance limits the customer or design requires. Specification limits define the boundary of acceptable: a dimension between an upper and lower tolerance, a property within a required range. In spec is purely about conformance to the requirement: is this output acceptable to the customer. Specification limits come entirely from the customer's needs and the design intent — they have nothing to do with how the process happens to behave. A part is in spec or out of spec regardless of whether the process making it is stable. In spec is a statement about the product meeting requirements, not about the process being predictable.
The heart of the confusion is that control limits and specification limits are completely unrelated, even though both can appear as lines on a chart. Control limits come from the process — they are calculated from its actual variation and describe what it does. Specification limits come from the customer — they are set by the requirement and describe what is needed. They can sit anywhere relative to each other: control limits inside the spec (a capable, stable process), control limits straddling or outside the spec (a stable process that still makes out-of-spec parts), or a process drifting unpredictably within a wide spec. Because the two sets of limits have different origins and different meanings, in control and in spec are independent properties that must be assessed separately.
Consider four situations for a machined dimension. In control and in spec: the process is stable, and its predictable band sits comfortably inside the customer tolerance — the ideal, a capable and stable process. In control but out of spec: the process is perfectly stable and predictable, but its natural band falls partly outside the tolerance, so it reliably, predictably produces some bad parts — stability is no comfort here. In spec but out of control: every part currently happens to fall within tolerance, but the process is unstable and unpredictable, so there is no guarantee it will stay that way. Neither: unstable and producing out-of-spec parts. Only the first situation is genuinely satisfactory, and you cannot tell which one you are in without assessing control and spec separately.
The two properties relate in a specific way: stabilise first, then make capable. A process must be in control before you can meaningfully judge whether it is capable of meeting spec, because an unstable process has no predictable band to compare against the tolerance — its capability is undefined while special causes come and go. So the sequence is: first eliminate special causes to bring the process into control (stable and predictable), then assess and improve its capability so its stable band fits within the specification. Chasing capability on an out-of-control process is futile; the numbers shift every time a special cause appears. Control is the foundation; capability — fitting the stable process inside the spec — is built on top of it.
The distinction sits underneath the quality factor of OEE. OEE's quality factor counts conforming, in-spec, good units — but a process that is merely in spec today without being in control is a quality problem waiting to happen, because it is unpredictable and could drift out of tolerance at any time. Conversely, a stable in-control process whose band sits outside the spec will steadily depress the quality factor with predictable scrap. Reading OEE quality alongside whether the process is in control and capable is what separates a temporary good number from a reliably good one, and connects to distinguishing common cause from special cause in the trend.
Fabrico trends quality output over time, which is what lets a team see not just whether parts are in spec right now but whether the process producing them is stable and predictable. A quality factor that looks fine today but bounces unpredictably is a different situation from one that is steady and capable — and the difference is exactly what tells you whether your good OEE is reliable or fragile. Seeing conformance and stability together keeps the quality number honest. Book a demo to see process stability behind your quality results.
In control means a process is statistically stable and predictable, with only common cause variation within control limits derived from the process itself. In spec means the output meets the customer's specification or tolerance. They are independent: one describes process behaviour, the other product conformance.
Yes. A perfectly stable, predictable process can have a natural variation band that falls partly outside the customer tolerance, so it reliably produces some out-of-spec parts. Stability does not guarantee conformance.
Control limits are calculated from the process's own variation and describe how it actually behaves. Specification limits come from the customer's requirements and describe what is acceptable. They have different origins and are unrelated, even though both can appear on a chart.
Because an unstable process has no predictable band to compare against the specification — its capability is undefined while special causes come and go. You stabilise the process first, then assess and improve whether its stable band fits within the spec.
OEE's quality factor counts in-spec good units, but a process that is in spec without being in control is unpredictable and could drift out of tolerance anytime. Reading quality alongside process stability separates a reliably good OEE from a fragile one.
Login
Blog
