You're in spec, but... (part 1)
...your process just shifted.
You get a new lot of bar stock Monday morning. The operator loads it, runs parts, spot-checks every fifth on the CMM. Everything checked passes.
Wednesday, your customer calls about out-of-spec parts - but you checked all the measurements. What happened?
Specs don't track the process
Spec limits answer one question: does this part meet the requirement? If the bore diameter is 50.0 ± 3.0 mm, any measurement between 47.0 and 53.0 ships.
Specs can't tell you whether something about your process has changed.
A control chart tracks your process over time. It calculates its own limits from your data — not from the print, not from the customer. These limits represent what your process normally does. Usually your measurements will land between these limits. When a point lands outside them, the process has changed.
The first 15 parts below are on the old lot. Part 16 is where the new lot starts.
Every point is between the purple spec lines (LSL/USL) - every part ships. But five red points break above the upper control limit - the process shifted.
That shift means your measurements now are consistently closer to the upper spec limit. There's less margin for error, and a greater chance parts will be out of spec.
So now what
The chart doesn't know it's a new material lot. It just knows when the change happened. You see the jump at part 16 and the five points breaking your upper control limit, and ask: what was different Monday morning?
New lot of bar stock - maybe the hardness is different, maybe the diameter is on the high side of the incoming tolerance band. You can call your supplier now, while you have the lot number and the data to back it up.
Without the chart, you don't see it coming. The process keeps running at the new level — most parts pass, but some parts now exceed specs. Spot checks catch some, but the rest ship. By Wednesday, your customer has a box of them.
Passing isn't the same as stable
A part that's in spec is passing. A process that's in control is stable — it's doing what it always does, and you can predict what it'll do tomorrow.
When you have both, you're in good shape. When specs say yes but the chart says something changed, your process is less predictable and your specs may say no tomorrow.
Try it yourself — the process shift example on zonesix.ai/learn shows this pattern with interactive charts you can explore.