PDCA / Plan-Do-Check-Act

Plan — state what you want to obtain, how you will go about it, and above all how you will know it worked. A plan without a success criterion makes the check step impossible.

Do — implement, on a small scale. A trial on one station tells you more than a roll-out across the whole line, and costs infinitely less when it fails.

Check — compare what is observed with what was planned. It is the step that gets skipped, because it is the only one that can say you were wrong.

Act — only two outcomes, and one must be chosen: either the result is achieved and you standardise, or it is not and you go back to planning with what you have learnt. "We’ll see later" is not a third option, it is abandonment.

The wheel is commonly attributed to W. Edwards Deming, who had it from Walter Shewhart. Deming in fact argued for PDSA — study instead of check — because "check" suggests inspection, whereas the step consists of studying what happened, including what had not been anticipated.

The nuance is not academic: it changes what is done at that step. Checking is ticking a box. Studying is looking for why the gap exists, and sometimes discovering that the initial hypothesis was wrong for an interesting reason.

The full picture also has a wedge under the wheel: that is the standard. Without it, the wheel rolls back down the slope it has just climbed — the next cycle starts again from a level that was thought secured.