Services See all → Strategic planningCulture and engagementAccountability and executionProblem solvingHigh performance teamsLean and kaizen
Methods See all → Fishbone diagram5SPoka yokeFMEA
AboutCase studiesBlog Book a call Español
Quality and measurement · November 5, 2026

PDCA cycle: the lap almost nobody finishes

Four curved segments in gold, sand and teal forming one continuous circle

PDCA gets taught on a single slide, a circle with four letters: plan, do, check, act. On the floor it almost never breaks because the slide was wrong. It breaks at the same step every time, and it is not the step most people expect.

The PDCA cycle, also called the Deming cycle or the Shewhart cycle, is a four step continuous improvement method that repeats: plan a change, do it at small scale, check whether it worked using data, and act on the result, either by standardizing the change or by adjusting it and starting again. Walter Shewhart developed it and W. Edwards Deming popularized it in Japan after the war, which is why it also carries his name.

Four steps, one circle that repeats. It sounds like common sense, and it is. The problem is never understanding the concept. It is knowing where the lap actually stalls once your team runs it on your own line.

Plan: the half everyone does and the half almost nobody does

Plan has two halves. The first is naming the problem or the opportunity, and that half almost always goes fine, because everybody on the floor already knows what is hurting them. The second half is defining, before you touch anything, what data will tell you whether the change worked. That second half is the one people skip. Without it, checking turns into opinion.

Do: on purpose, and small

Do means testing the change at a scale where a bad result does not cost much: one shift, one station, one lot, never rolling it out across the whole line. The temptation is to skip the small scale because it already looks like it is going to work. That temptation is what turns a PDCA into a process change with no pilot.

Check: against the criterion you wrote down in advance

This is where the cycle earns its keep or becomes a decorative lap. Checking means comparing the result against the criterion you wrote in the plan step, not against the impression that things look better. If you never defined the criterion up front, by the check step you end up moving the goalposts to fit the result, and that is the exact point where PDCA stops being improvement and becomes justification.

Act: where most cycles stop

We have covered plan, do and check. Most of the PDCA cycles I see on the floor make it this far, land on a good or a bad result, and stop right there. Nobody acts. If the change worked and nobody standardizes it, the line drifts back to the old way within two months, because no one made the new method official. If the change did not work and nobody adjusts the plan or runs another lap, the lesson disappears with the next shift rotation.

Act splits into two paths, not one. If it worked, you standardize it: document it, train on it, audit it like any other standard. If it did not, you adjust the plan with what you learned and start another lap. Both paths end at the same question. Who owns making sure this does not get lost.

Why the second lap moves faster

The first honest lap of a PDCA, with a real criterion defined in the plan step, takes longer than anybody wants to spend. The second lap on the same problem moves faster, because the verification criterion already exists and so does the habit of testing small before rolling out. That is what separates a plant that keeps improving from one that only runs isolated projects: the cycle is installed.

Where this lives inside a kaizen event

A kaizen event is, underneath the agenda, several fast laps of PDCA run with the whole team in the room. The difference between an event that leaves a real change and one that leaves a group photo comes down to whether somebody kept the act step after the team went back to their regular shift. That follow up is the part that falls hardest, and it is the same part that falls in any standalone PDCA.

Start here

Take a change your team already wants to make. Before touching anything, write down what data will tell you it worked and at what small scale you will test it. If you cannot answer both questions, you are still in the plan step and did not know it. If you want to install the full cycle in your operation, with follow through all the way to act, that is part of what I work on in lean and kaizen.

What does PDCA stand for?

Plan, do, check, act. It is a repeating continuous improvement cycle, not a one time project.

What is the difference between PDCA and PDSA?

PDSA replaces check with study. Deming preferred PDSA because he insisted that checking a result is not enough, you have to understand why it happened before you act on it. Most plants in practice still call it PDCA.

How often should you run a PDCA cycle?

There is no fixed frequency. It takes as long as testing the change at small scale and confirming the result takes. What is fixed is the order: you do not act before you have defined how you will check.

← Back to the blog

Next step

What is holding your operation back today?

Tell me the concrete problem. I will tell you whether it is mine to solve or someone else’s, and what I would do first.