What does PDCA stand for?

PDCA stands for Plan-Do-Check-Act. Plan a change and predict its effect, Do it on a small scale, Check the actual result against your prediction, and Act by standardizing what worked or discarding what did not. It is a four-step loop for improving a process by experiment, repeated continuously.

Who invented the PDCA cycle?

The cycle originated with the physicist Walter A. Shewhart at Bell Telephone Laboratories in the 1920s, who bent the straight-line sequence of specification, production, and inspection into a repeating circle in his 1939 book. W. Edwards Deming, who studied under Shewhart, carried it to Japan in the 1950s, where it became known as the Deming wheel.

What is the difference between PDCA and PDSA?

They are the same loop with one word changed. Deming preferred PDSA (Plan-Do-Study-Act), which he set out in his 1993 book The New Economics and called the Shewhart Cycle for Learning and Improvement. He objected that Check could be read as inspect or hold back, while Study better captures analyzing what the test taught you.

Why is the Check step so important in PDCA?

Because it is what makes PDCA an experiment rather than a guess. Check compares the actual result against the prediction you wrote during Plan. Without a prediction and an honest comparison, PDCA degrades into changing things for the sake of change. Deming renamed the step Study to emphasize learning from the result, especially the surprises.

How is PDCA related to kaizen?

PDCA is the engine; kaizen is the vehicle. A kaizen event is a compressed run of PDCA cycles: grasp the condition, test changes on the floor, check the results, and standardize the winners. Kaizen is the culture of continuous improvement, and PDCA is the disciplined loop it repeats at every scale.

What is SDCA and how does it relate to PDCA?

SDCA is Standardize-Do-Check-Act, the twin of PDCA. Where PDCA improves a process to a new level, SDCA holds it there: standardize the best method, run to it, check it is being followed, and act on drift. Teams alternate the two, PDCA to climb and SDCA to hold, so improvement becomes a staircase where each gain sticks.

Is a failed PDCA cycle a failure?

No. A cycle that disproves your hypothesis did its job: it stopped you from rolling out a change that would not have helped and taught you something true about the process. The only real failure is a cycle with no prediction and no honest Check, because then nothing was learned.

The PDCA Cycle (Plan-Do-Check-Act)

Deming's improvement loop, its real history, and how to run small experiments that stick.

The PDCA cycle (Plan-Do-Check-Act) is a four-step loop for improving a process by experiment: Plan a change and predict its effect, Do it on a small scale, Check the result against the prediction, and Act by standardizing what worked or discarding what did not. It is the engine under kaizen.

PDCA is the most quoted improvement cycle in manufacturing and the most misused. Teams run "plan, do, plan, do" with no real Check and call it PDCA; others treat it as paperwork rather than a genuine experiment. Done right, PDCA is the scientific method sized for the plant floor: a hypothesis, a small test, an honest look at the data, and a decision. It sits at the center of lean manufacturing because it is how every other lean tool actually gets better over time. Get the history and the discipline right and PDCA stops being a poster and starts being how the floor learns.

Where did the PDCA cycle come from?

The cycle traces to the physicist Walter A. Shewhart at Bell Telephone Laboratories in the 1920s. In his 1939 book Statistical Method from the Viewpoint of Quality Control Shewhart bent the old straight-line sequence of specification, production, and inspection into a circle, arguing that the three steps should repeat and feed back into one another rather than run once. That closed loop was the seed of PDCA.

W. Edwards Deming, who had studied under Shewhart, carried the idea to Japan in the 1950s while teaching statistical quality control to Japanese engineers and executives. Japanese practitioners called the loop the "Deming wheel" or "Deming cycle," and it became a backbone of the continuous-improvement movement that grew into the Toyota Production System (ASQ, PDCA Cycle). The four-word Plan-Do-Check-Act phrasing came out of that Japanese adaptation of Shewhart's original three steps.

Deming himself grew uneasy with the word "Check." Late in his life, in his 1993 book The New Economics he set out the variant PDSA, Plan-Do-Study-Act, which he called the Shewhart Cycle for Learning and Improvement. He objected that "Check" could be read as "hold back" or "inspect," while "Study" better captured the point: analyze what the test taught you (The W. Edwards Deming Institute, PDSA Cycle). ASQ today recognizes PDCA, PDSA, the Deming cycle, and the Shewhart cycle as names for the same family of loop. The short version: Shewhart invented it, Deming spread it, Japan named it, and Deming later refined it into PDSA.

The classic image: PDCA is a wheel rolling uphill. Each turn tests and adopts an improvement, and standard work is the wedge that keeps the gain from sliding back before the next turn.

What happens in each step of PDCA?

Each letter is a distinct discipline, and skipping any one breaks the loop.

Plan. State the problem, study the current condition, form a hypothesis about the cause, and design a small change. The plan must include a prediction: what you expect the change to do to a specific number. No prediction, no way to Check honestly.

Do. Run the change on the smallest scale that will teach you something, one line, one shift, one machine, and record what actually happened, including anything that went sideways. Do is a controlled trial, not a full rollout.

Check (or Study). Compare the result against the prediction you wrote in Plan. Did the number move the way you expected, more, less, or not at all? Surprises are the most valuable output, because they mean your understanding of the process was wrong and you just learned something.

Act. Decide. If it worked, standardize it, write it into standard work so it becomes the new baseline, and spread it. If it did not, keep the learning and discard the change. Then turn the wheel again on the next problem or the next iteration.

The half of PDCA that gets skipped is the prediction in Plan and the honest comparison in Check. Without them, PDCA degrades into "try things and keep the ones that feel better," which is not learning; it is guessing with extra steps.

How do you actually run a PDCA cycle? A step sequence

Here is the loop as a repeatable sequence a floor team can run in a day for a small problem or over a week for a bigger one.

Grasp the current condition. Go to the gemba watch the process, and get a baseline number: the scrap rate, the changeover minutes, the giveaway per case. You cannot improve what you have not measured.
State a target and a hypothesis. Write the target condition ("changeover under 20 minutes") and your best theory of what change will get you there. Tie the theory to a cause, not a symptom, using root cause analysis if the cause is unclear.
Predict the result. Write down, before you run it, what you expect the change to do to the baseline number. This single sentence is what makes the cycle an experiment instead of an opinion.
Do it small. Run the change on one line or one shift. Change one thing at a time so you can attribute the result. Record what happened, including the unexpected.
Check against the prediction. Put the result next to the prediction. Match, miss, or surprise? Decide what the data says, not what you hoped it would say.
Act on the verdict. If it worked, standardize and roll out. If it did not, capture the learning and try the next hypothesis. Either way, the wheel turns again, because no process is ever finished.

Notice that a failed cycle is not a failed PDCA. A cycle that disproves your hypothesis did its job: it stopped you from rolling out a change that would not have helped, and it told you something true about the process. The only failed PDCA is the one with no prediction and no honest Check.

How does PDCA power kaizen and the rest of lean?

PDCA is the engine; kaizen is the vehicle. A kaizen event is essentially a compressed run of PDCA cycles: the team grasps the condition, tests changes on the floor, checks the results, and standardizes the winners, all in a few focused days. The smallest version, a point kaizen fix at a single spot, is one quick turn of the wheel. The point is the same at every scale: improvement is an experiment, and PDCA is the shape of the experiment.

PDCA also pairs with its quieter twin, SDCA, Standardize-Do-Check-Act. Where PDCA improves the process, SDCA holds it: you standardize the current best method, run to it, check that it is being followed, and act on drift. Toyota's teaching is that you alternate, PDCA to climb, SDCA to hold the new level, which is the same wedge-and-wheel image from earlier. Without the SDCA hold, every PDCA gain rolls back down before the next turn.

Improvement is a staircase, not a ramp. PDCA raises the process to a new level; SDCA standardizes and holds it flat until the next PDCA climb. The hold is what makes the gain permanent.

Where do teams go wrong with PDCA?

No prediction in Plan. Skipping the "what do we expect" step turns Check into a rationalization. If you never said what should happen, you cannot honestly say whether it did.
No real Check. The most common failure: running "plan, do, plan, do," changing things for the sake of change with no look back at the data. Deming renamed the step Study precisely to fight this.
Do too big. Rolling the change out plant-wide before checking it means a wrong idea contaminates everything, and you cannot untangle what caused what.
Act that never standardizes. A winning change that never makes it into standard work evaporates when the team moves on. No SDCA hold, no lasting gain.
Treating a failed cycle as a failure. A cycle that disproves the hypothesis is a success; it saved you a bad rollout and taught you something true. Punishing failed cycles kills the honesty the loop depends on.

How does PDCA compare to other problem-solving methods?

PDCA is the lightweight, general-purpose loop. When a problem is bigger or the cause is genuinely unknown, heavier structured methods take over, and PDCA often nests inside them. A full comparison of A3, 8D, DMAIC, PDCA, and Kepner-Tregoe is in our problem-solving methods comparison and the head-to-head with Six Sigma's roadmap is in DMAIC vs PDCA. The short guidance: reach for PDCA by default for the daily stream of small, clear problems, and escalate to a heavier method only when the cost of a wrong guess earns the extra rigor. Most plants under-use the fast loop and over-formalize the small problems.

Fact	What the record shows	Primary source
Origin	Walter Shewhart bent specification-production-inspection into a repeating cycle in his 1939 book at Bell Labs.	ASQ, PDCA Cycle
Spread to Japan	Deming taught the loop in Japan in the 1950s; it became known as the Deming wheel.	ASQ, PDCA Cycle
PDSA variant	Deming preferred Plan-Do-Study-Act, arguing "Study" better captures learning than "Check."	Deming Institute, PDSA
Recognized names	PDCA, PDSA, Deming cycle, and Shewhart cycle all name the same loop.	ASQ, PDCA Cycle

The documented lineage: Shewhart invented the loop, Deming carried it to Japan, and Deming later refined Check into Study as PDSA.

How does live floor data make the PDCA loop faster?

PDCA's speed is capped by how fast you can Check. If the result of a change lives on a clipboard that gets keyed into a spreadsheet three days later, the loop stalls between Do and Check, and the team loses the thread. Plants that capture output, scrap, and cycle data live from stations and machines can read the result of a test the moment it finishes, so a cycle that used to take a week closes in a shift. That is the practical value of a live factory visibility layer over existing systems, no rip-and-replace: it turns Check from a transcription project into a glance. Whether you are running one point kaizen fix or standardizing a change across a batch production line, the tighter the loop, the faster the floor learns. See how digitizing the floor first plays out in the CLS case study. PDCA is only as fast as its slowest step, and for most plants that step is Check.