PDCA and PDSA: Shewhart, Deming and the Improvement Cycle

Almost every continuous improvement method you will meet, kaizen included, runs on a simple loop: plan a change, try it, look at what happened, and decide what to do next. You will see it written as PDCA (Plan-Do-Check-Act) or PDSA (Plan-Do-Study-Act), sometimes called the Deming cycle or the Shewhart cycle.

The two versions look nearly identical, and many people use them interchangeably. Their history, though, explains a real difference in emphasis, and that difference changes how useful the cycle is in practice. This guide traces the cycle from its origins and then shows how to run one well.

Shewhart's circle

The story starts with Walter Shewhart of Bell Laboratories, whom the W. Edwards Deming Institute describes as Deming's mentor. In his 1939 book Statistical Method from the Viewpoint of Quality Control, Shewhart took the usual straight-line sequence of specification, production and inspection and argued that it should be a loop. As quoted in Ronald Moen's history of the cycle, Shewhart wrote that "these three steps must go in a circle instead of in a straight line."

Shewhart compared the three steps to the scientific method: specification, production and inspection correspond to making a hypothesis, carrying out an experiment and testing the hypothesis. That link to the scientific method is the thread that runs through every later version. Deming, then 39, edited the book.

The Deming wheel in Japan

In 1950, Deming presented a modified version of Shewhart's cycle at an eight-day seminar on statistical quality control sponsored by the Japanese Union of Scientists and Engineers (JUSE). According to Moen, Deming turned a straight line of design, produce and sell into a circle and added a fourth step: redesign through market research. He stressed constant interaction among design, production, sales and research, with quality of product and service as the aim. The Japanese called this the Deming wheel, or Deming circle.

How PDCA emerged

The Plan-Do-Check-Act version came next, and Deming did not write it. Moen and Clifford Norman, in a 2010 article republished by the Deming Institute, report that according to Masaaki Imai, Japanese executives recast the Deming wheel from the 1950 seminar into PDCA. Design became plan, production became do, sales became check (sales figures confirm whether the customer is satisfied), and research became act. Moen notes that no one person has claimed authorship.

In that Japanese PDCA cycle, which Moen dates to 1951, plan means defining a problem and a hypothesis about possible causes and solutions; do means implementing; check means evaluating the results; and act means going back to plan if the results are unsatisfactory, or standardizing if they are satisfactory. Moen writes that the cycle emphasized preventing errors from recurring by establishing standards and continually modifying them. Kaoru Ishikawa later expanded the plan step to include setting goals, targets and methods, and the do step to include training and education.

Deming's PDSA

Deming reintroduced the Shewhart cycle in his 1986 book Out of the Crisis and kept refining it through 1993, when he called it the Shewhart Cycle for Learning and Improvement, the PDSA cycle. Moen, who managed Deming's monthly visits to a General Motors division in the 1980s and helped at 70 of his four-day seminars, reports that Deming frequently warned seminar audiences that the plan-do-check-act version was inaccurate because the English word "check" means "to hold back."

Deming was not shy about the distinction. In a 1990 letter to Moen about a manuscript, he wrote: "be sure to call it PDSA, not the corruption PDCA." Moen's summary is that the Japanese developed PDCA from Deming's 1950 seminars, Deming never embraced it, and the two cycles are related only through the scientific method.

Check versus Study: why the word matters

The Deming Institute explains the difference this way. Check focuses on the implementation of a change, with success or failure, followed by corrections to the plan if it failed. Deming's focus was on predicting the results of an improvement effort, studying the actual results and comparing them, possibly revising the theory. In the Institute's words, he stressed that "the need to develop new knowledge, from learning, is always guided by a theory."

In 1991, Moen and his co-authors Nolan and Provost made that explicit by adding a required prediction, with its associated theory, to the plan step; the study step then compares the observed data to the prediction. That small addition turns the cycle from "did it work?" into "what did we learn about why it worked, or didn't?"

Here is a hypothetical example of the difference. A small team tries a new intake form to cut back-and-forth emails with clients. A check mindset asks whether emails went down. A study mindset writes down a prediction first, perhaps "follow-up emails will drop by about half, mostly because of the new deadline field," and then compares. If emails dropped but clients still asked the same deadline questions, the team has learned that its theory was wrong, even though the form "worked." That is knowledge it can use on the next cycle.

Whichever label your organization uses, you can run the cycle in the study spirit.

Running a cycle that teaches you something

The Institute for Healthcare Improvement (IHI) uses PDSA inside the Model for Improvement, a framework developed by Associates in Process Improvement. The model pairs the cycle with three questions: What are we trying to accomplish? How will we know that a change is an improvement? What change can we make that will result in improvement? IHI's guidance on testing changes offers practical steps that apply well beyond health care:

1. Plan: state the objective of the test and the questions it should answer, make predictions, and decide who, what, when and where, including what data to collect. IHI advises that early cycles "should be scoped as small as possible."
2. Do: carry out the test, and document problems and unexpected observations.
3. Study: complete the analysis, compare results to your predictions, and summarize what you learned.
4. Act: refine the change and plan the next test.

IHI also recommends planning a couple of cycles ahead, scaling tests down (sampling the next 10 cases instead of 200, or one room for one week), and testing under a range of conditions before making a change permanent. It notes that most changes will need many cycles, run in sequence.

Common mistakes

• Skipping the prediction, so there is nothing to compare results against.
• Making the first test too big, which makes it slow, expensive and hard to interpret.
• Treating Act as "roll it out everywhere" after one success, instead of testing under different conditions first.
• Stopping after a failed test. IHI points out that teams often learn very important lessons from failed tests of change.

Key takeaways

• The cycle began with Walter Shewhart's 1939 idea of putting specification, production and inspection in a circle, modeled on the scientific method.
• Deming presented his version in Japan in 1950; Japanese executives recast it as PDCA, which no single person has claimed to author.
• Deming preferred PDSA and called PDCA a "corruption," because Study emphasizes learning against a prediction rather than checking pass or fail.
• A good cycle starts with a stated prediction, keeps early tests small and runs many cycles in sequence.
• Failed tests are useful data, not wasted effort.

Sources

• The W. Edwards Deming Institute, PDSA Cycle
• Ronald Moen, Foundation and History of the PDSA Cycle (The W. Edwards Deming Institute)
• Ronald D. Moen and Clifford L. Norman, Circling Back: Clearing Up Myths About the Deming Cycle (November 2010)
• Institute for Healthcare Improvement, Model for Improvement
• Institute for Healthcare Improvement, Model for Improvement: Testing Changes