The PDSA cycle — Plan, Do, Study, Act — is Deming's iterative learning framework, a method for generating knowledge under conditions of uncertainty. It is adapted from Walter Shewhart's three-stage specification-production-inspection cycle, which Deming reformulated into a four-stage cycle and generalized beyond manufacturing into a universal epistemological tool. The cycle begins with a Plan (a prediction based on theory), proceeds to Do (carry out the plan on a small scale), then Study (compare results to predictions), and finally Act (adopt, adapt, or abandon based on what was learned). The output of each cycle is not just a decision but new knowledge — a refined theory that feeds the next iteration.
Deming was insistent on the word "Study" rather than "Check" in the third step. The PDCA variant (Plan-Do-Check-Act) is the Japanese adaptation that spread through JUSE and became standard in Japan's quality movement. Deming regarded "Check" as inadequate because it implies mere inspection — holding results against a standard — whereas "Study" implies learning, comparison of results to predictions, and the generation of new theoretical understanding. This is not a pedantic distinction. It reflects the core difference between quality control (detecting defects) and quality improvement (understanding and improving the system that produces outcomes). The PDCA/PDSA distinction is a window into the broader tension between Deming's philosophy and the TQM/Six Sigma movements that claimed his legacy while often missing his epistemological point.
The intellectual lineage of the PDSA cycle traces back through Shewhart to the pragmatist philosophy of C.I. Lewis and ultimately to the scientific method itself. Shewhart's original cycle mapped directly to the scientific method: specification (hypothesis), production (experiment), inspection (observation). Deming's extension added the Act step — the decision about what to do with the new knowledge — and reframed the entire cycle as a learning process rather than a production process. This is the connection to theory-of-knowledge: the PDSA cycle is Deming's operationalization of the claim that there is no knowledge without theory. Each Plan step requires a theory; each Study step tests that theory; each cycle deepens understanding.
The PDSA cycle prefigures and directly influenced several later iterative frameworks. Boyd's OODA loop (Observe-Orient-Decide-Act), developed independently in the military domain, shares the same iterative structure and the same emphasis on the cognitive step (Orient/Study) as the critical element (see pdsa-ooda-parallel-iterative-learning-cycles-across-domains for a detailed comparison). Eric Ries's Build-Measure-Learn cycle in lean startup methodology is a direct descendant — Ries explicitly acknowledges the Deming lineage through Steve Blank and the lean manufacturing tradition. In software development, Agile methodologies' sprint retrospectives are PDSA cycles. The common thread is the insight from statistical-process-control-and-variation-theory and from system-of-profound-knowledge: in a world of variation and uncertainty, learning must be iterative, theory-driven, and empirically grounded.
The PDSA cycle is also the practical mechanism through which the system-of-profound-knowledge operates. It is the method by which managers develop appreciation-for-a-system, learn to distinguish common-cause-vs-special-cause-variation, and build the theoretical frameworks that constitute genuine knowledge. Without the PDSA cycle, the SoPK would be a static philosophy; with it, the SoPK becomes a dynamic learning methodology. Deming demonstrated this in his seminars by running the-red-bead-experiment and then asking participants to study the results — the demonstration itself was a PDSA cycle, with the audience's predictions serving as the Plan, the bead draws as the Do, the statistical analysis as the Study, and the revised understanding of variation as the Act.