PDCA explained: how Plan-Do-Check-Act actually works in a modern QMS
Plan-Do-Check-Act is the engine behind ISO 9001 and almost every continuous improvement program in the world. Here is what each step really means and how to run a PDCA cycle that produces results.
Plan-Do-Check-Act, usually shortened to PDCA, is the four-step cycle that sits inside ISO 9001, ISO 14001, ISO 27001, and most modern improvement methodologies — from Six Sigma's DMAIC to lean's kaizen. If you only ever learn one quality model, learn this one.
Where PDCA comes from
Walter Shewhart developed the original three-step "specification – production – inspection" loop at Bell Labs in the 1920s. W. Edwards Deming extended it into the four-step cycle the world later named after him. PDCA is sometimes still called the Deming cycle.
The four steps, in plain English
Plan
Define the problem and the desired outcome. Establish the objectives, the processes needed, and how you will measure success. "Plan" is where most cycles die — either because the problem was never defined sharply enough or because success was never measurable.
Do
Implement the plan, ideally on a small scale first. Collect data as you go. "Do" is execution, not deliberation — by this stage the debate should be over.
Check
Compare the actual results against the expected results. What worked, what did not, and what surprised you? In a modern QMS this is where dashboards, internal audits, and management review live.
Act
Standardize what worked and roll it out at scale. Adjust or kill what did not. Then start the next cycle. Without this step, PDCA becomes PDC — endless analysis with no codified change.
How PDCA maps to ISO 9001
- Plan — Clauses 4, 5, 6: context, leadership, planning
- Do — Clauses 7, 8: support and operation
- Check — Clause 9: performance evaluation, internal audit, management review
- Act — Clause 10: improvement and corrective action
ISO 9001 is essentially a structured PDCA loop with the volume turned up. Every clause is doing one of those four jobs.
A concrete example
A precision parts manufacturer notices its scrap rate has crept from 1.8% to 3.4% over six months.
- Plan — hypothesis: tool wear on the new alloy is faster than the maintenance schedule assumes. Target: scrap below 2.0% within 90 days. Measure: daily scrap by machine.
- Do — pilot a tighter tool-change interval on two machines for four weeks.
- Check — pilot machines drop to 1.6% scrap. Control machines stay at 3.3%. Hypothesis confirmed.
- Act — update the maintenance standard across all 14 machines, retrain operators, and add the new interval to the preventive maintenance system. Schedule the next PDCA cycle on inspection sampling.
Why PDCA fails in practice
- Plan is skipped — the team jumps to a fix before the problem is defined
- Do is treated as the project — "we ran the pilot, we are done"
- Check is performed but never written down, so the lesson is lost the next time staff turns over
- Act is missed — pilot results are never standardized, so the gain disappears within a quarter
PDCA in a modern, AI-assisted QMS
Modern QMS platforms can shrink the cycle dramatically. "Check" used to mean a quarterly report; now it can be a live dashboard that flags drift the moment it begins. "Act" used to mean a memo; now it can be an automatic update to a controlled work instruction with version history. The framework does not change. The cycle just gets faster — and the organizations that run it fastest, win.
“PDCA is not a quality tool. It is how learning organizations learn. Everything else in ISO 9001 is scaffolding around it.”