PDCA: how improvement compounds
Plan, Do, Check, Act — the four-step wheel behind every kaizen. It looks trivial. But run it honestly and each turn ratchets up a proven gain that can’t slide back. Watch one experiment go around the loop, then watch the loop climb.
A change isn’t an improvement
Anyone can change something. An improvement is a change you proved worked — and then locked in so the next change starts from higher ground. PDCA is the discipline that turns guesses into gains. Let’s run one.
Four steps, one loop.
Plan a change, Do it small, Check what actually happened, Act on what you learned — then go again. It’s a learning loop, not a checklist: the output of one turn is the input to the next.
State a hypothesis, not a wish.
“A self-check-in kiosk will cut the morning wait.” Make it testable: a target (12 → 8 min) and a prediction. If you can’t say what you expect to happen, you can’t learn from what does.
Run it small.
One desk, one week — not a clinic-wide rollout. Do is a deliberately small test so a wrong guess is cheap and the signal is clean. Big-bang changes skip straight to Act and call it improvement.
Compare actual to predicted.
Wait dropped to 9 minutes — better, but not the 8 you predicted. And you learned something you didn’t plan for: older patients struggled with the kiosk. The gap between predicted and actual is the lesson.
Adopt, adjust, or abandon — then lock it in.
The kiosk works, so adopt it — but add a greeter for the patients who struggled. Then standardize: write it into the standard so it can’t quietly revert. That standard is the chock behind the wheel. Now plan the next turn.
Each proven gain is the new floor.
Zoom out. That single turn lifted the standard from 12 to 9 minutes — and the standard holds it there. The next turn doesn’t start from zero; it starts from 9. Improvement isn’t a leap, it’s a wheel climbing a slope, one proven turn at a time.
The chock is the whole trick.
Without the standard behind it, the wheel rolls straight back down between shifts and people. Standardized work is what makes a gain permanent — PDCA and standardized work are two halves of the same engine: improve, lock, improve, lock.
Firefighting looks like motion.
Skip Check and there’s no chock: you change things, they slide back, you change them again — busy forever, climbing nothing. A change you didn’t verify isn’t improvement; it’s a guess with good intentions.
Where each step breaks
PDCA is easy to draw and hard to run. Almost every failure is one of four shortcuts — one per step — each of which feels like progress while quietly stopping the wheel.
Every step has a way to cheat.
PDCA is easy to draw and hard to run, because each step has a shortcut that feels like progress. Here’s where the wheel usually cracks.
Jumping to a solution.
The most common one: skipping the hypothesis and the target entirely — “let’s just buy kiosks.” With no prediction, there’s nothing to Check against later, so you can never actually learn whether it worked.
The big-bang rollout.
Rolling the change out everywhere at once instead of a small test. When it half-works, you can’t tell what helped, you can’t cheaply undo it, and the mess makes the next experiment impossible to read.
Declaring victory.
Skipping measurement and trusting the story you wanted — confirmation bias with a nice slide. If you don’t compare actual to predicted, Check becomes a celebration, not a test, and the wheel stops turning.
Never standardizing.
Proving something worked and then not writing it down. The gain quietly reverts between shifts and people, and three months later someone “discovers” the same fix. No chock, no ratchet.
One loop, every cadence
The same four steps run at every altitude of the company — a huddle that turns in minutes, a strategy that turns once a year. Small fast wheels nested inside slow big ones, all turning the same way.
The same loop, every speed.
PDCA isn’t only for big projects. It’s the same four steps running at every level of the organization, just at different speeds — from a huddle that lasts minutes to a strategy that turns once a year.
The daily huddle.
A team stands at the board: yesterday’s plan vs actual (Check), a small countermeasure for today (Plan → Do). A micro-turn of the wheel, every shift. Fast, cheap, constant.
The kaizen event.
A cross-functional team takes one process off-line for a focused week — plan, try, measure, standardize — and makes a step change. One deliberate, faster turn.
The A3.
One important problem, one page, one careful turn of PDCA — background, target, root cause, countermeasures, check, follow-up. The A3 is a single, rigorous rotation of the wheel.
Hoshin.
At the top, the whole company’s direction is one slow annual turn: plan the breakthrough, deploy and do, review against the plan, adjust. Nested wheels — small fast ones inside slow big ones, all turning the same way.
One loop, everywhere.
PDCA is the engine inside the tools. An A3 is one big, careful turn of it; a kaizen event is a fast one; standardized work is the chock that holds each gain.
Write your first A3 →Frequently asked
- What is PDCA?
- PDCA — Plan, Do, Check, Act — is a four-step cycle for improving work by learning, not guessing. Plan states a testable change and predicts its result; Do runs it as a small test; Check compares what actually happened to what you predicted; Act adopts, adjusts, or abandons the change and then standardizes what worked so it holds. Because the output of one turn is the input to the next, PDCA is a learning loop rather than a checklist. It is also called the Deming cycle or the Shewhart cycle.
- What is the difference between PDCA and PDSA?
- They are the same cycle with one word changed: PDSA replaces Check with Study. W. Edwards Deming preferred Plan-Do-Study-Act because "Check" can suggest a simple pass/fail inspection, while "Study" stresses the point of the step — analyzing the gap between what you predicted and what actually happened, and learning from it. In practice teams use the terms interchangeably; PDCA is the more common label in lean and manufacturing, PDSA in healthcare and quality improvement. If your Check step has become a rubber-stamp, switching to the PDSA mindset is the fix.
- How does PDCA relate to continuous improvement and kaizen?
- PDCA is the engine inside continuous improvement. Kaizen — small, continual improvement — works because each PDCA turn that proves a change is then standardized, so the gain becomes the new floor rather than sliding back. Improvement compounds like a wheel climbing a slope: every proven turn lifts the standard, and the standard is the chock that keeps it from rolling back down. Without the Check step and the standardization in Act, you get motion without a ratchet — busy forever, climbing nothing.
- Why does standardized work matter to PDCA?
- Standardization is what makes a PDCA gain permanent. Once a turn of the wheel proves a better method, writing it into the standard is the chock behind the wheel — it stops the improvement from quietly reverting between shifts and people. PDCA and standardized work are two halves of one engine: improve, lock, improve, lock. Skip the standardize step in Act and the same "discovery" gets re-made months later by someone else.
- Where does PDCA usually go wrong?
- Almost every failure is one of four shortcuts, one per step. Plan breaks when a team jumps straight to a solution with no target or prediction, leaving nothing to Check against. Do breaks with a big-bang rollout instead of a small test, so you cannot cheaply undo it or read the signal. Check breaks when a team declares victory without measuring — confirmation bias with a nice slide. Act breaks when a proven change is never standardized, so the gain reverts. Each shortcut feels like progress while quietly stopping the wheel.
- Is PDCA only for big projects?
- No — the same four steps run at every altitude of an organization, just at different speeds. A daily huddle is a micro-turn of the wheel every shift: yesterday plan-vs-actual, a small countermeasure for today. A kaizen event is one deliberate, faster turn over a focused week. An A3 is a single, rigorous turn on one page. Hoshin (strategy deployment) is one slow annual turn of the whole company. They are nested wheels — small fast ones inside slow big ones, all turning the same way.
Related
Founder of Kaizumi, an AI-powered Lean training platform. More about Matthew →
Updated July 2026 · The clinic wait-time experiment and its figures are illustrative, created for teaching. PDCA — the Deming (or Shewhart) cycle — and the principle that standardization holds each gain follow the Toyota Production System and W. Edwards Deming’s work on continuous improvement.