Skip to content
LeanSuite

Practical guide · Kaizen and continuous improvement

PDCA: plan, do, check, act, with a worked shop-floor example

Short answer

PDCA is a four-step cycle for improving a process: plan a change, do it on a small scale, check the result against what you expected, then act by making it the standard or going round again. It is also called the Deming cycle.

PDCA stands for plan, do, check, act. In a plant it is used to test one change at a time. The plan names the problem, the baseline, the change and a predicted result. The do is a small trial. The check compares the result with the baseline, the prediction and the target. The act either writes the change into the standard or sends the team back to plan. The cycle only improves a process if the act ends in a standard, so the next turn starts from the new level.

Last reviewed:

What is PDCA?

The Lean Enterprise Institute defines PDCA as an improvement cycle based on the scientific method: propose a change in a process, implement it, measure the results and take appropriate action. It is also known as the Deming cycle or Deming wheel, and as PDSA when the third step is called study. It calls the cycle the foundation of kaizen.

On the shop floor that is one habit: write down what you expect, try the change small, look at the result, keep it only if it held. A kaizen event, an A3 report and a daily improvement board all run this loop under different names.

PDCA and PDSA compared
PDCAPDSA
Third stepCheck: did the plan meet its target?Study: why did the result differ from the prediction?
Named forDeming cycle or wheel, as taught in Japan in the 1950sCalled the Shewhart cycle by Deming, who wrote PDSA
Steps one, two and fourPlan, do, actPlan, do, act

Sources: Lean Enterprise Institute lexicon and Wikipedia, Plan-do-check-act.

The four phases of PDCA and what you write down in each

PDCA stands for plan, do, check, act, and the order matters. Plan names the problem, the change and a predicted result. Do runs the change as a small trial. Check compares what happened with the baseline and the prediction. Act makes the change the standard or sends the team back to plan.

The PDCA cycle

  1. P

    Plan: What is wrong, why do we think so, and what do we expect a change to do?

  2. D

    Do: Did the trial happen the way the plan said?

  3. C

    Check: What happened, compared with the baseline, the prediction and the target?

  4. A

    Act: Does the change become the standard, or do we go back to plan?

A wheel of four quarters running clockwise: plan, do, check and act. After act the wheel comes back to plan.

Read clockwise from the top left. The cycle has no end point: after act it comes back to plan, with the new standard as the starting line.
The four PDCA phases, the question each answers and what goes on the sheet
PhaseThe questionWhat you write down
PlanWhat is wrong, why do we think so, and what do we expect a change to do?The problem with a baseline number, the target, the suspected cause, the change, a predicted result, where and for how long the trial runs, and the measure.
DoDid the trial happen the way the plan said?What was actually done, on which line or shift, and every departure from the plan.
CheckWhat happened, compared with the baseline, the prediction and the target?The result, the three comparisons, the result split by reason or shift, and any side effect.
ActDoes the change become the standard, or do we go back to plan?The decision and why, the exact change to the standard work, who is trained, and the plan and prediction for the next turn.
  1. 1

    Plan: write a number for what the change should do

    In the Japanese version of the cycle that Moen and Norman describe, planning means defining the problem and hypothesizing possible causes and solutions. Get the cause from data, with a 5 Whys or a fishbone, not from the loudest opinion. Then write a number for what the change should do. A plan with no prediction cannot fail, so it teaches nothing.

  2. 2

    Do: a small trial, with departures noted

    One machine, one line, one shift, for a set number of runs. If the change turns out to be wrong, a small trial is cheap to undo. Note what went differently from the plan, because a trial that drifted cannot be checked against it.

  3. 3

    Check: a comparison, not a feeling

    Put the result next to the baseline and the prediction, and split it by reason, because a total can hide a change that worked beside a side effect that did not.

  4. 4

    Act: two exits, standardize or go round

    Moen and Norman give them as returning to the plan step when the results are unsatisfactory and standardizing the solution when they are satisfactory. The Lean Enterprise Institute adds how: the team standardizes the new method by updating the standardized work, so the improvement is stable.

At the Check: standardize, or go round again?

  1. Did the result beat the baseline?

    If no

    Go back to Plan. Write down what the check showed and change the plan, not the numbers.

  2. Did it land near the predicted result?

    If no

    Find out why before you standardize. A gap between prediction and result means the cause was not what the plan assumed.

  3. Did splitting the result by reason or shift show a side effect?

    If yes

    Put the side effect into the next plan, and standardize only what held.

  4. Every answer right

    Act: write the change into the standard work, train the people on it and use the result as the baseline for the next turn.

Three questions after a trial. If the result did not beat the baseline, go back to Plan. If it missed the predicted result, find out why before standardizing. If splitting the result showed a side effect, deal with it in the next plan. Answer all three well and Act writes the change into the standard work.

The check can send the team back at three points. Only a trial that clears all three becomes the standard.

How PDCA loops: the second turn starts from a standard

The circle on the poster hides the point of the cycle. A turn that ends in a written standard raises the starting point for the next one, so the baseline for turn two is the result of turn one, not the number you started with.

One cycle rarely closes a problem. Turn one removes the biggest cause and its check shows the next biggest. Turn two goes after that one, measured against the new standard. The standard work guide covers what a written standard holds.

Stop when the check shows the target met and the standard written. Then pick the next problem.

A worked PDCA example: first-hour scrap after a die change

Illustrative example. Press 4 scraps parts in the first hour after every die change. The team logs eight changes, with the reason for each scrapped part, and runs two turns of PDCA. The baseline is first-hour scrap per die change, which was 46, 52, 40, 55, 48, 44, 51 and 44 parts: 380 parts in eight changes, or 47.5 a change. The target is under 25.

Illustrative: the baseline, 380 scrapped parts over eight die changes, by reason
Reason from the scrap logParts in 8 changesParts a change
Shut height or feed pitch set wrong19023.75
Die not lubricated before the first stroke769.50
Everything else11414.25
Total38047.50

Turn 1: a written setup sequence

Illustrative: turn 1 as it sits on the PDCA sheet

Example numbers

  • Plan

    Entry
    Half of the scrap, 190 of 380 parts, is shut height or feed pitch set wrong, because each setter adjusts in their own order. Change: one written setup sequence with a gauge check of both before the first stroke. Prediction: two thirds of those parts go, so scrap falls to about (380 minus 127) / 8, which is 32 a change. Trial: the next six die changes on Press 4.

  • Do

    Entry
    Six die changes over two weeks. On the first the setter had to fetch the gauge from the tool crib, so it now lives at the press.

  • Check

    Entry
    Scrap per change: 33, 29, 35, 27, 31 and 30. That is 185 parts, 30.8 a change, down 35 percent from 47.5. Close to the prediction of 32, not under the target of 25. Shut height and feed pitch fell from 23.75 to 8.33 a change. Lubrication did not move: 9.50 to 10.33.

  • Act

    Entry
    The theory held, so adopt it. The sequence goes into the Press 4 setup standard and is taught on every shift. The target is still missed and the largest remaining cause is lubrication. Turn 2 starts from 30.8, not 47.5.
Turn 1 missed the target of 25 and was still adopted, because the check split the scrap by reason and the predicted cause had gone.

Turn 2: from the new standard

Illustrative: turn 2 as it sits on the PDCA sheet

Example numbers

  • Plan

    Entry
    Baseline is the new standard, 30.8 a change. Lubrication is 62 of the 185 parts, 10.33 a change. Change: add a lubrication check as a line in the sequence, before the first stroke. Prediction: lubrication scrap falls by about 70 percent, to about 3 a change, so total scrap goes to about 30.83 minus 7.23, which is 23.6. Target: still under 25.

  • Do

    Entry
    The next six die changes, run by all three shifts from the written standard.

  • Check

    Entry
    Scrap per change: 24, 22, 27, 21, 25 and 23. That is 142 parts, 23.7 a change. Under the target, and close to the prediction of 23.6. Lubrication fell to 3.17 a change.

  • Act

    Entry
    Adopt. The lubrication check joins the standard and the new baseline is 23.7. Turn 3 would go after the 11.83 a change under everything else, which the log does not split yet. Other presses are a separate cycle, because one press is one test.
Turn 2 plans from 30.8, the standard turn 1 left behind, and goes after the largest cause the turn 1 check found.
Illustrative: scrap per die change by reason, baseline and the two turns
ReasonBaseline (8 changes)Turn 1 (6 changes)Turn 2 (6 changes)
Shut height or feed pitch23.758.338.67
Lubrication9.5010.333.17
Everything else14.2512.1711.83
Total a change47.5030.8323.67

First-hour scrap per die change, before and after each turn

Example numbers

  • 47.5

  • 30.8

  • 23.7

  • Before PDCA

    8 die changes, 380 parts

  • Standard after turn 1

    6 die changes, 185 parts

  • Standard after turn 2

    6 die changes, 142 parts

  1. Turn 1: A written setup sequence with a gauge check of shut height and feed pitch.
  2. Turn 2: A lubrication check added to the same sequence.

First-hour scrap per die change on Press 4: 47.5 before PDCA, 30.8 after turn 1, 23.7 after turn 2, against a target of under 25.

Each bar is the level the written standard holds after a turn. Turn 2 started from 30.8, not 47.5. The dashed line is the target of under 25, first met after turn 2.

Turn 1 missed its target and still counted as a win, because the check split the result by reason. A pass or fail look at 30.8 against 25 would have thrown away a sequence that removed two thirds of its cause. Together the two turns took first-hour scrap from 47.5 to 23.7 a change, about half.

PDCA vs DMAIC vs A3 vs SDCA

They sit at different sizes and do different jobs.

PDCA next to DMAIC, A3 and SDCA

  • PDCA

    What it is
    A four-step loop for testing one change
    Use it when
    The cause is known well enough to try a fix small
    The check asks
    Did the change beat the baseline and the prediction?
    It ends with
    A new standard, or another turn

  • DMAIC

    What it is
    A five-phase project method: define, measure, analyze, improve, control
    Use it when
    The problem is large, the cause unclear and the data need real analysis
    The check asks
    Did the improvement hold?
    It ends with
    A control that keeps the gain

  • A3

    What it is
    A one-page report that carries a problem through a PDCA cycle
    Use it when
    One page is needed to think and coach on
    The check asks
    Did the countermeasure work, and what is left?
    It ends with
    A follow-up on the same sheet

  • SDCA

    What it is
    A loop that holds a standard in place: standardize, do, check, act
    Use it when
    A standard exists and you need to know it is being followed
    The check asks
    Is the process still running to the standard?
    It ends with
    A correction to the work or to the standard
Four tools, four jobs. PDCA tests one change, DMAIC runs a larger project, an A3 puts the thinking on one page and SDCA keeps a standard running.

A DMAIC project is the bigger frame, and an A3 is PDCA written on a page. To choose between them, see the problem solving methods guide or the lean method navigator.

SDCA is practitioner vocabulary. Training providers describe it as the maintenance loop that runs beside PDCA: PDCA moves a process to a better level and SDCA keeps it there. The term does not appear in the Lean Enterprise Institute entry or in Moen and Norman's history, so check how your own company uses it before you put it on a poster.

Common PDCA mistakes

Five PDCA mistakes and what to do about each

  • Mistake

    Skipping Check

    What goes wrong: The change is called a success because everyone worked hard. Nobody can say whether the number moved.

    Do this instead

    Block time for the check before the trial starts, and bring the baseline to it.

  • Mistake

    A plan with no prediction

    What goes wrong: Any result can be read as a win, so the team learns nothing about why the change worked or failed.

    Do this instead

    Write the expected number and the reason for it in Plan.

  • Mistake

    Act ends with no standard

    What goes wrong: The gain lives in people's heads and fades when they change shift.

    Do this instead

    Name the document that changes, the date, and who is trained.

  • Mistake

    Changing several things in one Do

    What goes wrong: If scrap falls, nobody knows which change did it.

    Do this instead

    One change per turn, or a trial design that separates them.

  • Mistake

    Stopping at one good run

    What goes wrong: A good week can be chance. The next shift brings the problem back.

    Do this instead

    Run enough repeats to see the spread, then test the other shift.

Most of the fixes are written before the trial starts: the check time, the prediction and the document the act will change.

PDCA templates and software

The free PDCA template is a one-page PDF and an Excel cycle sheet that checks each result against the goal. Six more full cycles with their numbers are in the PDCA examples, and the PDCA poster is a printable cycle for the line-side board.

In LeanSuite, Kaizen & Projects manages every improvement project in one visual hub, from a small kaizen event to a full DMAIC project. Lean Studio has 30 ready-made boards, including the 5 Whys, A3 and 8D that feed the plan step.

Continuous Improvement Software

Run PDCA cycles in LeanSuite

Use A3, DMAIC, PDCA, PDSA or just-do-it templates, or define your own method and phases.

See Continuous Improvement SoftwareBest continuous improvement software

Go deeper

This page is the short version. These guides cover each part in full.

All our articles on this subject are in the kaizen and continuous improvement topic.

Free templates and tools

Key terms

Lean glossary

Doing this in LeanSuite

See it on your own lines

Bring the problem you are working on to a 45-minute demo. Or look around with sample data first, or pilot LeanSuite on your own lines for 90 days, with the setup done by our team.

Book a 45-minute demoClick through the interactive demoStart a 90-day pilot

FAQ

Questions about pDCA

  1. It is the improvement cycle that stands for plan, do, check, act (the order is plan, do, check, act). Plan a change and predict its result, do it on a small scale, check the result against the baseline and the prediction, then act by standardizing the change or going back to plan. It is also called the Deming cycle.

  2. Plan, do, check, act. Some teams write PDSA, with study as the third step, as Deming preferred. The four steps are the same, and the Lean Enterprise Institute calls the cycle the foundation of continuous improvement, or kaizen.

  3. Not quite. Kaizen is the habit of continuous improvement, and PDCA is the loop a team runs to do it. The Lean Enterprise Institute calls PDCA the foundation of continuous improvement, or kaizen, so a kaizen event, an A3 and a daily improvement board all run PDCA under other names.

  4. The solution is identified in Plan, where the team picks the change and predicts its result, and implemented in Do, as a small trial. Act implements it for good, by writing it into the standard work, but only if Check showed the trial worked.

  5. Plan: define the problem, set a target, find the cause and choose a change with a predicted result. Do: try the change on a small scale and note what went differently. Check: compare the result with the baseline, the prediction and the target. Act: standardize the change if it worked, or go back to plan if not.

  6. Illustrative: Press 4 scrapped 47.5 parts in the first hour after each die change, half of them from wrong shut height or feed pitch. Turn 1 added a written setup sequence and predicted about 32. The trial gave 30.8, so the sequence became the standard. Turn 2 started from 30.8, added a lubrication check, predicted 23.6 and measured 23.7, under the target of 25.

  7. The steps are the same. In PDCA the third step, check, asks whether the plan met its target. In PDSA the third step, study, compares the result with a written prediction to learn why. Deming preferred the name PDSA. The PDSA guide on this site has the history.

  8. PDCA is a short loop for testing one change. DMAIC is a longer five-phase project method, define, measure, analyze, improve and control, used when the problem is large or the cause is unclear. Small PDCA trials often run inside the improve phase of a DMAIC project.

  9. Until the check shows the target met and the standard is written. Most changes take two or three turns: the first removes the biggest cause, the next goes after what the check showed was left. Each turn starts from the standard the last one produced.

Related guides and tools

More method guides

Or browse the topics, the full library and the lean glossary.

Rather see it on your own problem?

Bring the problem you are stuck on, and we will show you on a call how LeanSuite handles it.

Schedule Demo

Pass it on

We want LeanSuite to be the best place on the internet for lean help. If this was useful, send it to someone on your team or in your network who needs it.