Plan, Do, Check, Act: The PDCA Cycle in Everyday Work

Why good changes in a company so often fizzle out

A company decides on a change. In future, quotes should go out within three days, complaints should be acknowledged the same day, the workshop should be tidied every Friday. A few weeks later nobody knows exactly whether things have improved. Some stick to the new rule, others do not, and at some point the old way applies again. Not because the idea was bad, but because nobody checked whether it works.

The PDCA cycle is the best-known answer to this pattern. This leads to the guiding question of this article: how does a small company use the PDCA cycle in daily work so that changes become verified improvements? To answer it, we first clarify the four steps and their origin, look at what studies show about repeated cycles, weigh two ways of applying it, and finally describe one cycle over four weeks.

A change that is not checked is an opinion with an implementation deadline.

What is the PDCA cycle? The four steps

The PDCA cycle goes back to the statistician Walter Shewhart and was made known by W. Edwards Deming. It describes improvement as a loop of four steps. Shewhart originally described a cycle of three steps, Deming turned it into four (Saier 2017). Deming himself later preferred to speak of study instead of check, to emphasize that it is about learning and not about control. Both names, PDCA and PDSA, are common today and mean the same thing.

In the German title of this article, the fourth step is called improve; in other articles it is called act, or standardize or discard. It always means the same thing: what has proven itself becomes the standard, for example in a template or checklist. What has not proven itself is discarded, and the next attempt begins. As the article on continuous improvement on a small scale notes, the cycle treats every idea like a small experiment and not like a decision that has to be defended.

An analogy from the kitchen makes the idea tangible. Whoever cooks a new dish does not season according to the recipe and serve. They season, taste, compare with what they expected, and season again. Only when it is right do they write the amount into the recipe. The PDCA cycle is this tasting, applied to processes.

The thinking model: improvement as an experiment with a prediction

What is special about the PDCA cycle is less the sequence than the attitude behind it. Every change begins with a prediction: if we change this, this number will fall by roughly this much. Only the prediction makes the third step meaningful. Without it, any result can be interpreted as a success after the fact. With it, the team learns even from a failure, because a wrong prediction shows that the suspected cause was not right.

The cycle works on two levels. On a large scale, it is the framework of a quality management system. As the article on quality management without certification bureaucracy describes, the ISO 9001 standard is based on this cycle. On a small scale, it is the form of a single improvement attempt that is completed in a few weeks. For the daily work of a small company, the second level is the more important one.

What studies show about repeated improvement cycles

Most studies on the PDCA and PDSA cycle come from healthcare, where the method has been used for improvement projects for years. Bailey, Jennings, Alao, and colleagues (2025) describe a project in four cystic fibrosis care centers. The teams wanted to introduce a procedure for remotely identifying patients' social needs and connecting them to support services. For one year, each center ran a PDSA cycle every two weeks. In total there were 26 cycles. All four centers introduced the procedure, each adapted to its own processes.

Knippen, Woodcock, Papoutsakis, and Gonzales-Pacheco (2025) accompanied nutrition professionals at several sites in putting guidelines for treating gestational diabetes into practice. After a baseline measurement and training, each group set two goals and worked through them in repeated cycles. The analysis of treatment data showed improved process measures. In the interviews, participants highlighted above all the check step. It created structure and accountability that are otherwise lost in daily work. One professional described it as scheduling an appointment with ourselves.

For manufacturing companies there are the findings described in detail in the article on continuous improvement on a small scale. Kang, Li, and Kang (2025) found in data from 308 manufacturing plants in 15 countries and regions that suggestions strengthened continuous improvement above all when they were worked on in small groups, and that strongly centralized decisions weakened this effect. Ikica and Vrtodušić Hrgović (2026) showed in a survey of 244 managers in hotel companies that what mattered was the degree of actual implementation, while quality training had no measurable effect.

One large cycle or many small ones? Two ways of applying it

There are two readings of how the PDCA cycle should be used in a company. The first sees it as a framework for large undertakings: an annual plan, implementation over months, an evaluation at year-end, then adjustment. That is how a quality management system with management review works, and annual goals can be steered this way too. In its favor: large decisions also need a check.

The second sees it as a rhythm for many small experiments: each cycle takes two to four weeks, concerns a single change, and ends with a clear decision. The findings from healthcare speak for this. Short cycles provide feedback quickly, and a failed attempt costs little.

I consider both levels necessary, but in a clear relationship. The large cycle sets the direction once a year, for example in the planning described in the article on annual planning in the company. The actual improvement happens in the small cycles. Whoever only knows the large cycle learns once a year. Whoever runs many small ones learns every month.

One PDCA cycle in four weeks: how it works

The following procedure is intended for a single change. It requires one person responsible for the cycle, a number that can be collected without much effort, and a fixed date for the check step.

1. Week 0: Plan

   Describe the problem in one sentence, the suspected cause, and the change. Define the number and write down what value you expect after four weeks. Measure the starting value.

2. Weeks 1 to 3: Do

   Try the change on a small scale: one team, one type of order, one workplace. Note anything conspicuous, but do not change the experiment along the way.

3. End of week 3: Check

   Compare the measured value with the starting value and with your prediction. Ask those involved what they observed.

4. Week 4: Act

   Decide: standard, adjust, or discard. A standard is written into a template, checklist, or process page. An adjustment starts a new cycle.

The example shows how one cycle leads into the next. The first change sped up approval and in doing so made the next bottleneck visible. How a bottleneck is found is described in the article on process optimization step by step. Whoever already has a weekly meeting can anchor the check step there, as the article on weekly team planning provides.

The most common mistakes in the PDCA cycle

• No prediction: the change is implemented without saying beforehand what should change and by how much.
• Too much at once: several changes run at the same time, so nobody knows which one worked.
• Thinking too big: the change affects the whole company right away instead of one workplace or type of order.
• Checking without a date: the third step has no date and therefore never happens.
• Acting without a standard: what works stays in individual heads and disappears with the next vacation.

The last mistake has the greatest consequences. An improvement that is not written down no longer exists after a few months. How insights are recorded so that they take effect is described in the article on project closure and lessons learned.

Conclusion: how do changes become verified improvement?

The question was
a small company uses the PDCA cycle in daily work so that changes become verified improvements. The answer lies in three things: a prediction before every change, a fixed date for the check, and a standard at the end. The four steps are quickly explained. Their value only arises when all four are completed, and often and on a small scale.

As shown, the findings from healthcare and manufacturing point in the same direction. Short, repeated cycles work when they are actually carried out, and the check step creates the accountability that is otherwise missing in daily work. A company that completes one small cycle every month has a dozen verified answers after a year instead of a list of good intentions.

How to get started

1. Choose a problem that regularly causes frustration in the team and can be described with a number.
2. Fill in the PDCA card: problem, suspected cause, change, number, and prediction.
3. Measure the starting value and start the experiment on a small scale.
4. Put the check date three weeks out firmly in the calendar.
5. On the check date, decide explicitly: standard, adjust, or discard.