If continuous improvement has a heart, it is the PDCA cycle. Conceived by Walter Shewhart and popularized by W. Edwards Deming, it has become a common language for quality professionals worldwide.
From Shewhart to Deming
The cycle has its roots in Shewhart’s work on statistical quality control. He saw production as an application of the scientific method: specify what is wanted, produce it, then inspect in order to learn from what happened. Deming introduced the idea in his lectures in Japan, where it took the shape of the four steps known today. In his later work, Deming preferred the name PDSA, replacing "Check" with "Study", to stress that the aim is not merely to inspect results but to understand their causes and learn from them.
The four stages
- Plan: define the problem or opportunity, understand the current situation with data, set a measurable goal and an action plan. At this stage it helps to analyse root causes and to write down an explicit prediction of what will happen once the solution is applied.
- Do: implement the solution on a small scale, as a trial, and document what happens, including difficulties and unexpected effects.
- Check: compare the results with the goal and with the recorded prediction. Did things improve? Why? What have we learned?
- Act: if the solution works, make it the new standard and roll it out; if not, learn from the trial and start a new cycle.
Once a new solution is adopted, it has to be sustained: standardize the way of working, train the people concerned and keep tracking the indicator, so that performance does not slip back to where it was. This consolidation phase is sometimes described as the SDCA cycle: Standardize, Do, Check, Act.
A practical example
An administrative department noticed that processing requests took too long. In the Plan stage, the team measured the current lead time, identified the causes of delay and set a reduction target. In the Do stage, a unified request form was tested in one office. In the Check stage, results were compared with the other offices. Once the improvement was confirmed, the form was adopted everywhere and written into the procedures.
The details of this example reveal the logic of the cycle. The initial measurement showed that much of the delay came from incomplete files being sent back to applicants, so the proposed solution was a unified form with a clear list of required documents. A single office served as the trial site, so that it could be compared with a similar office where nothing had changed. In the Check stage, the team did not settle for the average lead time: it also tracked the number of incomplete files and gathered staff feedback, which brought to light a comment on the wording of some fields, corrected before roll-out. A second cycle was then launched to tackle another cause of delay.
How to launch a first cycle
The "Model for Improvement", widely used in the health sector, asks three questions before any cycle: What are we trying to accomplish? How will we know that a change is an improvement? What change can we make that will result in improvement? From there, the following steps apply:
- Choose a limited, tangible problem that users or staff experience and that can be measured.
- Form a small team that includes the people who actually do the work.
- Collect data on the current situation before proposing any solution.
- Set a measurable target and write down a prediction of the change’s effect.
- Test the change on a small scale and for a short period.
- Compare the results with the target and the prediction, then decide: adopt, adapt or abandon the change.
- Record and share the lesson learned, then start the next cycle.
Early cycles are best kept short, measured in days or weeks rather than months, so that the team sees results quickly and gains confidence in the method.
Common mistakes
- Jumping to action without understanding the problem.
- Skipping the Check stage and relying on impressions.
- Treating the cycle as a project that ends, when it is a loop that repeats.
- Rolling a solution out across the whole organization before testing it on a small scale.
- Changing several factors at once, so that it becomes impossible to tell which one produced the effect.
- Choosing an overly broad problem for the first cycle, so that the work drags on and enthusiasm fades.
- Neglecting standardization, so that performance slips back within a few weeks.
The power of PDCA lies in its simplicity: it makes improvement a methodical habit rather than an exceptional event.
PDCA in management systems
International management system standards follow the same logic: plan the system, operate it, evaluate its performance through audit and review, then improve.
In ISO 9001, for example, Plan corresponds to clause 6 on planning, Do to clauses 7 and 8 on support and operation, Check to clause 9 on performance evaluation and Act to clause 10 on improvement, with leadership (clause 5) at the centre of the cycle, fed by an understanding of the organization’s context (clause 4) and of the needs of interested parties. The cycle thus works at several levels: solving an everyday problem, running an improvement project and managing the whole system.
