The PDCA cycle (Plan, Do, Check, Act)

In one paragraph

A loop for improving anything: plan a change, make it small, check what actually happened, then either adopt it or try something else — and go round again.

Plan — identify the problem, work out what you think is causing it, and decide on a change and what result would count as success. Do — make the change, small and contained, so it can be undone. Check (or Study) — compare what happened with what you predicted. Act — adopt it, adapt it, or abandon it. Then start again.

Deming, who popularised it, preferred Study to Check, because Check implies looking for whether it worked while Study implies learning from the gap between prediction and result — including when it did work for reasons you did not expect.

Why the loop matters more than the steps

The steps are obvious. What businesses actually skip is the *repeat*. Improvement is treated as a project with an end, so a change is made, declared successful and never examined again. PDCA's claim is that improvement is a permanent habit of small cycles, not an initiative.

The other commonly skipped part is predicting the result *before* you make the change. Without a prediction, Study becomes a search for evidence that you were right.

How to actually use it

• Pick one irritation this week — a recurring error, a slow handover. Write down the change and what you expect to happen, try it for a fortnight, then compare.
• Keep the cycle small enough to reverse. Large changes cannot be studied because too many things moved at once.

Where it breaks down

• Suited to incremental improvement of an existing process. It will not produce a step change or tell you the process should not exist.
• It needs measurement. Where nothing is counted, Check collapses into an opinion about whether things feel better.

If you are being examined on it

Link it to kaizen and to quality assurance rather than control — it is prevention and improvement, not inspection.

More on operations