Understanding the PDCA Cycle in ISO 9001

Ask anyone who has sat through a quality induction to name the four steps of PDCA, and they will rattle them off without blinking. Plan, Do, Check, Act. It is on the poster by the coffee machine and on the first slide of every training deck ever built.

Most quality systems, however, overly focus on the first two at the expense of the second two.

They “plan”, sometimes beautifully. They “do”, often heroically. Then “check” becomes a spreadsheet nobody opens, and “act” becomes a meeting where everyone nods and agrees things are fine. The cycle stops being a cycle and turns into a straight line that runs off the edge of the desk.

In this article we’ll do three things. First, we’ll explain what PDCA actually is. Next we’ll show how ISO 9001 is built on top of it and lastly we’ll spend some time on the part that most companies miss: why the last 2 steps of the cycle is where quality systems quietly die, and what you can do to keep yours alive.

Where PDCA came from, briefly

PDCA is old. Not dinosaur old but still older than you might think. In fact, Walter Shewhart sketched the idea in the 1930s, and W. Edwards Deming carried it out into the world and made it famous, which is why you’ll sometimes hear it called “the Deming cycle”.

One detail from Deming’s work is worth penciling down into your Quality Policy. Late in his life he W. E. Deming stopped saying “Check” and started saying “Study”. As in “Plan-Do-Study-Act”. He felt “Check” had given people the impression this step of the processes could be quickly ticked off at a glance. “Study” asks you to actually sit with the results and understand them. Hold on to that word. It turns out to be the whole ballgame.

The four steps are simple enough to explain over coffee:

• Plan. Decide what you are trying to achieve, and how you will get there.
• Do. Execute the plan.
• Check. Compare what actually happened with what you expected.
• Act. Keep what worked, fix what did not, and feed the refinements back into the next plan.

That is the whole engine. Everything else is detail.

ISO 9001 is one big PDCA loop

To understand the management system that sits underneath this cycle, see our introduction to what is a QMS.

If you have read ISO 9001:2015, you have already come across PDCA, whether you noticed it or not. The standard is organised around it, and the structure lines up almost one to one.

Every article on this subject shows you that table for a reason. It tells you something the clause list on its own doesn’t quite portray. It is easy to read the clauses as a filing cabinet of separate requirements. Look again, and you will see a loop. Clause 10 hands back to Clause 6. The output of your improvement work becomes the input of your next round of planning.

When people say the standard feels like a pile of disconnected boxes to tick, it is almost always because someone handed them the boxes without the loop that joins them up. See the loop, and the boxes start to make sense.

Now, the part where most companies trip.

Check and Act are where it falls apart

“Plan” and “Do” are the visible half of the cycle. They produce things you can point at: a documented procedure, a delivered order, a product on a pallet. People are good at the visible half, because the visible half is obviously work, and work gets noticed.

“Check” and “Act” are the introverted siblings. Quiet work is easy to let slide, for three very human reasons. Measuring is uncomfortable, because it can tell you that you were wrong. Nobody gets a pat on the back for discovering a problem. And “Act” always has to compete with today’s fires, which are louder and feel more urgent than last month’s data.

So the second half of the cycle withers, usually without anyone deciding to let it. You can spot it from a mile off once you know the signs:

• Data gets collected faithfully and then never looked at.
• Internal audits only ever seem to find formatting problems in documents.
• Management review is an annual ritual, run off much the same slide deck as last year.
• Corrective actions replace the one faulty unit and close the ticket, while the process that produced it carries on untouched.
• There is a wall of KPIs, and not one of them has ever caused anyone to change a decision.

Here is why this is more than untidy. A system that only “plans” and “does” cannot improve. It can only repeat. It will continue to make the same product, with the same defects, at the same cost, until something outside of it forces a change. That is what W. E. Deming was reaching for with “Study.” “Checking” is glancing at the result. “Studying” is understanding it well enough to do something about it.

PDCA is not one loop. It is many.

There is a second reason PDCA gets misunderstood. People picture a single cycle turning slowly once a year, in sync with the management review. A working quality system runs many loops at once, at different speeds.

• An operator checking a part against a gauge and adjusting the machine is running a PDCA loop that takes seconds.
• A team looking at scrap figures at the end of each week is running one that takes days.
• Management reviewing performance against objectives is running one that takes months.
• A corrective action opened after a customer complaint is a loop of its own, running on its own clock until it is closed and proven.

They are nested, like gears of different sizes. The fast little loops on the shop floor feed the slower ones above them. Furthermore, the slow loops set the direction for the fast ones. When we hear from a company that their PDCA cycle is the annual review, we know straight away that most of the real improvement in their business is happening by accident, if it is happening at all. An annual loop takes far too long to catch anything while it’s still relevant.

What an auditor actually looks for

We build these systems, and we also audit them, so we can confidently share what it looks like from the auditor’s side of the table. Our work focuses on the connection between the management system, the product and the evidence needed to demonstrate that the system works. Learn more about Robur Systems.

It usually takes less than an hour to be able to tell whether a PDCA loop is real or just decorative. The tell is always the same, and it lives in the second half of the cycle.

An auditor who knows the job does not ask to see your procedures and stop there. Nope. Evidence is always required to ensure that the policies and documentation are followed

• “Can you show me a decision you changed because of something the data told you?”
• “Show me a corrective action that changed the process, not just replaced the broken part.”
• “Was there an objective you reset or dropped at the last management review, and why?”

Good answers arrive with evidence attached, and they arrive unhurried, because the loop is a habit. Weak answers come with a pause, then a search for a document that turns out to describe the intention rather than the act. That gap, between a system that is documented and a system that is true, is exactly where audit findings live. It is also, as it happens, where most of the value of the standard is either captured or thrown away.

How to make the second half real

None of this is hard to fix. It mostly takes deciding to, and then building a few small habits that make “Check” and “Act” as routine as “Plan” and “Do”.

Give every loop a cadence. Decide how often each loop should turn, from the shop-floor checks and data reviews to the team meetings and management reviews. Put each one in the calendar as a standing commitment rather than a when-we-get-to-it. A loop with no rhythm does not turn.

Give every loop an owner. Not a department, a person. “Quality owns the corrective action process” means, in practice, that no one owns it. A name creates follow-through.

Make each phase leave a trace. “Plan” leaves objectives. “Do” leaves records. “Check” leaves the analysis, not just the raw numbers. “Act” leaves the change you made and the proof it worked. If a phase leaves nothing behind, it probably did not really happen, and an auditor will reach the same conclusion.

Make Check honest. Measure against what you predicted, not against a target you can quietly move afterwards. And build a culture where writing down “we were wrong” is treated as a good day’s work, because it is. Every honest Check is a problem caught while it is still cheap to fix.

The half that makes it worth doing

PDCA is not a diagram for the wall. It is a habit of turning experience into improvement, over and over, at every level of the business. The poster by the coffee machine has it right as far as it goes. The trouble is that most systems live on the left-hand side of it, in “Plan” and “Do”, and treat the right-hand side as paperwork.

The second half is the half that makes the business improve. Get “Check” and “Act” turning honestly, on a rhythm, with someone’s name on each loop, and ISO 9001 stops feeling like a standard you comply with and starts working like the improvement engine it was always meant to be.

If your “Check” and “Act” have gone quiet, that is the best place to start. Pick one loop this week, give it a cadence and an owner, and watch what it turns up. You will almost certainly discover an opportunity that has been sitting there the whole time.

Want to explore more defence quality topics?

→ AQAP 2110 and its companion standards
→ What is a QMS?
→ Five things we learned at MSPO 2026
→ Back to the Robur Systems blog

Need help assessing your own QMS? Let’s talk.