A year after certification, the security policies look exactly as they did on audit day, the risk register has the same rows, and the only change is the dates on the cover pages. During that year the company rolled out two new systems and switched email providers, yet the management system documents show no trace of either. On paper the system exists; in practice it has stopped.
The PDCA cycle, short for plan, do, check, act, is a simple model of the difference between a living system and a dead one. Management students learn it as a quality improvement tool, while in an information security management system (ISMS) it is a way of reading the requirements of ISO/IEC 27001, even though the standard has not named it explicitly since the 2013 edition. Every turn of the cycle should leave a trace you can show: a revised risk assessment, a measurement result, a closed corrective action.
What is the PDCA cycle?
The PDCA cycle is a four-step model of continual improvement: plan a change (Plan), carry it out (Do), check the results (Check) and, based on them, either adjust the way you work or lock the solution in (Act). After the last step the cycle starts again, this time with what the previous turn taught you.
The Japanese version of the cycle from 1951, as cited by Ronald Moen and Clifford Norman, covered defining a problem and hypothesising possible causes and solutions, implementing a solution, evaluating the results, and returning to the plan step if the results were unsatisfactory or standardising the solution if they were satisfactory (R. D. Moen, C. L. Norman, “Circling Back”, Quality Progress, November 2010).
The model does not tell you what to plan or how to measure. It is a frame that enforces an order: first the goal and the plan, then the action, then the evidence, then the decision. In information security that order protects against two extremes: deploying controls without checking whether they work, and endless analysis after which nothing changes.
The word “cycle” matters more than the letters. Implementing a system is one turn; a management system exists only when the turns repeat and each one starts from the conclusions of the last.
Where does the PDCA cycle come from, and why did Deming speak of PDSA?
The cycle grew out of Walter Shewhart’s work, was developed by W. Edwards Deming and, according to Masaaki Imai’s account, was given its PDCA form by Japanese executives; Deming himself preferred the PDSA version, in which the third step is “Study”, meaning examining the results.
According to Moen and Norman, in his 1939 book “Statistical Method from the Viewpoint of Quality Control” Shewhart described three steps, specification, production and inspection, which he later closed into a circle (Moen and Norman, 2010). Deming presented his own version in 1950 at an eight-day seminar in Japan organised by the Japanese Union of Scientists and Engineers (JUSE), and his wheel connected design, production, sales and research (Moen and Norman, 2010). The same authors cite Masaaki Imai’s account that Japanese executives recast the Deming wheel from the JUSE seminar into the PDCA cycle.
Deming consistently distinguished his version from PDCA. The W. Edwards Deming Institute stresses that he emphasised PDSA, with Study rather than Check: checking focuses on whether an implementation succeeded or failed, while studying compares predictions with actual results in order to revise the theory (The W. Edwards Deming Institute, “The PDSA Cycle”). Moen and Norman add that Deming warned that the English word “check” also means “to hold back”.
For an ISMS the distinction has practical meaning. A Check step reduced to ticking off that a control exists teaches nothing. A Study step starts by asking what you expected and ends with a conclusion about what to change in the plan.
Does ISO/IEC 27001 require you to use the PDCA cycle?
Not by name: since the 2013 edition, ISO/IEC 27001 has not mentioned the PDCA cycle explicitly, but its clause structure corresponds to it, with clause 6 covering planning, 8 operation, 9 performance evaluation and 10 improvement.
The requirement is continual improvement. In the 2022 edition, clause 10 opens with 10.1 on continual improvement, followed by 10.2 on nonconformity and corrective action; in the 2013 edition the order was reversed. The new order captures the point of the clause well: correcting mistakes is part of improvement, not the other way round.
Since the standard does not name PDCA, it does not impose that model either. You can describe your system with the PDCA cycle or another scheme, as long as the system plans, operates, evaluates its results and improves. PDCA nevertheless remains a convenient way to explain to a team why the clauses of the standard exist, when at first glance they look like separate obligations.
An overview of all the requirements of the standard is in the complete guide to ISO 27001, and a short definition in the ISO 27001 glossary entry. Other articles on the standard are collected in the ISO 27001 knowledge base.
What does the Plan stage look like in your ISMS?
In an ISMS, the Plan stage means deciding what the system must protect against, what it wants to achieve and by what means: in practice, a risk assessment, risk treatment decisions and information security objectives with measures and deadlines.
A plan is not written once. It changes with the company: a new system, a new supplier, a new location or new legislation all call for another look at the risks. The requirements themselves change too. Amendment ISO/IEC 27001:2022/Amd 1:2024 of 23 February 2024 added to clause 4.1 a requirement to determine whether climate change is a relevant issue for the organisation (IEC Webstore, ISO/IEC 27001:2022/AMD1:2024).
A good plan has a feature that is easy to overlook: it can be checked. An objective such as “improve email security” will not survive the Check stage, because nobody knows what to measure. An objective such as “all administrative accounts covered by multi-factor authentication by the end of the quarter” can be verified with a single report from the identity system.
What does the Do stage mean in an information security management system?
The Do stage is carrying out the plan in day-to-day work: deploying controls, following procedures and delivering the actions in the risk treatment plan in a way that leaves a trace.
The trap at this stage is mistaking a document for an action. A procedure for granting access rights described in a policy is a plan. A request with a line manager’s approval and the date the account was created is evidence that the procedure works, and without such traces the Check stage has nothing to assess.
Do also covers people. Training after which staff know how to report an incident is an action; an attendance list with no check on whether anyone reported anything afterwards is only a formal trace.
In a mature system, some actions run automatically: an access review generated from the identity system, a backup report, scan results. Automation does not remove the need for decisions, because someone still has to read the report and respond to a deviation.
How do you check the effectiveness of controls in the Check stage?
The Check stage answers the question of whether controls work, not merely whether they exist; in ISO/IEC 27001 it corresponds to clause 9 on performance evaluation, and in practice it consists of measurements, internal audits and a review by management.
A measurement makes sense when it has a threshold and an owner. A metric such as “open critical vulnerabilities older than the agreed deadline”, with a threshold of zero and a named owner, leads to a decision, whereas a chart without a threshold leads to a slide. This is where Deming’s lesson helps: before measuring, write down what you expect, and afterwards compare the result with your prediction.
For example, before changing the leavers process, the team records the assumption that after a quarter there will be no active accounts belonging to people who have left the company. If the access review finds such accounts, the result does not simply say that “the control does not work”. It shows which part of the assumption was wrong: the data from HR, the deactivation deadline or the assigned responsibility, and that is the part to change in the next plan.
An internal audit checks whether what has been described actually happens. How to prepare it so that it is a tool rather than a ritual is covered in the article on the ISO 27001 internal audit.
For entities covered by the NIS2 Directive, assessing effectiveness is an obligation. Article 21(2)(f) lists among the minimum measures “policies and procedures to assess the effectiveness of cybersecurity risk-management measures”, and Article 20(1) requires management bodies to approve those measures and oversee their implementation (Directive (EU) 2022/2555). The directive applies through national law; in Poland, through the Act on the National Cybersecurity System, as amended by the act of 23 January 2026 (Journal of Laws 2026, item 252), in force since 3 April 2026 (Sejm of the Republic of Poland, ELI). Legal position as of 15 October 2026.
Key points:
- ISO/IEC 27001 does not name PDCA, but clauses 6, 8, 9 and 10 line up with its four steps.
- The Check stage assesses whether controls work, not whether they exist.
- Every turn of the cycle should leave evidence you can show an auditor.
What happens in the Act stage?
In the Act stage, the organisation turns the conclusions of the evaluation into decisions: it deals with nonconformities, takes corrective action and changes the plan for the next turn; in ISO/IEC 27001 this corresponds to clause 10 on improvement.
A corrective action differs from a simple correction in that it targets the cause. A correction closes a specific former employee’s account, while a corrective action changes the leavers process so that future accounts close without reminders. If the nonconformity register shows the same items in successive audits, the organisation is making corrections instead of corrective actions.
Act also has a second side: locking in what worked. In the Japanese description of PDCA cited by Moen and Norman, satisfactory results lead to standardising the solution. In an ISMS, that means updating a procedure, a template or a baseline configuration so that the solution does not depend on one person.
After the Act stage, the conclusions return to the Plan stage as a revised risk assessment, new objectives and improved controls. Without that step, the cycle becomes a straight line that ends in a report.
What makes your PDCA cycle spin in place?
The cycle spins in place when one of its stages exists only on paper and the conclusions of the evaluation do not return to the plan.
The first cause is a plan without measures: objectives worded so that they cannot be checked, so the Check stage evaluates activity rather than results. The second is a Check stage reduced to a calendar, in which the audit happens because a year has passed and its findings have no owners or deadlines.
The third cause is a management review run as a presentation rather than a decision. If the review produces no changes to budget, priorities or risk acceptance, management has been informed but the system has not improved. The fourth is a lack of connection with the rest of the business: the ISMS cycle runs alongside IT changes, so a new production system reaches the risk assessment only at the next audit.
In all four cases, there is no trace of the transition between stages. If you cannot point to which decision in the plan follows from which evaluation result, the cycle exists only in a slide deck.
How do you map out one turn of the PDCA cycle in an ISMS?
It is worth mapping out one turn of the cycle so that, at every stage, it is clear which question it asks and which evidence it leaves.
| Stage | ISO/IEC 27001:2022 clause | Control question | Evidence left behind | Sign that the cycle has stalled |
|---|---|---|---|---|
| Plan | 6 — planning | Do the risks and objectives match the company as it is today? | dated risk assessment and treatment plan, objectives with a measure and a deadline | risk register unchanged since the last audit |
| Do | 8 — operation | Do the planned controls work in day-to-day practice? | requests, change logs, system reports, training confirmations | the procedure exists, but nobody can show a trace of it |
| Check | 9 — performance evaluation | Do we know the controls work, not just that they exist? | measurement results with thresholds, internal audit report, management review minutes | measurements with no thresholds and no owner |
| Act | 10 — improvement | What did we change after what we learned? | register of nonconformities and corrective actions with status, changes to the plan | the same nonconformities in successive audits |
In a pre-audit self-assessment, each warning sign in the table points to the place where a turn of the cycle broke off, and that is where to start putting things right.
The cycle does not need a single rhythm. A turn of the whole system is set by internal audits and management reviews, while smaller turns cover individual controls: a monthly access review has its own Plan, Do, Check and Act.
How can nFlo help you improve your ISMS?
In the governance, risk and compliance area, nFlo provides the ISMS Review, Audit and Advisory service: a review of the system documentation (policy, statement of applicability, procedures, registers), evidence-based verification of how effectively the implemented controls work, and independent internal audits. It is the Check stage carried out from the outside, with conclusions ready to turn into Act.
For organisations that are still building their system, we help through the ISO 27001 - Implementation and Certification Preparation service: from gap analysis, through ISMS documentation and risk analysis aligned with ISO 27005, to the internal audit, management review and support during the certification audit. The certificate is granted by an accredited certification body.
If the cycle in your ISMS has stalled at the documentation stage and your cybersecurity programme needs evidence of effectiveness, book a conversation about reviewing your system before the next audit.
What else should you know about the PDCA cycle?
Is the PDCA cycle mandatory in ISO 27001?
Not by name. Since the 2013 edition, ISO/IEC 27001 has not mentioned the PDCA cycle explicitly, but it requires continual improvement of the system, and the structure of its clauses on planning, operation, performance evaluation and improvement corresponds to the four steps of the cycle.
How does PDCA differ from PDSA?
PDSA is the version W. Edwards Deming promoted: the third step is Study, meaning examining the results and comparing them with the prediction. In an ISMS this means measurements that teach rather than merely confirm.
Who is responsible for the PDCA cycle in an organisation?
Top management is responsible for the system, while risk and control owners run the individual steps, coordinated by the person responsible for the ISMS. In entities covered by NIS2, management bodies approve risk-management measures and oversee their implementation; the directive applies through national law. Legal position as of 15 October 2026.
How often should the PDCA cycle turn?
There is no single pace. The measure is not frequency but whether each turn ends with a change to the plan.
Is the PDCA cycle used only in ISO 27001?
No. The cycle comes from quality management in industry, and in information security it also structures the NIS2 requirement to assess the effectiveness of risk-management measures.
