If the same finding is written again at the next audit, the corrective action closed the symptom but left the cause in place. In information security nonconformities, the most common reason for this loop is writing “staff carelessness” or “lack of training” in the root cause field and moving on. Below, correction, corrective action and root cause are first separated; then the 5 Whys and fishbone (Ishikawa) methods are applied step by step to a realistic access review finding, and how to measure effectiveness is shown.
The difference between correction, corrective action and root cause
ISO/IEC 27001:2022 Clause 10.2 requires that, when a nonconformity occurs, the organization first reacts (takes action to control and correct it, deals with the consequences), then evaluates the need for action to eliminate the causes so that it doesn’t recur or occur elsewhere, implements any action needed and reviews its effectiveness. The nature of the nonconformity, the actions taken and the results of corrective action are kept as documented information.
| Concept | Question | Example |
|---|---|---|
| Remediation | How do we fix this nonconformity now? | Closing the leaver accounts that remained open |
| Root cause | Why did this happen, and why wasn’t it noticed? | The leaver notification doesn’t reach the ERP; there’s no control mechanism |
| Corrective action | How do we prevent it from recurring? | Tie the HR–IT notification to a mandatory workflow, weekly reconciliation |
| Effectiveness verification | Did the action work? | No exceptions in reconciliation over the next two quarters |
The most often skipped part of Clause 10.2 is evaluating whether similar nonconformities exist elsewhere. If the finding was detected in one application, the other applications fed by the same process are checked too. The auditor finds one example. You scan the whole system.
The right problem statement for root cause analysis
The quality of the analysis is limited by the problem statement. A general statement like “access control is inadequate” pulls in every direction, and the analysis stays general. A good problem statement is measurable, dated and has a clear scope. Let the finding in the sample scenario be:
In the audit, it was found that for 3 of the 14 staff who left in the last six months, the ERP account remained active for more than 30 days after the leaving date. All Active Directory accounts were disabled on time. The quarterly ERP permissions review was not carried out in the last two quarters.
This statement carries three pieces of information: the problem is in the ERP, not AD; the problem isn’t continuous but in a specific period; and the detective control (the review) didn’t work either. So the analysis has two branches: why it happened and why it wasn’t noticed. Asking the two questions separately prevents getting stuck with only a preventive or only a detective solution. The numbers in the scenario are specific to this example and used to illustrate the method.
The 5 Whys method: applying it to a sample finding
5 Whys aims to move from the symptom to the process by asking “why?” again at each answer. Five isn’t a rule but an approximate depth; you stop when the answer reaches a process or management system gap. Every answer must be based on evidence, not assumption. The ticket queue, email distribution list, procedure revision date — all are opened and checked.
Branch A: Why did the account stay open?
- Why weren’t the ERP accounts closed? No closure request reached the ERP administrator. (Evidence: no tickets in the ERP queue for the three leavers concerned.)
- Why didn’t the request reach them? HR’s leaver notification goes by email only to the AD team; the ERP is managed by a separate team.
- Why isn’t the ERP team on the distribution list? ERP administration was handed over to an outsourcing firm a year ago; the leaving procedure wasn’t updated after this change.
- Why wasn’t the procedure updated? The service handover wasn’t treated as a change; its impact on the ISMS wasn’t assessed.
- Why wasn’t its impact assessed? No ISMS impact assessment step is defined for supplier changes.
Branch B: Why wasn’t the situation noticed?
- Why didn’t the quarterly review catch it? It wasn’t done in the last two quarters.
- Why wasn’t it done? The person doing the review changed roles; the task wasn’t assigned to anyone.
- Why wasn’t it assigned? Responsibility for the review is defined by a person’s name, not a role; there’s no calendar reminder.
Branch B stops at the third step because it reaches a management system gap there. Instead of stopping at the answer “staff forgot”, the analysis goes down to the structure that made forgetting possible (person-bound responsibility, no automatic tracking). This is a picture you may come across: the AD side works automatically, business applications stay with a separate team and the leaving chain breaks there.
Multi-factor analysis with fishbone (Ishikawa)
5 Whys follows a single chain. When the problem arises from a combination of several factors, the chain falls short. An Ishikawa diagram sorts possible causes into categories and makes the team scan every area. The classic 6M categories from manufacturing can be adapted to information security like this:
| Category | Possible cause in the sample scenario | Verified with evidence? |
|---|---|---|
| Method (process) | The leaving procedure doesn’t cover the ERP | Yes, the procedure revision was reviewed |
| People (role) | Review owner is tied to a person | Yes, no role definition in the RACI |
| System (tool) | No integration between the HR system and the ERP | Yes |
| Data | ERP user names don’t match staff ID numbers | Partly; some accounts are shared users |
| Measurement (monitoring) | Leaving-to-closure time isn’t measured | Yes |
| Environment (organization) | No ISMS impact assessment for the outsourcing handover | Yes |
The “data” row in the table didn’t come up in the 5 Whys chain: shared ERP accounts that don’t match staff ID numbers also make it harder to find a leaver’s account. That’s exactly the value of Ishikawa. The cause on each bone is a hypothesis until verified. Only causes verified with evidence are written in the corrective action record as root causes.
Corrective action plan and responsibilities
Each verified root cause has at least one corresponding action. Actions are defined with a role, a date and evidence of completion; sentences like “the relevant units will take the necessary measures” aren’t a plan.
- Leaving workflow: The leaving record in the HR system is tied to a workflow that opens separate tasks for AD and the ERP. Owner: IT process owner. Evidence: the workflow definition and task records for the first three leavers.
- Role-based review responsibility: The ERP permissions review is assigned to a role in the RACI; a calendar reminder and a backup owner are defined.
- Cleaning up shared accounts: ERP accounts that don’t belong to a person are inventoried, unneeded ones are closed and the rest are assigned owners.
- Weekly reconciliation: An automatic comparison is made between HR’s list of leavers and active accounts.
- Supplier change step: An ISMS impact assessment for supplier handovers is added to the procedure.
- Extent check: Whether the same gap exists in applications other than the ERP (e.g. a cloud-based CRM) is investigated.
The reconciliation step starts on the AD side with a simple script. The PowerShell example below lists the user names from HR’s leaver list that are still enabled (requires the ActiveDirectory module):
Import-Module ActiveDirectory
Import-Csv .\leavers.csv | ForEach-Object {
Get-ADUser -Filter "SamAccountName -eq '$($_.SamAccountName)'" -Properties LastLogonDate
} | Where-Object { $_.Enabled } |
Select-Object SamAccountName, Name, LastLogonDate |
Export-Csv .\left-but-enabled.csv -NoTypeInformation -Encoding UTF8
The first thing to check in the console is whether the user name column in the CSV is exactly the same as SamAccountName in AD; the HR list often carries first and last names or staff ID numbers, and rows that don’t match silently return nothing. On the ERP side, the same comparison is done by exporting the application’s user report. When the output file is dated and kept, it becomes both operating evidence and the data source for measuring effectiveness. How this evidence is documented within the periodic review is explained in the article access review evidence.
How is effectiveness verified?
An action being completed and being effective are different things. Completion shows the workflow went live; effectiveness shows the problem didn’t recur over a period. The effectiveness criterion should be written when the corrective action is opened, together with the action plan:
- Verification period: the two quarters after the action is completed.
- Measure 1: zero active accounts more than one working day after the leaving date in the weekly reconciliation reports.
- Measure 2: two quarterly ERP permissions reviews completed and approved on the planned dates.
- Measure 3: no similar findings in the other applications reviewed in the extent check, or any found recorded under a separate corrective action.
Verification is done not by the person who implemented the action but by someone independent (usually the internal auditor or ISMS manager), who adds it to the corrective action record with evidence. If the criterion isn’t met, the corrective action isn’t closed. The analysis is reopened. The difference between a control merely being designed and being operated throughout the period is covered in detail in the article control effectiveness testing.
Typical corrective action records the auditor rejects
- Root cause = repeating the finding: “The accounts weren’t closed because closing wasn’t done.”
- Corrective action = correction: Only the three accounts in the finding were closed; no change that prevents recurrence.
- “Staff will be trained”: An action written without proving the cause was lack of training, and with no measurable target.
- No effectiveness criterion: The corrective action was closed on the day the action was completed.
- Extent not checked: Other systems in the same process weren’t reviewed.
Whether a finding is classified as major or minor doesn’t change the logic of root cause analysis; what changes is when and how the certification body verifies it. For what the classes mean, see the article major vs minor nonconformity. For support analyzing findings raised after a certification audit, see our post-audit nonconformity closure and corrective action consulting, and for integrating the internal audit cycle with the corrective action process, our internal audit and corrective action management service.
Frequently asked questions
Is root cause analysis needed for every nonconformity?
ISO/IEC 27001:2022 Clause 10.2 requires evaluating, for every nonconformity, the need for action to determine and eliminate its causes. The depth of the analysis is adjusted to the finding’s impact; a short 5 Whys is enough for a single error, while multi-factor analysis with Ishikawa suits systemic findings.
Do you always have to ask five times in the 5 Whys method?
No. Five is an approximate depth. You stop when the answer reaches a process or management system gap that can be changed. Supporting every answer with evidence matters more than the number of questions.
When can a corrective action be closed?
After the actions are completed and the predefined effectiveness criterion has been met during the verification period. Corrective actions closed on the day the action is completed are questioned in audits because their effectiveness hasn’t been shown.
Can lack of training be written as the root cause?
It can, but it must be verified with evidence: it must be shown that the person didn’t know the relevant responsibility and that there’s no training record. In most cases, a gap in the process or role definition that made the error possible is a more accurate root cause.