Security Compliance

ISO 27001 Implementation Guide: From Scope to Audit

Photo: unknown (CC0 1.0)
In this article6 sections

What this covers and who it is for

ISO/IEC 27001 implementation means building an information security management system (ISMS) that can pass a certification audit: a defined scope, a repeatable risk assessment, a Statement of Applicability, the Annex A controls you decided to run, and the records proving those controls operate. Certification is issued against ISO/IEC 27001:2022. Its requirements sit in Clauses 4 to 10; its Annex A holds 93 controls across four themes — organizational (A.5), people (A.6), physical (A.7) and technological (A.8).

This hub is written for the person who owns the project rather than the certificate: the IT manager at a 60-person manufacturer handed the job in a Tuesday meeting, the security lead at a SaaS company whose largest customer put ISO 27001 in the contract, the MSP or consultant building an ISMS on a client’s behalf, and the compliance manager tracking a Stage 2 date. It assumes certification is already a requirement and you now need to build the thing without stalling the business.

Every section maps to a clause or an Annex A control, so you can trace a requirement to a record and a control to evidence. Topics that deserve their own treatment — risk methodology, incident handling, supplier security — are linked at the end.

The short version

If you read nothing else, this is the sequence. Each step produces a document or record the auditor will ask for.

  1. Define the scope (Clause 4.3). One page: what is in, what is out, why, and where the boundary leaks.
  2. Put leadership on record (Clause 5.1, 5.2). A signed policy and a named owner beat a steering-committee slide.
  3. Inventory assets and obligations (Clause 4.2, A.5.9). Systems, data, suppliers, and the legal and contractual requirements you must satisfy.
  4. Assess risk (Clause 6.1.2). A method with defined scales, criteria, owners and a review cadence, applied the same way every time.
  5. Treat risk and write the Statement of Applicability (Clause 6.1.3). Every Annex A control gets a decision and a justification.
  6. Set measurable objectives (Clause 6.2) and put the operating rhythm in place (Clause 8.1).
  7. Implement controls and capture evidence as you go. Retrospective evidence is the most common failure.
  8. Train people and record it (Clause 7.3, A.6.3).
  9. Run an internal audit (Clause 9.2) and a management review (Clause 9.3).
  10. Close findings (Clause 10.1), then book Stage 1 and Stage 2 with an accredited certification body.

The order matters less than the linkage. An auditor is not checking that you own a file called “risk assessment”; they are checking that the risk assessment drove the SoA, the SoA drove the controls, and the controls produced records.

Key concepts you need first

A management system, not a control checklist. Clauses 4 to 10 follow a plan–do–check–act loop: context (4), leadership (5), planning (6), support (7), operation (8), performance evaluation (9), improvement (10). Annex A controls sit inside that loop. This is why a team that “implements all the controls” but never audits itself, never reviews objectives and never records a corrective action fails — those are Clause 9 and 10 requirements, audited directly.

Risk assessment and risk treatment are different steps. Clause 6.1.2 is assessment: identify risks, analyze them against defined scales, evaluate them against an acceptance criterion, and record an owner for each. Clause 6.1.3 is treatment: select options, determine the controls needed, compare them with Annex A, produce the Statement of Applicability, and get the risk owner to approve the residual risk. A missing approval line is the easiest finding to raise, because the gap is visible on the page.

The Statement of Applicability is a decision record. Required by Clause 6.1.3(d), the SoA lists every Annex A control with an applicability decision, a justification, and implementation status. It is not a scorecard and not a marketing asset. An exclusion is legitimate only when the risk assessment justifies it — excluding A.7.4 physical security monitoring because the scope contains no facilities you operate, for example. Excluding A.8.11 data masking with the justification “not applicable” and nothing else is the row an auditor picks first.

2022 versus 2013. Templates written before the 2022 revision are organized by 14 domains and 114 controls. The current standard restructures these into four themes and 93 controls and adds entries such as A.5.7 threat intelligence, A.8.9 configuration management, A.8.10 information deletion, A.8.11 data masking and A.8.16 monitoring activities. Rebuild the SoA from the 2022 Annex A rather than mapping old rows across; a half-mapped SoA is worse than a fresh one.

Evidence is the product. Clause 7.5 requires the documented information the standard calls for, plus whatever the organization decides it needs for the ISMS to be effective. In practice that means policy, scope, risk assessment and treatment records, the SoA, internal audit results, management review outputs, nonconformity and corrective action records, and operational evidence that controls actually ran. Build the trail as you build the control. Reconstructing a year of access reviews and patch reports the week before an audit produces thin, unconvincing artifacts.

How to apply this

Step 1: Write a scope you can defend (Clause 4.3). Name the legal entity, locations, products or services, supporting infrastructure, and the interfaces to anything outside the boundary. If a payroll provider or data center sits out of scope, say so and say how the risk is managed. A scope that reads “all information systems of the company” while the company runs three subsidiaries creates a scoping finding at Stage 1 and an expensive re-scope afterwards.

Step 2: Make the risk method repeatable (Clause 6.1.2). Pick a five-point likelihood and impact scale, define what each number means, set the acceptance threshold, and record the criteria before assessing anything. Assess at the level where decisions get made — a business process or an asset group, not “the network”. Each entry needs an owner who can accept residual risk, because the auditor will trace that acceptance back to a name.

Step 3: Trace one risk all the way to evidence. Suppose a 120-person B2B SaaS company scopes its customer platform and support operations. The risk register carries: misconfigured object storage makes customer report exports publicly readable. Likelihood scores 4 and impact 5 against the defined criteria, the score exceeds the acceptance threshold, and the treatment plan assigns four controls:

  • A.8.9 Configuration management — storage configuration held in version control with a reviewed baseline.
  • A.8.3 Information access restriction — public access blocked at the account level; exports served through signed URLs only.
  • A.8.15 Logging — storage access logs shipped to the SIEM and retained.
  • A.8.28 Secure coding — the export feature runs a pre-merge check for public bucket policy during code review.

The SoA row for each control records applicability, the justification, and implementation status:

A.8.9 Configuration management
Applicability: Applicable
Justification: R-014 public exposure of customer report exports;
               baselines required for cloud storage and network config
Status:        Implemented (2022 Annex A)
Owner:         Platform Engineering Lead
Evidence:      repo infra/baselines, drift alert CFG-01, ticket OPS-2317

The evidence pack then holds the baseline file and its approving review, the drift-detection alert that fires on a policy change plus one closed ticket showing an alert investigated, a redacted sample of access logs, and the internal audit working paper where the auditor sampled two buckets and confirmed each is private.

That chain — risk R-014 to SoA row A.8.9 to baseline commit to drift ticket to audit sample — is what “implemented” means in ISO 27001 terms. A tool deployment on its own is not evidence.

Step 4: Audit yourself before someone else does (Clause 9.2). Plan the internal audit programme across the certification cycle, covering every clause and every applicable Annex A control. Assign auditors independent of the area under review — the person who wrote the SoA should not be the only person testing it. Record criteria, scope, findings and conclusion. Then hold management review (Clause 9.3) with the required inputs: status of previous actions, changes in external and internal issues, performance feedback including nonconformities and monitoring results, audit results, and risk assessment status. Output decisions on improvement and resources.

Step 5: Prepare for the two audit stages. Stage 1 is a readiness check: scope, documentation, the SoA, and whether you are ready for Stage 2. Stage 2 tests implementation — the auditor samples controls, interviews owners and looks for records. Expect sampling: two or three assets, a handful of starters and leavers, a sample of changes, a named supplier. Certification then runs on a three-year cycle with surveillance audits in years one and two and recertification in year three, so the ISMS must keep producing records, not just produce them once.

Where people get this wrong

  • Treating Annex A as a checklist. Implementing all 93 controls with no risk rationale creates work and an indefensible SoA. Treating every control as optional produces exclusions no auditor will accept.
  • Building the SoA last. It is an output of Clause 6.1.3. Assembled after the audit is booked, its justifications describe the control instead of the risk.
  • Risk registers nobody can score twice. If two people assess the same risk and get different numbers, the criteria are undefined. Fix the criteria, not the risk.
  • Evidence that lives in one person’s drive. Evidence must be retrievable by role, not by asking the consultant who built it. Store it where the process owner works.
  • Internal audit without independence. Clause 9.2 requires objectivity; auditing your own work is a finding waiting to be written.
  • Corrective actions with no root cause. Clause 10.1 wants the cause identified and effectiveness reviewed. “Retrained the user” repeats next cycle.
  • Confusing certification with customer assurance. A certificate does not answer a security questionnaire on its own; buyers still ask about A.5.19 supplier security, A.8.8 technical vulnerability management and A.5.24 incident management planning in detail.

Where to go next

Pick the next step by where you actually are, then use the linked guides for depth.

If you are close to an audit date, the highest-value work is usually not another control deployment but an independent internal audit and a pre-Stage 2 readiness review, booked early enough that findings can be closed and re-tested. Both are engagements to scope before the certification body is scheduled, not after.

Nathan Cole

Vulnerability management research, Dominion Cyber

Nathan Cole writes about vulnerability management and emerging threat research — why CVSS score alone is a poor prioritisation input, how patch operations actually get run, the mechanics of adversary-in-the-middle phishing, and the security model of AI assistants and browser extensions.

Get the weekly security brief

One email a week: what is worth patching, what is worth watching, and what is worth reading. No spam, unsubscribe any time.