How to Write an Information Security Policy (ISO 27001 Clause 5.2)
The information security policy required by ISO 27001 Clause 5.2 is a single top-level document established by top management. It must be appropriate to the organization's purpose, include security objectives or a framework for setting them, commit to satisfying applicable security requirements, and commit to continual improvement of the ISMS. It must be documented, communicated within the organization, and available to interested parties as appropriate. It is not your policy library — access control, acceptable use, and incident response belong in separate sub-policies. One or two pages is the right length, and the practical test is whether a founder could summarise it accurately without reading it.
The information security policy required by ISO 27001 Clause 5.2 is a single, top-level document established by top management. It states what the organization commits to and why — and it is deliberately short. One to two pages is normal. Ten is a sign that something which belongs in a sub-policy has wandered in.
This is the document most often confused with an entire policy library. If what you actually need is the set of policies — acceptable use, access control, incident response and the rest — start with the minimum policy set mapped to Trust Services Criteria or our nine internal company policies with templates. This article is about the one policy that sits above all of them.
What Clause 5.2 Requires
The standard is unusually specific here. Top management must establish an information security policy that:
- Is appropriate to the purpose of the organization. A payments company and a design agency should not have the same policy. Generic text is the fastest way to signal that nobody senior engaged with it.
- Includes information security objectives, or provides the framework for setting them. Either state the objectives, or state how and by whom they get set and reviewed. Most organizations choose the framework option and keep the objectives themselves in a separate, more frequently updated document.
- Includes a commitment to satisfy applicable requirements related to information security — legal, regulatory, and contractual.
- Includes a commitment to continual improvement of the information security management system.
And the policy itself must:
- Be available as documented information
- Be communicated within the organization
- Be available to interested parties, as appropriate
Those last three lines are where audits actually go wrong. Writing the policy is the easy half; proving it reached people is the half teams forget. More on that below.
A Section-by-Section Skeleton
A structure that satisfies Clause 5.2 without bloating:
- Purpose and scope — one paragraph on why the policy exists and what it applies to, aligned with your ISMS scope statement. Don't restate the scope; reference it.
- Commitment statement — what the organization commits to protecting, in plain language: the confidentiality, integrity, and availability of customer and company information.
- Objectives or the framework for setting them — either the objectives themselves, or a sentence explaining that objectives are set annually by management review and tracked as documented information.
- Commitment to applicable requirements — legal, regulatory, and contractual obligations relevant to information security.
- Commitment to continual improvement — of the ISMS, referencing Clause 10 in practice if not by number.
- Roles and responsibilities — who owns the ISMS, who approves policy, what is expected of every employee. Keep it to named roles rather than named people so the document doesn't churn with every hire.
- Supporting policies — a list of the sub-policies that sit beneath this one, so a reader knows where the detail lives.
- Consequences of non-compliance — a short, honest statement, consistent with your employment terms.
- Approval and review — who approved it, when, and the review cadence.
Ready to Streamline Your Compliance?
Discover how AuditBadger can simplify your compliance management process.
What Does Not Belong in It
The clearest way to keep this document short is to be strict about what goes elsewhere. Operational detail belongs in sub-policies and procedures:
- Password length and MFA specifics → access control policy
- What employees may install on a laptop → acceptable use policy
- Severity levels and escalation paths → incident response plan
- Backup schedules and retention periods → backup and continuity documentation
- How suppliers are assessed → supplier security policy
The reason isn't tidiness. It's change velocity. The top-level policy should be stable enough that top management can reapprove it once a year without reading a diff. Anything you expect to change quarterly needs to live somewhere that can change quarterly without a management-review cycle. For running that lifecycle across the whole set, see internal policy management.
The Test Worth Applying
Hand the policy to a founder or a senior manager who did not write it, wait a week, and ask them to summarise what the company commits to. If they can — accurately, without the document in front of them — it is appropriate to the purpose of the organization in the way Clause 5.2 means. If they produce a shrug, you have a template with your logo on it, and an auditor interviewing that same person will reach the same conclusion.
This is not a hypothetical audit technique. Auditors do interview staff, and "top management demonstrates leadership and commitment" is assessed partly through whether the people at the top can speak to their own policy. See Clause 5 on leadership for what else falls under that heading.
Communicating It — and Proving You Did
"Communicated within the organization" is a requirement, and an auditor will ask you to evidence it. Publishing the document to a shared drive is not communication; it is availability. The two are separate requirements in the clause for a reason.
What works as evidence:
- Acknowledgement records — a per-employee record that they read the current version, with a date and the version they acknowledged.
- Onboarding coverage — the policy is part of induction, with a record for each new joiner.
- Re-acknowledgement on material change — when the policy changes substantively, people acknowledge the new version rather than the old one silently remaining "accepted."
- Availability to interested parties — where appropriate, a public summary or a version supplied to customers under NDA. A trust centre is a common way to satisfy this without publishing internal detail.
Version control matters more than teams expect here. If acknowledgements aren't tied to a version, you cannot show who accepted what, and the record stops being evidence.
Approval and Review Cadence
The policy is approved by top management — in a startup, a founder or the CTO, whoever genuinely holds authority over the ISMS. Annual review is the common cadence, usually folded into management review under Clause 9.3, plus an out-of-cycle review when something material changes: a new regulatory obligation, a significant incident, a change of scope, or an acquisition.
Record the approval. A policy whose approval history is "last modified by" metadata in a document editor is thin evidence that top management established anything.
The Takeaway
Write one or two pages that sound like your company, cover the four things Clause 5.2 names, and point to the sub-policies where the detail lives. Then spend your remaining effort on the part that actually fails audits: getting it acknowledged, versioned, and reviewed on a cadence you can show.
Next: Clause 5 in full, the minimum policy set for what sits beneath this one, or the whole Clause 4–10 walkthrough.
Frequently asked questions
What does ISO 27001 Clause 5.2 require in an information security policy? +
Top management must establish a policy that is appropriate to the organization's purpose, includes information security objectives or a framework for setting them, commits to satisfying applicable security requirements, and commits to continual improvement of the ISMS. The policy must also be available as documented information, communicated within the organization, and available to interested parties as appropriate. Those last three are separate requirements, and the communication one is where audits most often fail.
How long should an information security policy be? +
One to two pages for the top-level policy required by Clause 5.2. If yours runs to ten, something that belongs in a sub-policy has wandered in — password rules belong in an access control policy, laptop rules in acceptable use, severity levels in the incident response plan. The reason is change velocity: the top-level policy should be stable enough for management to reapprove annually without reading a diff.
What is the difference between the information security policy and other security policies? +
The information security policy required by Clause 5.2 is a single top-level document establishing what the organization commits to and who is accountable. Everything operational — access control, acceptable use, incident response, backup, supplier security — lives in separate sub-policies beneath it. The top-level policy should list those sub-policies so a reader knows where the detail sits, but it should not contain their content.
Who approves the information security policy? +
Top management — in a startup, a founder or the CTO, whoever genuinely holds authority over the ISMS. Clause 5.2 places the obligation on top management specifically, and auditors assess leadership commitment partly by interviewing those people about their own policy. Record the approval properly: a policy whose approval history is only "last modified by" metadata in a document editor is thin evidence that top management established anything.
How do you prove an information security policy was communicated? +
Publishing to a shared drive proves availability, not communication — Clause 5.2 lists those as separate requirements. What works as evidence: per-employee acknowledgement records tied to a specific policy version and date, coverage in onboarding for every new joiner, and re-acknowledgement when the policy materially changes. Version control matters most here: if acknowledgements aren't tied to a version, you cannot show who accepted what.