Governance, Policy, and the Human Layer

Lesson 5: Governance, Policy, and the Human Layer

CC1 through CC5 and CC9 account for a large share of the Common Criteria, and they are not engineering problems. They are the questions: does leadership set the tone, is risk assessed, are controls monitored, are policies real, are people competent and supervised, are vendors managed, and can the business survive disruption. Companies that arrive at SOC 2 as a technical exercise lose weeks here β€” and it is the cheapest area to fix, because it is mostly writing things down and then actually doing them once a year.

The policy set

You do not need 40 documents. You need one approved, versioned information security programme policy plus a modest library of supporting policies, each with an owner, an approver, an effective date, a review cadence (annual is standard), and acknowledgements from staff. A typical set:

Governance & peopleTechnology & operationsRisk & continuity
Information security programme
Acceptable use
Code of conduct / ethics
HR security (screening, onboarding, offboarding)
Security awareness training
Access control
Confidentiality & NDA
Whistleblower / anonymous reporting
AI acceptable-use (increasingly expected)
Change management
Secure SDLC
Encryption & key management
Asset management
Vulnerability & patch management
Logging & monitoring
Backup
Physical & environmental security
Data classification, retention & disposal
Risk management
Vendor / third-party management
Incident response
Business continuity & disaster recovery
Privacy (if in scope)

Policy hygiene is a real test. Auditors check the document's approval (board or executive), its date, whether the version in force during the window is the one you can produce, and whether employees acknowledged it. A policy revised in November does not govern June. If your practice differs from the policy, the policy is the specification you will be measured against β€” so write down what you actually do, at the standard you can sustain, rather than an aspirational document you'll violate by February.

Governance structure

  • A named security owner β€” a CISO, a head of security, or a fractional/consulting arrangement that is documented. Someone must be accountable for the programme, and that person must have authority and a reporting line.
  • Oversight body β€” a security steering committee or equivalent meeting on a documented cadence with minutes; and board or independent oversight with periodic reporting. Minutes are evidence.
  • Defined roles and responsibilities β€” an org chart, a control-owner matrix, and job descriptions that reference security duties where relevant.
  • Competence and accountability β€” hiring standards, background checks where lawful and applicable, performance reviews, and a disciplinary process that is documented and understood.
The keystone control is the risk assessment. One properly executed annual risk assessment β€” plus a change-triggered refresh when you ship a new product, adopt a new vendor, or enter a new region β€” satisfies multiple criteria at once (CC3 and parts of CC9), drives your control selection, and gives the auditor the narrative for everything else. Format matters less than substance: identified risks, likelihood and impact, in-scope assets, existing controls, treatment decision, owner, due date, and evidence of closure. Most companies do this once for the audit and then let it go stale; the criteria expect ongoing assessment, not a 2024 artefact.

The employee lifecycle is tested end to end

StageControlEvidence
Before hireBackground check per policy; confidentiality agreement signedScreening record, signed NDA
OnboardingSecurity awareness training within a defined window (often 30 days); policy acknowledgement; access provisioned from an approved requestTraining completion report, acknowledgement log, access ticket
In roleAnnual refresher training; performance review; acceptable-use adherenceCompletion records, review forms
Role changeAccess re-evaluated and adjusted, not accumulatedChange ticket showing removals
OffboardingAccess revoked on a defined SLA; assets returned; knowledge transferOffboarding checklist with timestamps

If you use contractors or an outsourced support team, they are in scope with the same obligations. This is one of the most common description-vs-reality gaps.

Vendor and subservice management (CC9)

Maintain a vendor inventory with a tier (critical / high / low) based on data access and business dependency. For each higher-tier vendor: a pre-contract security review, contractual security and privacy terms (DPA, breach notification, subprocessor list), and an annual reassessment. For vendors that are themselves service organisations, collect and review their SOC report, note the CUECs, and record the review. β€œWe have their SOC 2 in a Google Drive folder” is a file, not a control β€” the control is the documented review with a date and a conclusion.

Incident response and continuity

  • Incident response plan β€” severity classification, roles, escalation, internal and external communications, legal/regulatory and contractual notification obligations and their clocks, evidence preservation, and a post-incident review with corrective actions tracked.
  • Tested, not shelved. An annual tabletop exercise with a written record of participants, scenario, decisions and follow-ups. If you had a real incident during the window, that is fine β€” the finding is having no documented response, not having the incident.
  • Business impact analysis identifying critical services and dependencies, feeding the BCDR plan with RTO/RPO targets, and an annual test whose results are retained.
  • Vulnerability to people β€” phishing simulation is optional for the criteria but cheap insurance against the finding that training exists on paper only.

Monitoring and internal audit (CC4)

The criteria expect you to evaluate your own controls, not just run them. In practice: a control self-assessment or internal audit cycle covering the control set, results documented, deficiencies communicated to the responsible owner, and remediation tracked to closure with dates. Combine it with a management review of the security programme at least annually β€” metrics, incidents, risk changes, training completion, open findings β€” with minutes. This is also where you catch problems before the auditor does, which is the entire point.

The most common governance failure: a beautiful policy library and no evidence that anything in it happened. Policies describe intent; the report tests operation. If your access review policy says quarterly and you have two reviews for four quarters, the policy has made your position worse, not better β€” it converted a vague weakness into a documented, measurable deviation.
The cheap win: put everything on a calendar with an owner and a destination folder β€” annual risk assessment, annual policy review, quarterly access reviews, annual training, annual pen test, DR test, vendor reassessments, tabletop, management review, board update. One calendar, twelve recurring items, and the governance half of your report becomes boring. Boring is the goal.

🧠 Knowledge Check

1. Your access-control policy states reviews happen quarterly. You performed two reviews in the 12-month window. What has the policy done to your position?

2. Which control satisfies multiple criteria at once and anchors the whole programme?

3. You store a critical vendor's SOC 2 report in shared drive. Is that sufficient vendor management?

Further Reading