← All guides
Security Policies9 min read5 June 2026

Change Management Policy for SaaS: SOC 2 CC8, ISO 27001 A.8.32, and Deployment Controls (2026)

How to build a change management policy that satisfies SOC 2 CC8, ISO 27001 Annex A.8.32, and HIPAA change control requirements. Covers code review, deployment gates, change approval workflows, emergency changes, and evidence collection.

Why change management is a critical SOC 2 control

Change management is one of the most-tested areas in SOC 2 Type II audits. It's also where many fast-moving SaaS startups fail — not because they're reckless, but because their practices are good and informal rather than documented and evidenced.

SOC 2 CC8 (Change Management) requires that changes to infrastructure and software are authorised, tested, and reviewed before deployment to production. The question isn't whether you do code review — almost everyone does. The question is whether you can prove it with evidence auditors can see.

What frameworks require

FrameworkReferenceRequirement
SOC 2CC8.1Changes to infrastructure, data, software, and procedures are authorised, designed, developed, tested, configured, and implemented to meet change management controls
ISO 27001:2022A.8.32Changes to information processing facilities and systems subject to change management procedures
ISO 27001:2022A.8.8Technical vulnerabilities managed — patch management as a change management process
HIPAA§164.308(a)(5)(ii)(B)Protection from malicious software — includes change control to prevent unauthorised software deployment
PCI DSSReq 6.3–6.5Security vulnerabilities addressed, development and testing separated, code reviews required
NIS2Art. 21(2)(e)Security in network and information systems acquisition, development and maintenance, including vulnerability handling and disclosure

The core change management controls

1. Separation of development, staging, and production environments

The most fundamental change management control: code should never go directly to production from a developer's machine. At minimum, you need:

  • Development environment: where code is written and unit tested
  • Staging/pre-production: where changes are tested against production-like data and infrastructure
  • Production: live customer-facing environment

Production access should be limited to the deployment pipeline and specific on-call roles — not every engineer. This is a common SOC 2 gap: developers with direct production database write access because "we might need to fix things fast."

2. Code review requirements

Code review is your primary change management control for software changes. SOC 2 auditors look for:

  • At least one peer review required before merge (documented in PR)
  • Branch protection rules enforced in your repository (GitHub: protected branches with required reviews)
  • No self-approvals (the author cannot also be the sole reviewer/approver)
  • Security-sensitive changes (authentication, access control, payment processing, data exports) require review by a senior engineer or security reviewer

Evidence auditors will ask for: screenshots of branch protection settings, sample PRs with review approvals, pull request history for the audit period.

3. Testing requirements

What testing is required before deployment? At minimum for most SOC 2 assessments:

  • Automated unit/integration tests passing in CI pipeline
  • Staging environment test before production deploy (especially for major changes)
  • Security testing: dependency vulnerability scanning (Dependabot, Snyk, npm audit), SAST (Semgrep, CodeQL) for security-critical code paths
  • Regression testing for changes that affect existing functionality

4. Deployment authorisation

Who can authorise and trigger a production deployment? This should be explicit in your policy. Common models:

  • Auto-deploy on merge to main: Works if your branch protection and review requirements are strong. CI/CD pipeline IS the authorisation gate. Document this explicitly.
  • Manual deploy trigger: Engineer triggers deploy after merge. Creates an additional approval moment but adds friction.
  • Release manager model: Designated person authorises each production release. Required for HIPAA-adjacent environments or higher-risk changes.

Whatever model you use, document it — and make sure your CI/CD pipeline logs are retained as evidence.

5. Change documentation

Changes should be documented before deployment. The minimum documentation set:

  • Description of change and business justification
  • Affected systems/components
  • Risk assessment (low/medium/high)
  • Test results
  • Rollback plan
  • Approver and timestamp

For most SaaS companies, the pull request IS the change documentation. A well-maintained PR with description, test evidence, reviewer sign-off, and linked ticket satisfies this requirement if you can produce it during an audit.

6. Emergency change procedures

Every policy needs an emergency change procedure — what happens when production is down and you need to deploy a fix immediately without the normal review cycle. The key principle: emergency changes bypass normal process but are documented and reviewed retroactively.

  • Designated emergency approvers (on-call lead + CTO, for example)
  • Emergency change logged immediately (ticket, Slack #incidents, PagerDuty record)
  • Post-emergency review within 24–48 hours: root cause, what was deployed, was it appropriate
  • Emergency changes tracked separately and reviewed in regular change review meetings

SOC 2 auditors will ask: "How do you handle emergency changes?" Having a documented process — even one that's rarely used — demonstrates mature change management.

Evidence collection: what SOC 2 auditors actually want

SOC 2 Type II auditors typically sample 10–25 changes from your audit period and review evidence for each. What they look for:

ControlEvidenceWhere to Find It
Code review requiredPRs showing ≥1 approving review from someone other than the authorGitHub/GitLab PR history
Branch protection enforcedScreenshot of branch protection settings requiring PR + reviewsGitHub Settings → Branches
Automated testing in pipelineCI/CD pipeline logs showing tests run and passed before deployGitHub Actions / CircleCI / Jenkins logs
Staging test before productionDeployment logs showing staging deploy preceded production deployCI/CD logs, deployment records
Change description documentedPR descriptions or change tickets with description and justificationGitHub PRs, Jira tickets
Production deploy authorisationApproval in pipeline or sign-off in change ticket before production deployCI/CD approval gates, Jira workflow
Emergency changes reviewed retroactivelyPost-emergency review ticket or meeting notesJira, Notion, incident log

Infrastructure changes: same process applies

Change management doesn't just apply to application code. Infrastructure changes — Terraform/Pulumi changes, Kubernetes config updates, cloud IAM changes, database schema migrations — follow the same principles:

  • Infrastructure as Code (IaC) in version control: Terraform/Pulumi changes go through pull requests with the same review requirements as application code
  • Plan before apply: terraform plan output reviewed before terraform apply in production
  • Automated drift detection: infrastructure should match code; untracked manual changes are a red flag
  • Database migrations: schema changes tested in staging, migration rollback plan documented

Security patches and vulnerability management as change management

ISO 27001 A.8.8 requires vulnerability management. Patches are a type of change and should flow through your change management process — but with tighter timelines:

SeverityPatch TimelineChange Process
Critical (CVSS 9.0+)24–72 hoursEmergency change process; immediate patch or mitigating control; post-patch review
High (CVSS 7.0–8.9)7–14 daysPrioritised in next sprint; standard review + testing; staging before prod
Medium (CVSS 4.0–6.9)30 daysStandard change process; included in scheduled maintenance windows
Low (CVSS <4.0)90 daysTracked and addressed in normal development cycle

Common change management failures in SaaS audits

FailureSOC 2 ImpactFix
PRs merged without reviewCC8.1 finding — no evidence of authorisationEnforce branch protection: required reviews, cannot self-approve
No documented emergency change processGap finding — "what happens when you need to move fast?"Write a 1-page emergency change SOP; train on-call engineers
Direct pushes to main/masterCritical finding — branch protection not enforcedProtect main branch; force all changes through PRs
Developers with prod database write accessCC6 + CC8 overlapping findingRestrict prod DB access to DBA role + deployment pipeline service account only
No staging environmentTesting gap — changes not tested before productionEven a basic staging environment on the same cloud satisfies this; document it
CI/CD logs not retainedCan't produce evidence for sampled changesRetain CI/CD logs for minimum 1 year; export to log management system

Related guides and generators

⚠️ This guide is for informational purposes only and does not constitute legal advice. SOC 2 requirements depend on your Trust Service Criteria scope and assessor interpretation. Engage a qualified SOC 2 auditor for your specific programme.