Cloud Compliance Shared Responsibility Matrix: AWS, Azure, and Google Cloud Control Tasks
cloud securityshared responsibilityAWSAzureGoogle Cloudcompliance

Cloud Compliance Shared Responsibility Matrix: AWS, Azure, and Google Cloud Control Tasks

KKeepsafe Cloud Editorial Team
2026-11-16
7 min read

Compare AWS, Azure, and Google Cloud responsibilities and use a recurring matrix to track SaaS cloud security and compliance controls.

A cloud provider can secure the underlying infrastructure, but your SaaS company still owns many of the controls that determine whether customer data is protected and compliance evidence is complete. This cloud shared responsibility matrix compares AWS, Azure, and Google Cloud at a practical control level, shows what your team should track, and provides a recurring review schedule for cloud compliance.

Overview

Shared responsibility means that cloud security is divided between the provider and the customer. The provider generally operates and protects the physical facilities, core networking, host infrastructure, and managed service platform. Your company remains responsible for how services are configured, who can access them, what data is stored, how applications are developed, and whether required evidence is retained.

The exact boundary changes by service. A virtual machine usually leaves more operating system and patching work with the customer than a managed database. A serverless function may reduce infrastructure maintenance while still leaving your team responsible for identity permissions, application code, secrets, logging, and data handling. Do not treat a provider-level compliance certification as evidence that your particular workload is compliant.

For a useful cloud compliance checklist, document three things for every important control:

  • Owner: the cloud provider, your company, or both.
  • Implementation: the technical or procedural measure that satisfies the control.
  • Evidence: the record an auditor, customer, or internal reviewer can inspect.

Keep this mapping tied to your actual architecture. A generic AWS, Azure, or Google Cloud checklist is a starting point, not a substitute for reviewing your accounts, regions, services, repositories, and data flows.

What to track

Provider comparison at a control level

AWS, Microsoft Azure, and Google Cloud all publish shared responsibility guidance and assurance information. Their service names, configuration options, and documentation can change, so use the relevant provider documentation and assurance reports when validating a control. The high-level allocation below is intentionally service-neutral.

Control areaUsually owned by the providerUsually owned by the SaaS companyShared review question
Physical securityFacilities, hardware, environmental protections, and physical access to provider infrastructure.Physical security for your own offices, devices, and any customer-managed equipment.Does the provider assurance evidence cover the services and regions your environment uses?
Identity and accessIdentity features, authentication options, and platform access controls.Account lifecycle, least privilege, privileged access, service identities, MFA, and access reviews.Can every sensitive permission be linked to an owner and reviewed on schedule?
EncryptionEncryption capabilities for eligible services and protection of the underlying platform.Key selection, key access, application-level encryption, secrets management, and data classification.Are encryption settings, key custodians, rotation expectations, and exceptions documented?
Logging and monitoringAvailability and operation of service-level logging features.Enabling logs, selecting events, protecting log integrity, alerting, retention, and investigation.Would your team detect and investigate an unusual administrative or data-access event?
Backups and recoveryDurability and resilience of the provider platform as described for the selected service.Backup scope, schedules, retention, access restrictions, restore testing, and recovery objectives.Have you tested restoration for critical data rather than merely confirming that backups exist?
Vulnerability managementPatch and vulnerability management for provider-controlled infrastructure.Application code, operating systems you manage, containers, dependencies, images, and configuration weaknesses.Are findings prioritized, assigned, remediated, and verified within defined timeframes?
Incident responseDetection and response for incidents affecting the provider-controlled environment.Your applications, accounts, data, response plan, customer communications, and regulatory assessment.Do provider notification paths and your incident response policy work together?

Identity, encryption, and configuration

Identity is often the most important customer-owned control. Maintain an inventory of human users, privileged roles, service accounts, API keys, workload identities, and external integrations. Remove unused credentials, separate production access from development access, and require an approval path for elevated permissions. Record periodic access reviews as evidence rather than relying on a statement that reviews occur.

For encryption, record which data is sensitive, where it is stored, how it moves between systems, and which keys or secrets protect it. Encryption does not replace access control: a person or workload with legitimate access may still retrieve decrypted data. Track key ownership, recovery procedures, rotation practices, and emergency access separately.

Logging, backups, and vulnerabilities

Define the events that matter to your business before enabling every available log. Administrative activity, authentication events, changes to security settings, access to sensitive data, and changes to backups are common starting points. Restrict who can alter or delete logs, set a retention period that supports investigations and compliance needs, and test that alerts reach an accountable responder.

Backups should be mapped to business-critical systems and tested through restoration exercises. Track the last successful backup, the last restore test, the person who reviewed the result, and any unresolved limitation. Vulnerability management should cover cloud configuration as well as code and infrastructure. A patched host does not correct an overly broad role, exposed storage, or an unintended network path.

For related evidence planning, see the SOC 2 evidence collection guide and the SOC 2 controls mapping matrix for SaaS.

Cadence and checkpoints

A recurring review works best when each check has a named owner, a defined output, and a place to retain evidence. Adapt the cadence to your risk, architecture, and contractual obligations.

At least monthly

  • Review privileged users, newly created identities, inactive accounts, and high-risk access changes.
  • Check critical security alerts, open vulnerabilities, exposed resources, and failed backup jobs.
  • Confirm that logging and alert pipelines are operating for production systems.
  • Review changes to public endpoints, firewall rules, storage permissions, secrets, and service integrations.
  • Record exceptions with an owner, business reason, compensating control, and expiry date.

At least quarterly

  • Reconcile the cloud asset inventory with your architecture diagram, data flow records, and subprocessor register.
  • Perform a formal access review for administrators, production users, service accounts, and external vendors.
  • Test restoration of selected critical backups and document recovery results.
  • Review provider shared responsibility documentation, assurance reports, service changes, and regional availability relevant to your environment.
  • Run a tabletop incident exercise covering provider escalation, customer communication, evidence preservation, and regulatory assessment.

After a material change

Repeat the relevant review when you add a cloud service, migrate a workload, change regions, introduce a new data type, alter identity architecture, acquire a company, or change a critical vendor. Update the control matrix before or during the change, not after an audit request exposes the gap.

How to interpret changes

Not every provider update creates a compliance problem, and not every unchanged configuration remains safe. Classify a change by its effect on ownership, exposure, evidence, and recovery.

  1. Ownership change: a managed service may transfer patching or platform maintenance to the provider, while introducing new customer responsibilities for configuration and identity.
  2. Exposure change: a new endpoint, region, integration, or public setting may change the data-access or threat model.
  3. Evidence change: a new provider report, renamed service, altered log format, or changed retention setting may require your evidence collection process to be updated.
  4. Recovery change: a migration or service replacement may affect backup coverage, restore procedures, recovery time, or recovery point expectations.

When a control changes, update the owner and implementation fields first. Then identify the evidence that proves the revised control is operating. Avoid copying provider language directly into an internal policy unless it accurately describes your configuration. This distinction is especially important when answering a vendor security questionnaire or preparing SOC 2 evidence.

For privacy and healthcare workloads, connect the cloud matrix to your data protection obligations. Document where personal or regulated data is processed, which vendors can access it, and how incidents are escalated. A cloud control may support GDPR, HIPAA, or customer contractual requirements, but the matrix should not imply that infrastructure controls alone satisfy those obligations.

When to revisit

Set a calendar reminder for a monthly operational review and a deeper quarterly review. Also revisit this article and your internal matrix whenever AWS, Azure, or Google Cloud changes a relevant service, responsibility boundary, assurance report, regional option, or configuration workflow. The same applies when your company adds a cloud account, launches a product feature, changes a subprocessor, responds to an audit request, or experiences a security incident.

Use this practical closeout checklist:

  • List every production cloud account, subscription, project, region, and critical service.
  • Map identity, encryption, logging, backups, vulnerability management, and incident response controls to an accountable owner.
  • Attach current evidence links or records to each control, including access reviews, scan results, restore tests, and incident exercises.
  • Mark shared or unclear responsibilities for confirmation through provider documentation, contracts, or your internal security review.
  • Assign remediation deadlines and schedule the next review before closing the current one.

A useful shared responsibility matrix is a living operating record, not a one-time audit document. Keeping it aligned with your architecture helps your team answer customer questions, prepare for assurance work, and identify cloud security gaps before they become incidents.

Related Topics

#cloud security#shared responsibility#AWS#Azure#Google Cloud#compliance
K

Keepsafe Cloud Editorial Team

Cybersecurity and Compliance Editors

Senior editor and content strategist. Writing about technology, design, and the future of digital media. Follow along for deep dives into the industry's moving parts.