| Scope and covered activity | The provider's scope is the cloud service and the infrastructure, platform, application, facilities, personnel, and support processes it operates for that service. | The customer's scope is its use of the service: selected features, identities, configurations, workloads, data, integrations, endpoints, and customer-run processes. | Name the purchased service and technical layer. The boundary changes between IaaS, PaaS, and SaaS and can also change by feature or service tier. |
|---|
| Who acts | Provider owners commonly include the service owner, platform or infrastructure operations, security operations, supplier management, incident response, and customer support. | Customer owners commonly include the service owner, security and cloud engineering, identity administrators, data owners, application teams, supplier management, and the risk owner. | Assign a named owner to each action. A shared control needs at least one provider action and one customer action, not a single 'shared' label. |
|---|
| Trigger or threshold | Provider action is triggered by service design and operation, provider changes, incidents, vulnerabilities, capacity or resilience events, supplier changes, and commitments made to customers. | Customer action starts at selection and onboarding and continues through configuration changes, access changes, new workloads or data, incidents, assurance review, renewal, and exit. | Record scheduled reviews and event triggers. A change on either side can invalidate the previous allocation. |
|---|
| Core obligations | The provider should agree and document role allocation, secure its service boundary, disclose relevant capabilities and constraints, support customer requirements, and operate the controls assigned to it. | The customer should define cloud security requirements, assess service gaps, confirm it can perform its assigned role, configure and monitor its part of the service, and add controls where needed. | Write actions as testable statements. For example: provider supplies infrastructure logs; customer enables and retains workload logs. |
|---|
| Evidence and records | Useful provider evidence includes the service description, agreement, responsibility schedule, change notices, backup and logging specifications, incident process, assurance reports, and provider test results. | Useful customer evidence includes the approved architecture, configuration baseline, identity and access review, workload logs, key records, restore tests, vulnerability tickets, and risk decisions. | Link each matrix row to current evidence on both sides. Provider evidence cannot prove a customer action, and customer evidence cannot prove an opaque provider control. |
|---|
| Timing and cadence | Provider timing follows service releases, maintenance, assurance periods, incident procedures, vulnerability handling, supplier reviews, and contractual notice periods. | Customer timing follows onboarding, configuration and access reviews, workload changes, risk reviews, contract milestones, incident exercises, and exit. | Track each side's period and due date. An annual provider report does not set the customer's access-review or restore-test cadence. |
|---|
| Assurance and accountability | The provider can support assurance with service documentation, contractual commitments, independent reports or certificates, customer responses, and evidence permitted by the agreement. | The customer should assess whether provider evidence is sufficient and separately evidence its own controls through operational records, internal audit, risk review, or other applicable assurance. | ISO/IEC 27017 is guidance rather than a law or a standalone certification scheme. State the separate contract, certification scope, or legal requirement that makes a control consequential. |
|---|
| Overlap and reuse | Provider reports, certificates, service descriptions, logs, and test results can support provider-operated portions of several customer controls. | The customer can reuse provider evidence only for the covered entity, service, region, period, and control. It still needs evidence for every complementary customer action. | Record the exact claim each artifact supports, its limits and exceptions, and the customer evidence that completes the control. |
|---|
| Practical decision rule | Assign an action to the provider when only the provider can operate or evidence the relevant service layer. | Assign an action to the customer when it depends on the customer's choice, configuration, identity, workload, data, endpoint, or internal process. | Mark a control shared when the outcome depends on both actions. Then describe the dependency and test the end-to-end result. |
|---|