CISA SCuBA cloud security implementation
CISA SCuBA cloud security: assess, harden and prove Microsoft 365 or Google Workspace settings
A hands-on implementation guide for teams securing Microsoft 365 or Google Workspace. It shows how to define the tenant boundary, choose the right CISA baselines, run the assessment tools safely, turn findings into approved changes, and prove that security settings and cloud logs keep working. Federal teams get a separate path for Binding Operational Directive 25-01. Other organizations can use SCuBA as a strong engineering reference, not as a certificate or a claim of federal compliance.
- Issuing authority
- Cybersecurity and Infrastructure Security Agency (CISA)
- Timing
- For one established tenant, use a 6 to 12 week internal planning range: 1 to 2 weeks to settle scope and owners, 1 to 2 weeks to map policies and access, 1 week for the first assessment, 2 to 5 weeks for risk-ranked changes, then 1 to 2 weeks for retesting and executive sign-off. Multiple tenants, restrictive change windows, license dependencies or a large backlog can take longer. These are project estimates, not CISA deadlines. BOD 25-01’s original 2025 inventory, assessment and mandatory-setting deadlines have passed. If an in-scope agency is late, agree an accurate recovery plan and reporting route with its CISO, counsel and CISA; never backdate an inventory, assessment, exception or policy change.
- Reviewed
- Inside this guide
- 6 steps · 6 templates
Who it applies to
CISA’s Secure Cloud Business Applications project publishes architecture guidance, secure configuration baselines, assessment tools and visibility guidance for Microsoft 365 and Google Workspace. CISA recommends these resources for organizations across sectors. BOD 25-01 is a separate federal directive for in-scope Federal Civilian Executive Branch production tenants with applicable final baselines; agency counsel and the CISA implementation guidance should confirm scope, required policies, reporting and exceptions. The directive does not turn a private organization’s SCuBA assessment into an authorization, certification or CISA endorsement. Keep FedRAMP authorization, FISMA program obligations, customer contract terms and any sector rules as separate workstreams.
What you’ll produce
A signed tenant and service boundary; a versioned baseline applicability matrix distinguishing mandatory SHALL requirements from recommended settings; a controlled and repeatable assessment run; a prioritized, owner-assigned remediation backlog; approved changes and tested exceptions; a cloud visibility map tied to detection and response; and an executive evidence pack that shows current status without confusing a tool result with proof of effective security.
A September 2026 DHS OIG audit reported that 86% of FCEB agencies did not meet the June 2025 SCuBA implementation deadline. That figure describes the audit’s federal milestone finding, not today’s configuration state, every cloud customer or a certification failure rate. A separate CISA-led review of the 2023 Storm-0558 intrusion found that 22 organizations and more than 500 people worldwide had Exchange Online mailboxes compromised. The review is a useful reminder that tenant configuration, provider identity controls and usable telemetry are connected parts of the same defense. CISA makes its assessment tools available without a tool purchase fee; budget separately for tenant licenses, staff time, change delivery, support, telemetry retention and recurring assessment. There is no universal SCuBA implementation price because tenants, enabled products and remediation backlogs differ. The tools compare settings with policy; they do not change a tenant, prove that a control operated all year or replace incident detection. Read every failed, manual, excluded and errored check in context.
Run the project
Go to the roadmapDecisions for leadership
- What applies to each entity, product and activity?
- Have Legal or Regulatory confirm scope and applicable deadlines for each entity, activity and jurisdiction. Assign owners for incident reporting, filings and lifecycle changes where required.
- Who will do the work and who can sign?
- Appoint a sponsor, a day-to-day lead and a named owner for every workstream. Set aside their time and decide who can approve the outcome, accept remaining risk or authorize a filing.
- Which deadlines can we promise?
- Separate fixed external dates from internal targets. Confirm dependencies such as a vendor, test lab or assessor, and approve a backup plan if their dates slip.
How to spend the first week
- Ask the first-step owner to bring the most recent system, activity or product inventory and the contract or rule driving the work.
- Walk a real activity through the systems, suppliers and teams that run it. Verify the rule, authority and deadline that apply to the observed facts.
- Record what the sponsor approved, open questions, work owners and the next date a decision is needed. Send the brief to everyone who needs to contribute evidence.
What to review each week
Monitor obligations without a named owner, filings or notifications approaching deadline, and changes that should reopen the scope decision.
Kickoff brief to share with your team
Fill this in with your sponsor and send it to the people who own the work.
CISA SCuBA cloud security implementation kickoff brief Business reason and required outcome [Contract, obligation or approval] Scope and exclusions [Entities, systems or activities, and reasons] Fixed deadlines and internal milestones [Date, source and owner] Executive sponsor and authorized signer [Names] Project lead and backup [Names] First workstream Set the legal scope, cloud boundary and accountable owners · CIO or CISO sponsor + Cloud service owner + Agency counsel + Identity lead + Procurement [Name and due date] People, effort and external costs [Estimate, owner and contingency] Evidence folder and reviewer access [Location, access owner and review date] Decisions and blockers [Item, options, owner and decision date] First week 1. Ask the first-step owner to bring the most recent system, activity or product inventory and the contract or rule driving the work. 2. Walk a real activity through the systems, suppliers and teams that run it. Verify the rule, authority and deadline that apply to the observed facts. 3. Record what the sponsor approved, open questions, work owners and the next date a decision is needed. Send the brief to everyone who needs to contribute evidence.
How to hand off evidence
- Record the requirement or claim, the person responsible, the period and the result. Link to the original report, configuration, test or case record.
- Include the version, system or product, date, and how you gathered the record. When a report covers a group of items, retain its filters and reconcile the count to the source.
- Ask the reviewer to open the evidence and reproduce the key check. Record their decision and date. Keep missing items or failures visible with an owner, a due date and a retest when needed.
- Share through the agreed channel. Keep sensitive originals in a restricted location and link any redacted copies to the corresponding originals.
How this works in practice
Two questions to check scope
- Is this tenant in the federal directive’s scope, or are we using the CISA baselines voluntarily?
- Which exact baseline version and policies apply to our tenant edition, licensing and workload?
- Who may authorize Graph or Google API access for the assessment tool, and where may reports be stored?
- How will a failed, manual, not-applicable or uncollected check be distinguished in leadership reporting?
- Which logs let the SOC see changes to identity, sharing, forwarding, admin roles and security policy?
Worked scenario
A federal agency has two Microsoft tenants, a small Google Workspace collaboration tenant and a separate high-side environment. The cloud service owner puts all four in the tenant register, then asks counsel and the CISA liaison to confirm which are production tenants covered by BOD 25-01 and which baselines are final and applicable. The team runs the approved assessment against one tenant, stores the raw output in its restricted evidence location, and turns a failed external-sharing policy into a change record. Privacy reviews the partner workflow, the service owner pilots the new rule with a test group, and the SOC confirms alerts still fire when an approved external-sharing exception is used. The executive scorecard reports assessment coverage and open high-risk items separately; it does not call every passing tool result an effective control.
What to read next
- Federal FISMA audit readiness
Use the federal annual audit guide to organize system evidence, control testing and corrective actions.
- FedRAMP
Use FedRAMP when a cloud service provider needs an agency authorization for federal use.
- NIST Risk Management Framework
Use NIST RMF to connect security categorization, control selection, assessment and authorization decisions.
Official milestones
See the deadline calendar- February 21, 2025; tenant inventory refreshed annually in the first quarter
Inventory in-scope cloud tenants for BOD 25-01
The original inventory deadline has passed. Confirm the current CISA update process and reconcile tenants before the next first-quarter update; do not use the old list as a current inventory.
Official reference (opens in a new tab) - April 25, 2025
Deploy assessment tools and begin required reporting for in-scope BOD tenants
This original deadline has passed. Confirm the currently accepted tool version, tenant configuration, reporting method and frequency with CISA before submitting.
Official reference (opens in a new tab) - June 20, 2025
Implement the mandatory configurations then in scope
This original deadline has passed. Check the live Required Configurations page for later additions or changes, and manage any outstanding deviation through the current agency and CISA process.
Official reference (opens in a new tab) - Before authorizing a new in-scope production tenant
Assess applicable baselines and document the security decision
Confirm the exact new-tenant requirement and evidence with the agency authorizing official and current CISA implementation guidance.
Official reference (opens in a new tab) - Monthly, as an internal operating target
Review changes, failed assessments, exclusions and telemetry gaps
A monthly control review helps identify drift; use any shorter interval imposed by the agency or threat response.
Official reference (opens in a new tab) - Quarterly, as an internal operating target
Brief executives on coverage, remediation and accepted risk
Report denominators, aging, exceptions and evidence freshness. Never describe a single tool run as continuous compliance.
Official reference (opens in a new tab)
- 01Set the legal scope, cloud boundary and accountable owners, CIO or CISO sponsor + Cloud service owner + Agency counsel + Identity lead + Procurement, about 10 days
- 02Freeze the right baseline and requirement set, Cloud security architect + M365 or Google product owner + Licensing lead + Compliance counsel, about 7 days
- 03Approve and run a safe, repeatable baseline assessment, Tenant administrator + Cloud security engineer + Change manager + Security operations, about 7 days
- 04Triage every result and turn it into owned work, Control owners + Cloud security architect + SOC + Privacy + Internal audit, about 5 days
- 05Implement changes through a tested change and exception process, Product owner + Change advisory board + Identity or messaging admins + Privacy + User support + Risk owner, about 28 days
- 06Prove cloud visibility, report status and keep the baseline current, SOC lead + Cloud service owner + CISA or agency reporting liaison + CISO + Internal audit, about 14 days
Step by step
Checklist, instructions and 6 templates.
View plan as a table
| Step | Owner | When to act | Estimated target | Deliverables | Status |
|---|---|---|---|---|---|
| Set the legal scope, cloud boundary and accountable owners | CIO or CISO sponsor + Cloud service owner + Agency counsel + Identity lead + Procurement | Before tool access, baseline selection or any federal compliance statement | Set a start date | Signed tenant inventory, boundary diagram, applicability decision and owner matrix | Open |
| Freeze the right baseline and requirement set | Cloud security architect + M365 or Google product owner + Licensing lead + Compliance counsel | After scope approval and before defining assessment success | Set a start date | Versioned baseline crosswalk, applicability rationale and policy-to-owner register | Open |
| Approve and run a safe, repeatable baseline assessment | Tenant administrator + Cloud security engineer + Change manager + Security operations | After admin consent and data handling review; repeat on the agreed monitoring cycle | Set a start date | Approved assessment runbook, permission approval and reproducible run manifest with protected raw outputs | Open |
| Triage every result and turn it into owned work | Control owners + Cloud security architect + SOC + Privacy + Internal audit | Within five business days of each assessment as an internal service target | Set a start date | Normalized assessment results, exception log, prioritized remediation backlog and management summary | Open |
| Implement changes through a tested change and exception process | Product owner + Change advisory board + Identity or messaging admins + Privacy + User support + Risk owner | Use approved maintenance windows; start with a pilot; track every deviation until retest | Set a start date | Approved change tickets, user-impact and rollback plan, exception approvals and before/after retest evidence | Open |
| Prove cloud visibility, report status and keep the baseline current | SOC lead + Cloud service owner + CISA or agency reporting liaison + CISO + Internal audit | Continuous monitoring and reporting at the cadence required by current federal direction; monthly operational review and quarterly executive review are internal targets | Set a start date | Cloud visibility coverage map, tested detection evidence, current CISA reporting package and signed executive scorecard | Open |
Targets use illustrative calendar-day estimates and assume the steps run in sequence. Check official dates and agree real durations with your team.
- 01
Set the legal scope, cloud boundary and accountable owners
List every tenant, cloud environment, legal owner, business purpose, workloads, production status, administrator group, contract and data class. Reconcile the list against identity domains, cloud bills, procurement records, SaaS discovery, authorization packages and exit plans. For federal systems, have counsel and the CISA liaison document whether the entity and each production tenant are in BOD 25-01 scope, what national-security or other exclusions apply, and what reporting route is current. Name one accountable tenant owner and delegates for identity, Exchange or Gmail, collaboration, endpoint, logging, privacy, records and change approval.
What to produce
Signed tenant inventory, boundary diagram, applicability decision and owner matrixTenant register and signed scope memo
Gather first
- Identity domain and tenant exports
- Cloud bills and procurement records
- SaaS discovery
- System boundaries
- Data classification
- Agency organizational chart
- Contracts and FedRAMP packages
How to prepare it
- Create one row per tenant, including tenant ID, primary domain, cloud environment, product, legal owner, mission, data class, production status, authorization boundary, administrator group and authoritative source. Record the extract date and owner for every source list.
- For each federal tenant, document the BOD 25-01 scope decision with counsel: FCEB entity, production or operational tenant, finalized applicable CISA baseline, any exclusion and the accountable agency liaison. Keep voluntary use by non-federal organizations clearly separate from a binding requirement.
- Name the accountable tenant owner and delegated administrators for identity, collaboration, messaging, logging, privacy, records, change management and incident response. Have a second owner reconcile the register against cloud billing, identity domains, CMDB and authorization records; investigate every unmatched or duplicate tenant.
Worked example
Fictional details. Use the same fields with evidence from your work.
- Fictional tenant row
- Tenant ID 8f21…c402 | tenant org.example.gov | Microsoft 365 GCC | production | owner: Deputy CIO for collaboration | FCEB agency: yes | BOD scope: counsel review complete | source: Entra tenant export dated 14 Nov | reconciled: yes
- Fictional exception
- A dormant trial tenant appears in an identity export but not in the production CMDB. The owner records the validation steps, data check, disabled status, closure ticket and evidence before excluding it from the operational boundary.
Check before handing it over
Every tenant has an accountable owner, a reproducible scope decision and a traceable source. An agency-wide total can be rebuilt from the row-level register.
Obtain sign-off from the CIO delegate, counsel, cloud service owner and CISA reporting liaison before granting the assessment identity access.
- 02
Freeze the right baseline and requirement set
Download the current CISA baseline, required-configuration list, tool release notes and supported-environment guidance. Record retrieval time, version, source URL, tenant cloud (commercial, GCC, GCC High, DoD or Google Workspace edition), relevant workloads and license prerequisites. For BOD reporting, mark each policy as SHALL, SHOULD, manual, automated, configurable, not applicable or outside scope according to the current official materials. Only treat a policy as mandatory when the current directive and required-configurations source say it is. Map each requirement to the current tenant setting, intended state, implementation owner, dependency, test method, expected evidence and exception route.
What to produce
Versioned baseline crosswalk, applicability rationale and policy-to-owner registerPolicy-to-setting crosswalk
Gather first
- Current CISA baseline files
- BOD required configurations
- Tenant edition and license matrix
- Service configuration export
- Prior exceptions
- Product-owner list
How to prepare it
- Freeze the CISA retrieval date, baseline release, BOD list version and tool release. Save the original files in the controlled evidence repository and give each import a stable hash or repository revision.
- Create one row per policy ID. Capture exact version, product, policy text, SHALL or SHOULD indicator, automated or manual method, required configuration, applicable tenant environments, license dependency, current setting, intended setting and test source.
- Record applicability with a reason and approver. Map each SHALL policy to an accountable owner, change dependency, proof of implementation, compensating control if allowed, and a retest. Reconcile mandatory-policy totals with the live CISA required-configurations source before reporting.
Worked example
Fictional details. Use the same fields with evidence from your work.
- Fictional policy row
- MS.EXO.4.1v2 | Exchange Online | SHALL under the current checked directive list: yes | automated check: yes | license dependency: confirm with tenant contract | owner: Messaging Engineering | actual: fail | intended: enable approved external forwarding block | evidence: change CHG-4821 | retest: pending
- Fictional not-applicable decision
- The policy checks a product not licensed or provisioned in this tenant. The owner links the service inventory and license record, confirms no data or configuration exists, asks the compliance lead to approve the decision and schedules a recheck after procurement changes.
Check before handing it over
The file distinguishes mandatory from recommended policies, errors and not-applicable rows. Baseline and tool versions remain visible when results are compared across runs.
Send the approved crosswalk to tenant administrators, privacy, procurement and the assessment owner; review it whenever CISA changes a baseline or a service is added.
- 03
Approve and run a safe, repeatable baseline assessment
Read the current tool prerequisites, API permission list, supported tenant environments, installation instructions and output schema before execution. Use an approved workstation or managed runner, least privilege, a dedicated approved identity and a documented tenant identifier. Pin and record the tool and baseline versions. For ScubaGear, use the required configuration file for BOD 25-01 work; verify the selected organization and cloud before running. ScubaGear produces HTML, JSON and CSV results; ScubaGoggles has its own current setup and execution requirements. The assessment tool is not a remediation script. Save run ID, start/end times, account, tenant, command or configuration, errors, product coverage, output hashes and protected storage location. Do not put raw tenant reports in broad-access tickets or public repositories.
What to produce
Approved assessment runbook, permission approval and reproducible run manifest with protected raw outputsControlled runbook and evidence manifest
Gather first
- Tenant register
- Tool release and prerequisites
- Current API permissions
- Admin consent
- Baseline and YAML configuration
- Approved runner
- Security logging and storage rules
How to prepare it
- Document supported host, network egress, tool install source, pinned version, dependencies, admin role, app permissions, service account or interactive identity, tenant ID and selected environment. Obtain written approval for permissions and sign-in method before use.
- For Microsoft, record the ConfigFilePath, Organization, product selection, cloud environment and approved operator identity. Use the current ScubaGear documentation for any BOD-required configuration. For Google, follow ScubaGoggles prerequisites, APIs, authentication and configuration guidance rather than copying M365 steps.
- Capture run ID, start and end time, exact command or config hash, tool and baseline version, tenant context, per-product success or error, counts of automated/manual/applicable policies, file names and cryptographic hashes. Store HTML, JSON, CSV and logs under the approved tenant evidence permissions and retention rule.
Worked example
Fictional details. Use the same fields with evidence from your work.
- Fictional run manifest
- Run 2026-11-19-014 | tenant 8f21…c402 | cloud GCC | ScubaGear version recorded from Invoke-SCuBA -Version | ConfigFilePath hash 71ab…91d0 | operator sec-assess-02 | selected products: Exchange, Entra, SharePoint | result: 3 products complete, 1 collection error | reports: restricted/SCuBA/2026-11-19/ | raw files: 6, hashes recorded
- Safety gate
- The security architect confirms permissions, approved sign-in method, tenant domain, output storage and incident contact. A version update is reviewed in a test run before it replaces the pinned production runner.
Check before handing it over
A second operator can reproduce the assessment, prove which tenant and versions were evaluated, and locate outputs without searching broad collaboration channels.
Provide a sanitized summary to the remediation owners. Restrict raw tenant configuration and identity data to approved reviewers.
View template for Approved assessment runbook, permission approval and reproducible run manifest with protected raw outputsView example for Approved assessment runbook, permission approval and reproducible run manifest with protected raw outputsCISA ScubaGear documentation and prerequisites (opens in a new tab) - 04
Triage every result and turn it into owned work
Reconcile expected policy rows with tool output so missing or errored checks do not disappear from the denominator. For each result record policy ID and version, test source, actual state, expected state, automated or manual status, applicability, affected users or data, exploitability, business impact, confidence, evidence link and owner. Separate verified failure from collection error, policy not applicable, documented exception and incomplete coverage. Risk-rank the queue using exposure, privilege, external access, sensitive data and detectability. Start with global administrators, authentication, legacy access, external sharing, forwarding, app consent, audit logging and recovery controls when those policies apply.
What to produce
Normalized assessment results, exception log, prioritized remediation backlog and management summaryFindings register and scorecard
Gather first
- Run manifest
- Full tool outputs
- Baseline crosswalk
- Asset and user populations
- Threat model
- Prior findings
- Change and risk registers
How to prepare it
- Import one row for every expected policy, not only rows present in the tool report. Record result as pass, fail, manual review, not applicable, excluded, error or not assessed; preserve the tool’s exact finding and link to the raw output.
- For every failure or error, record tenant and product, policy ID/version, affected population, actual and expected state, sensitive data or privilege exposure, external access, exploitability, confidence, control owner, due date, interim safeguards and required approvals.
- Assign priority from risk and exposure, not from a simple count of failed controls. Show separate totals for mandatory SHALL, recommended SHOULD, manual checks, errors, exceptions and incomplete coverage. Age actions from detection to retest and require a senior risk owner for overdue mandatory deviations.
Worked example
Fictional details. Use the same fields with evidence from your work.
- Fictional finding
- Result: fail | policy MS.SHAREPOINT.1.3v1 | population: 3 sites allow anonymous links | evidence: restricted report row 12 | impact: public access to internal planning docs | owner: Collaboration Product Director | priority: urgent | interim action: disable new anonymous link creation | due: 5 business days | permanent action: approved allow-list and retest
- Fictional dashboard
- Coverage 94% means 94 of 100 expected checks produced interpretable results. It is not 94% compliance. Show 11 mandatory failures, 4 errors, 6 not-applicable decisions and a separate aging chart with data date.
Check before handing it over
Results reconcile to the expected policy population; uncertainty, manual checks and collection failures stay visible. Management can tell what is exposed, who owns the fix and when it will be tested.
Give high-risk exposures to the incident and risk processes immediately. Do not wait for a quarterly scorecard to contain active exposure.
View template for Normalized assessment results, exception log, prioritized remediation backlog and management summaryView example for Normalized assessment results, exception log, prioritized remediation backlog and management summaryCISA ScubaGear and ScubaGoggles assessment tools (opens in a new tab) - 05
Implement changes through a tested change and exception process
For each policy, document the current behavior, target setting, dependency, license requirement, user impact, rollback trigger, communications and validation query. Test in a non-production or limited pilot where feasible, then stage rollout and monitor service impact. Do not bulk-apply settings simply to turn a report green. A deviation needs a named risk owner, business reason, affected population, compensating control, approver, expiry or review date, and evidence that it is permitted under the applicable CISA and agency process. Re-run the same policy check after a change and independently verify the setting in the admin console or API.
What to produce
Approved change tickets, user-impact and rollback plan, exception approvals and before/after retest evidenceChange dossier and retest record
Gather first
- Prior assessment
- Target policy
- Product configuration guide
- License and user-impact analysis
- Change calendar
- Test account
- Support and rollback procedures
How to prepare it
- For each change state the current value and tenant scope, target value, reason, dependent services, user groups, license requirement, expected security result, outage risk, owner, approver and support message. Link the exact baseline policy and check the current vendor configuration instructions.
- Test the setting on a restricted pilot group or non-production tenant where available. Preserve pre-change and post-change exports, representative user journeys, alert tests, admin audit events and rollback trigger. Expand deployment in controlled waves with a service-health watch.
- Retest with the same versioned assessment method and independently inspect the setting in the administrator interface or API. A policy exception needs the permitted approving authority, business reason, scope, risk, compensating control, expiry or review date and monitoring owner. Record unresolved results and reopen a change if the retest fails.
Worked example
Fictional details. Use the same fields with evidence from your work.
- Fictional change
- CHG-4821 | block auto-forwarding to external addresses | owner Messaging Engineering | pilot: 12 users across 3 business units | support check: CRM notification workflow remains functional | rollback: restore approved connector if named business critical flow breaks | approvals: CAB and data owner
- Fictional closure
- The follow-up report shows the policy now passes, but the reviewer also checks the Exchange setting, samples the pilot users, verifies the change audit event and tests a safe alert. The evidence set records the policy version and an independent approver.
Check before handing it over
Closure is based on an effective setting, a controlled rollout and a successful retest, not an administrator assertion or a closed ticket.
Notify the SOC of meaningful changes and test that the expected audit and alert events remain available after the change.
View template for Approved change tickets, user-impact and rollback plan, exception approvals and before/after retest evidenceView example for Approved change tickets, user-impact and rollback plan, exception approvals and before/after retest evidenceCISA secure configuration baseline and BOD requirements (opens in a new tab) - 06
Prove cloud visibility, report status and keep the baseline current
Use CISA’s eVRF to map security questions to required telemetry, source system, observation point, collector, retention, detection rule, responder and test event. Cover tenant audit and identity logs plus relevant endpoint, secure access, email-security and cloud-provider signals. Prove events arrive with correct tenant context and timestamps, can be searched for the required period, generate an alert and reach a staffed response queue. Submit federal status only through the current approved CISA method; reconcile tenant inventory, applicable policies, tool results, exclusions, deviations and corrective actions before a senior official approves it. On every CISA baseline or tool update, assess changed policy IDs, licensing and tenant impact, rerun affected checks, update evidence and brief owners.
What to produce
Cloud visibility coverage map, tested detection evidence, current CISA reporting package and signed executive scorecardVisibility workbook and executive evidence pack
Gather first
- CISA eVRF guide and workbooks
- M365 or Google audit sources
- Identity provider
- Endpoint and secure access telemetry
- SIEM rules
- CISA reporting instructions
- Assessment trend
- Exceptions and POA&Ms
How to prepare it
- For every security question, map the threat or policy to the required event, source product, observation point, collection method, tenant context, expected delay, retention, detection rule, responder and evidence link. Include configuration changes and user activity such as new admin role, suspicious consent, external sharing, mailbox forwarding and audit-policy changes when supported and relevant.
- Test each path end to end using an approved benign test: create or change a controlled setting, confirm the audit event carries the right tenant and actor, measure time to SIEM, trigger the expected detection and confirm the on-call team receives and can investigate it. Record unavailable fields, expired retention, unsupported license features and unstaffed alerts as gaps.
- For federal reporting, reconcile the current tenant population, applicable policy set, assessment output, exclusions, deviations and remediation status to the latest CISA instructions. Obtain required agency approvals before submission and preserve the submitted file, validation response, receipt and correction history. Build an executive view with data date, denominators, coverage, overdue exposure, repeat findings, exceptions and ownership.
Worked example
Fictional details. Use the same fields with evidence from your work.
- Fictional visibility row
- Threat: unexpected global admin grant | event: directory role assignment | source: Entra audit | collector: agency SIEM | tenant ID included: yes | delay target: 10 min (internal) | detection: high-risk role outside change window | responder: identity on-call | test date: 19 Nov | result: alert received and triaged
- Fictional executive metric
- 3 of 4 tenant environments have current, interpretable assessments; 82 of 90 applicable policies produced results; 7 mandatory items remain failed; 2 collection errors; 4 approved exceptions expire within 60 days. The scorecard names each owner and includes the evidence date.
Check before handing it over
An independent reviewer can trace an executive metric to tenant rows, policy results, raw evidence and a tested telemetry path. Reporting identifies coverage gaps and risk acceptance instead of hiding them inside one percentage.
CISO signs the reporting and residual-risk summary. The CIO funds recurring tool operations, log retention and corrective work; internal audit samples the evidence independently.
View template for Cloud visibility coverage map, tested detection evidence, current CISA reporting package and signed executive scorecardView example for Cloud visibility coverage map, tested detection evidence, current CISA reporting package and signed executive scorecardCISA Extensible Visibility Reference Framework guidebook (opens in a new tab)
Official sources
- CISA · Secure Cloud Business Applications project (opens in a new tab)
- CISA · Binding Operational Directive 25-01 (opens in a new tab)
- CISA · BOD 25-01 required configurations (opens in a new tab)
- CISA · Free cloud security tools (opens in a new tab)
- CISA · ScubaGear for Microsoft 365 (opens in a new tab)
- CISA · ScubaGoggles for Google Workspace (opens in a new tab)
- CISA · SCuBA Technical Reference Architecture (opens in a new tab)
- CISA · Extensible Visibility Reference Framework guidebook (opens in a new tab)
- DHS OIG · CISA SCuBA implementation audit OIG-26-30 (opens in a new tab)
- CISA Cyber Safety Review Board · Summer 2023 Exchange Online intrusion review (opens in a new tab)